# SearchLift AI: full plain-text index

SearchLift AI is an AI SEO platform. It audits a website for Google and for AI assistants,
applies reversible fixes on approval, tracks AI visibility (what ChatGPT, Claude, Perplexity,
Gemini and Grok tell customers about the business, which is the measuring half of ChatGPT SEO),
and counts the visits that arrive from those assistants. Built by Dude Lemon LLC, United States.
Contact the team through https://trysearchlift.com/contact. Support runs 24/7 on every plan,
Free included, from that form or from "Contact us" inside the console. Create a free account at
https://searchlift.dudelemon.cloud/register.

Site: https://trysearchlift.com | Generated from the same content the pages render.

## Pricing

### Free: $0
Start in a minute
- Up to 10 pages per audit
- 1 AI blog / month
- 2 AI fixes / month
- 1 tracked question
- Monthly AI answer checks on Gemini
- Weekly audits
- 30-day analytics history

### Plus: $8.99 a month
Growing sites
- Up to 50 pages per audit
- 2 AI blogs / month
- 25 AI fixes / month
- 2 tracked questions
- AI answer checks every 2 weeks on all 5 engines
- Weekly audits + auto-heal
- 90-day analytics history

### Premium: $17.99 a month
High-volume sites
- Up to 100 pages per audit
- 3 AI blogs / month
- 50 AI fixes / month
- 3 tracked questions
- AI answer checks every 2 weeks
- Audits every 3 days + auto-heal
- 6-month analytics history

### Enterprise: $59.99 a month
High-volume + teams
- Up to 400 pages per audit
- 4 AI blogs / month
- 100 AI fixes / month
- 4 tracked questions
- AI answer checks every 2 weeks
- Audits every 2 days + auto-heal
- Audit report PDF
- Developer handoff PDF
- 1-year analytics history

### Platinum: $129.99 a month
Largest sites + agencies
- Up to 800 pages per audit
- 5 AI blogs / month
- 300 AI fixes / month
- 6 tracked questions
- AI answer checks every 2 weeks
- Audits every 24 hours + auto-heal
- Audit report PDF
- Developer handoff PDF
- 2-year analytics history

### Plan limits, line by line

Columns: Free | Plus | Premium | Enterprise | Platinum

Coverage:
- Pages per audit: 10 | 50 | 100 | 400 | 800
- Products per audit: 10 | 50 | 100 | 400 | 800
- Automatic audit: Weekly | Weekly | Every 3 days | Every 2 days | Daily
- AI page reviews per scan: 15 | 50 | 150 | 400 | 800
- History kept: 30 days | Forever | Forever | Forever | Forever
- Analytics look-back: 30 days | 90 days | 6 months | 1 year | 2 years

AI visibility:
- Tracked questions: 1 | 2 | 3 | 4 | 6
- AI assistants tracked: Gemini | All 5 | All 5 | All 5 | All 5
- Answer checks: Monthly | Every 2 weeks | Every 2 weeks | Every 2 weeks | Every 2 weeks
- Competitor gap analyses: 1 / month | 5 / month | 10 / month | 20 / month | 30 / month
- Accuracy alerts: yes | yes | yes | yes | yes
- AI referral traffic: yes | yes | yes | yes | yes

Fixing:
- AI fixes: 2 / month | 25 / month | 50 / month | 100 / month | 300 / month
- Autopilot and auto-heal: no | yes | yes | yes | yes
- One-click undo: yes | yes | yes | yes | yes
- AI blogs: 1 / month | 2 / month | 3 / month | 4 / month | 5 / month
- Schema markup generations: 2 / month | 15 / month | 30 / month | 100 / month | 150 / month
- Image alt text: 2 / month | 25 / month | 50 / month | 150 / month | 200 / month
- llms.txt generations: 2 / month | 1 / month | 3 / month | 5 / month | 10 / month

Reporting:
- Search Console suite: yes | yes | yes | yes | yes
- Core Web Vitals re-check: Monthly | Every 2 weeks | Every 2 weeks | Weekly | Every 3 days
- Audit report PDF: no | no | no | yes | yes
- Developer handoff PDF: no | no | no | yes | yes
- Monthly report by email: no | no | no | yes | yes

Monthly, no contract, cancel from the billing screen. One plan covers one connected website.
Support runs 24/7 on every plan, Free included.

## How it works

Connected in minutes. No agency, no retainer, and no code on the connected platforms.

01. Connect your site: Connect your CMS or store in one click, give a headless site a signed webhook, or just paste a
URL. Within minutes you have an AI SEO score, a list of issues in order of impact, and the
questions your buyers are asking AI assistants.
02. Approve the fixes: Approve fixes one at a time, switch on autopilot inside limits you set, or export a work order
for your developer. Every change is saved before and after, so undo is one click.
03. Watch the answers change: Your AI visibility, citations per assistant and visits from AI, all tracked against the day each
fix went live: fourteen days before against fourteen days after.

## Everything included

- AI SEO audit: Every page scored twice: once the way Google reads it, and once the way an AI assistant has to read it to quote it.
- AI fixer: Rewrites, structured data, robots access and canonical fixes, applied when you approve them and always reversible.
- AI visibility monitor: ChatGPT, Claude, Perplexity, Gemini and Grok, asked your buyer questions with web search on, and every answer recorded.
- Accuracy guard: Alerts you when an AI assistant says something about your business that is wrong.
- Competitor gap analysis: Fetches the page the assistants cite instead of yours and tells you what it would take to win it back.
- Search Console suite: Queries you nearly rank for, pages getting fewer clicks than their position should earn, and index problems that went unnoticed.
- Content decay alerts: Pages losing traffic month over month, flagged while a refresh still takes an hour instead of a rewrite.
- Cannibalization finder: Your own pages competing for the same query, with the one to keep named.
- Broken links and redirects: Dead links found across the site, with the closest live page suggested for your own URLs.
- Core Web Vitals: Loading, responsiveness and layout stability measured against Google thresholds, page by page.
- AI article studio: Long articles written so an AI assistant can quote them, published straight to your CMS where the platform allows it.
- llms.txt and AI crawler access: Checks whether the AI crawlers are even allowed in, then writes the file that tells them what you sell.

## Features (https://trysearchlift.com/features)

Every feature and what it does for the business, in six areas. Scenarios are illustrative, not
customer data. Limits per plan are on the pricing page.

### What AI says about you

When a customer asks ChatGPT, Claude, Perplexity, Gemini or Grok who to buy from, or reads
Google’s own AI answer at the top of the page, a few businesses get named. These features show
you whether one of them is you.

AI visibility tracking (https://trysearchlift.com/features#ai-visibility): We put the questions
your customers ask to ChatGPT, Claude, Perplexity, Gemini and Grok, and tell you whether they
named you.
In practice: A wedding florist in Austin tracks the question "best wedding florist in Austin".
Every check shows which assistants named the florist, which named a competitor, and how that
changes over time.
- Named or missed, for each assistant and each question
- Where in the answer you appeared, and whether you were recommended or only mentioned
- Each assistant’s exact words, including what it thinks a good choice in your category needs
- Your AI visibility score over time, with the date of every fix marked on the line
Plans: Every plan, Free included. Console screen: AI Visibility.

Google’s AI answer, tracked too (https://trysearchlift.com/features#google-ai-overviews): Google
now answers many searches itself, at the top of the page, in an AI Overview. SearchLift reads
that answer for the questions you track and tells you whether you are in it, and which sites
Google leaned on.
In practice: A boutique hotel learns that Google’s own answer to "best boutique hotel in Lisbon"
does not name it, and that the three sites Google cited are travel guides the hotel has never
spoken to.
- Google’s AI Overview read for every question you track
- Named or not, and the sites Google cited
- Asked in your own country and language, so you see what your customers see
- Side by side with ChatGPT, Claude, Perplexity, Gemini and Grok
Plans: Paid plans. Console screen: AI Visibility.

Questions worth tracking next (https://trysearchlift.com/features#question-suggestions): You do
not have to guess what your buyers ask. SearchLift suggests the questions, drawn from what
Google shows and suggests around your business and from your own search data, and you track the
ones you like with one click.
In practice: A physiotherapy clinic is offered "physio or chiropractor for lower back pain?" and
"does physio help sciatica?", both taken from what people already search around the clinic’s own
terms.
- Questions phrased the way a person asks an assistant
- Each one shows where the demand came from
- Track any of them with one click
Plans: Every plan, Free included. Console screen: AI Visibility.

The websites AI trusted (https://trysearchlift.com/features#sources): See which web pages each
assistant read before it answered, and whether your own site was one of them.
In practice: ChatGPT recommended a rival plumber because a local directory listed them and not
you. Now you know exactly which directory to get listed on.
- The pages each assistant used, per question
- Whether your site was among them at all
- Which third-party page is doing the talking when you are missing
Plans: Every plan, Free included. Console screen: AI Visibility.

Who won instead, and why (https://trysearchlift.com/features#competitor-gap): When a competitor
gets recommended, we fetch the page the assistant used, compare it with yours, and list what
yours is missing.
In practice: A gym loses "best gym for beginners in Leeds" to a rival. The comparison shows the
rival’s page states prices, opening hours and a beginner class timetable. The gym’s page has
none of those.
- The competitor named for every question you lose
- Their page compared with yours
- What an article or page would need to contain to win the answer back
Plans: Every plan, Free included. Console screen: AI Visibility.

Wrong facts, caught (https://trysearchlift.com/features#accuracy-guard): You list the true facts
about your business. If an assistant says something different, you get one email so you can
correct it.
In practice: An assistant tells people a restaurant closes at 9 pm. It closes at 11. The owner
gets an email that day instead of finding out from an empty dining room.
- Facts you type in, one per line, such as "We ship worldwide"
- Phone, address and hours checked automatically where your store platform shares them
- One email per new problem, never a flood
Plans: Every plan, Free included. Console screen: AI Visibility.

Counted under every name you go by (https://trysearchlift.com/features#other-names): If an
assistant recommends you under your founder’s name or an older trading name, it still counts as
you.
In practice: A dental clinic is usually recommended as "Dr Patel", not by the clinic’s name.
Adding that name means those answers count as wins instead of misses.
- A list of other names you are known by, in Settings
- Answers that use any of them counted as yours
Plans: Every plan, Free included. Console screen: Settings.

Visitors who came from AI (https://trysearchlift.com/features#ai-traffic): Visits that arrive
from an AI assistant are counted on their own, so you can see which assistant is sending people
to you.
In practice: A software company sees that Perplexity sent 40 visitors last month and ChatGPT
sent 12. The AI work stops being a theory.
- Visits from each assistant, separately
- Their share of all your traffic, where site analytics are available
Plans: Every plan, Free included. Console screen: Analytics.

### Fixing your website

The audit finds what is holding your pages back with Google and with AI, and the fixer changes
it for you, with an undo on every change.

Two scores for every page (https://trysearchlift.com/features#audit): Every page gets a score
for how well Google can read it and a score for how easily an AI assistant can pull an answer
out of it. Every issue is spelled out, biggest first.
In practice: A bakery’s "Wedding cakes" page scores 41. The top issue: it has no description, so
Google writes its own, badly.
- One score for Google and one for AI, per page
- Every issue ranked by what will move first
- Products in your store scored the same way as pages
- Re-checked automatically, so the score is never stale
Plans: Every plan, Free included. Console screen: SEO Audit.

The fix is a button (https://trysearchlift.com/features#fixer): For each problem we write the
actual fix: a better title, a description, structured data, alt text. You read it, approve it,
and it goes onto your site.
In practice: The bakery approves a new description for "Wedding cakes". It is written to the
site through the platform’s own connection. Nothing to copy, nothing to paste.
- A preview of every fix before it goes live
- Applied through your platform’s own connection where one exists
- Elsewhere, handed to you as an exact instruction to paste
Plans: Every plan, Free included. Console screen: SEO Audit.

Undo, in one click (https://trysearchlift.com/features#undo): Every change keeps the value it
replaced. If you do not like a fix, put it back with one click.
In practice: A new page title reads oddly to the owner. One click, and yesterday’s title is
back.
- The old and new value side by side, on every change
- An undo button beside each one
Plans: Every plan, Free included. Console screen: Change Log.

Autopilot, with a ceiling (https://trysearchlift.com/features#autopilot): Switch it on and fixes
are applied for you, and applied again if a site edit undoes them. It works inside limits you
set, so you stay in charge of your own site.
In practice: A shop owner turns on autopilot in March and does nothing else. Titles and
descriptions are kept in shape all month, and the change log shows exactly what was done.
- Fixes applied for you, without a click
- Applied again when a theme change or an edit removes them
- Limits you set, so automation never gets ahead of you
Plans: Paid plans. Console screen: Settings.

Are the AI crawlers allowed in? (https://trysearchlift.com/features#crawlers): AI assistants
send crawlers to read your site. If your site blocks them, they cannot cite you. We check, and
open the door where the platform lets us.
In practice: A furniture store had been blocking OAI-SearchBot, the crawler ChatGPT uses, for a
year without knowing. One fix, and ChatGPT could read the site again.
- A check of your robots.txt for every AI crawler name
- The fix applied where the platform allows, or the exact lines to add
Plans: Every plan, Free included. Console screen: SEO Audit.

Ghost pages (https://trysearchlift.com/features#ghost-pages): Some sites say "page found" even
for pages that do not exist, so search engines index junk. We spot it and correct it where the
platform allows.
In practice: Every mistyped address on a store returned a working page. Google had indexed
hundreds of them, diluting the pages that matter.
- A test of how your site answers a page that does not exist
- The fix applied where the platform allows it
Plans: Every plan, Free included. Console screen: SEO Audit.

Structured data, written for you (https://trysearchlift.com/features#schema): Structured data is
a small block of code that tells Google and AI exactly what a page is: a product, an article, a
business. We write it, check that it is valid, and add it.
In practice: A product page gains a block that states the price, the stock status and the brand
in the form Google reads directly, so the listing can show the price in search results.
- The right type for each page, generated and validated
- Added to the page where the platform allows, offered as a suggestion where it does not
Plans: Every plan, Free included. Console screen: Toolkit.

Image descriptions (https://trysearchlift.com/features#alt-text): Every image should carry a
short description for people who cannot see it and for search engines. We write the missing
ones.
In practice: A photographer’s portfolio has 80 images and no descriptions. Each one gets a
sentence saying what is in the picture.
- Images without a description, listed
- A description written for each, ready to approve
Plans: Every plan, Free included. Console screen: Toolkit.

How quotable your pages are (https://trysearchlift.com/features#citability): A score for each
page on the things that have been measured to change whether an AI assistant quotes it: a clear
answer up front, figures, dates, plain language.
In practice: A consultancy’s services page scores low because it never states a price, a
timeframe or a plain sentence saying what it does. The biggest single win is shown first.
- A score per page, and the average across your site
- Each check passed or failed, with the reason
- The one missing thing that would lift the most pages
Plans: Every plan, Free included. Console screen: Toolkit.

Page speed (https://trysearchlift.com/features#speed): Loading time, responsiveness and layout
jumps, measured against Google’s own thresholds, page by page.
In practice: A home page loads in four seconds on a phone. The report says which of the three
measures failed and re-checks after a change.
- The three Core Web Vitals per page, against Google’s pass marks
- Re-checked automatically after changes
Plans: Every plan, Free included. Console screen: Toolkit.

Broken links (https://trysearchlift.com/features#broken-links): Links that lead nowhere are
found across your site, and for your own pages the closest live page is suggested.
In practice: A blog post from 2022 links to a product that was renamed. The link is flagged and
the current product page is suggested as the replacement.
- Every dead link, with the page it sits on
- The nearest live page suggested for your own URLs
Plans: Every plan, Free included. Console screen: Toolkit.

llms.txt (https://trysearchlift.com/features#llms-txt): A small text file that tells AI
assistants what your business is and sells, and which pages to read first. We check whether you
have one and write it for you.
In practice: A store with 300 products gets a one-page file naming the business, the categories
and the pages that matter, ready to publish at the root of the site.
- A check of the live file, if there is one
- A valid file written from your top pages, ready to publish
Plans: Every plan, Free included. Console screen: SEO Audit.

A to-do list for your developer (https://trysearchlift.com/features#work-order): If someone else
looks after your site, export the fixes as a page-by-page list with the exact before and after.
They do it without logging in, and we notice when it goes live.
In practice: An agency client ticks twelve findings and sends the PDF to their web developer. A
week later, SearchLift spots that the changes are live and starts measuring them.
- A PDF or spreadsheet, grouped page by page, with exact values
- Each item watched nightly until the change is really made
- A before-and-after measurement on that exact page
Plans: Enterprise and Platinum. Console screen: SEO Audit, Search Console.

The audit as a PDF (https://trysearchlift.com/features#audit-report): The whole audit as a
report you can send to a client, a boss or a board.
In practice: A marketing manager attaches the report to the monthly update instead of
screenshots.
- Scores, issues and fixes in one document
- Generated from the audit you are looking at
Plans: Enterprise and Platinum. Console screen: SEO Audit.

### Your Google and Bing data

Google and Bing already hold the honest numbers about your search traffic. These features read
them for you and turn them into a short list of things worth doing.

Google Search Console, read for you (https://trysearchlift.com/features#search-console): Connect
Google Search Console and we read Google’s own numbers about your site: what people search for,
what they click, and which pages Google has indexed.
In practice: A café owner has had Search Console for years and never opened it. Connected, it
becomes a short list of things worth doing this week.
- Google’s real figures, never estimates
- Connected with your Google account in one step
- Your property added and your sitemap submitted if they are not there yet
Plans: Every plan, Free included. Console screen: Search Console.

Almost ranking (https://trysearchlift.com/features#striking-distance): Searches where you
already sit just off the first page. Moving one of those up is the best-value work in SEO, and
we draft the new title and description for you.
In practice: A locksmith ranks eleventh for "emergency locksmith Bristol" with real searches
behind it. A better title moves it onto page one.
- Searches where you already sit just below the first page
- A title and description drafted from the page’s own top searches
Plans: Every plan, Free included. Console screen: Search Console.

Pages getting fewer clicks than they should (https://trysearchlift.com/features#ctr-gap): A page
ranking third should get more clicks than a page ranking ninth. When it does not, the title is
the problem, not the ranking. We list those pages.
In practice: A recipe page ranks second but gets the clicks of a page ranking eighth. Its title
is the recipe’s Latin name. A plainer title fixes it.
- Each page’s clicks compared with what its position should earn
- The pages losing clicks they have already won, ranked
Plans: Every plan, Free included. Console screen: Search Console.

Pages Google has lost (https://trysearchlift.com/features#index-health): Every page is checked
against Google’s own record of what it has indexed. Pages that quietly dropped out, and new
pages Google has not found, are listed.
In practice: A store’s new collection page was published a month ago and Google still has not
found it. It appears in the list, with Google’s own verdict beside it.
- Your sitemap plus every page Google has really served, checked one by one
- Google’s verdict on each, in Google’s words
Plans: Every plan, Free included. Console screen: Search Console.

Pages slowly losing traffic (https://trysearchlift.com/features#decay): Pages whose clicks are
sliding month by month, flagged while a refresh is still an hour’s work instead of a rewrite.
In practice: A guide that brought 400 visits a month in January brings 150 in June. Nobody
noticed, because the site total stayed flat. Now it is flagged.
- Pages losing clicks or impressions over time
- Pages whose content is getting old without review
- Every flag computed from the numbers, so it can be checked
Plans: Every plan, Free included. Console screen: Search Console.

Your own pages competing (https://trysearchlift.com/features#cannibalization): When two of your
pages chase the same search, both lose. We group them, name the one to keep, and write the steps
to merge the rest.
In practice: A clinic has three pages about teeth whitening. Google keeps swapping which one it
shows and none of them ranks well. Keep one, merge two.
- Your pages grouped by the search they share
- The page to keep, named
- The steps to merge the others
Plans: Every plan, Free included. Console screen: Search Console.

Searches become questions (https://trysearchlift.com/features#query-mining): The searches people
already use to find you on Google are turned into questions to track with the AI assistants, and
into article topics.
In practice: People find a tailor through "made to measure suit Manchester". That becomes a
tracked AI question and an article topic, so Google and AI work from the same demand.
- Real searches suggested as AI visibility questions
- The same searches suggested as article topics
Plans: Every plan, Free included. Console screen: Search Console.

Bing too (https://trysearchlift.com/features#bing): Connect Bing Webmaster Tools and see your
Bing and Copilot search performance beside Google.
In practice: A B2B firm finds a fifth of its search traffic comes through Bing, because that is
what its customers’ office computers default to.
- Bing clicks and impressions, connected the same way as Google
- Shown alongside your Google figures
Plans: Every plan, Free included. Console screen: Bing Webmaster.

### Articles

Articles chosen from real demand, written from your own site, in the shape AI assistants quote,
and published where you already publish.

Topics chosen from evidence (https://trysearchlift.com/features#topics): Article ideas come from
three real places: searches you nearly rank for, questions where an AI recommended someone else,
and the page that beat you. Each idea says why it was chosen.
In practice: A pet store is losing "best food for a puppy with a sensitive stomach" to a rival’s
article. That question is the next topic, with the reason attached.
- A short list of topics, each with its reason
- Nothing chosen from a generic keyword list
Plans: Every plan, Free included. Console screen: Blog Studio.

Written from your own site, in the shape AI quotes
(https://trysearchlift.com/features#written-from-your-site): Before writing, we read your own
pages and products, so the article says what you really sell. Then it is written with the direct
answer first, the shape assistants quote.
In practice: The pet store’s article opens with the answer, names the two foods the store
actually stocks, states their prices, and carries the structured data an assistant looks for.
- Built from your own content, products and stated facts
- The answer up front, then definitions, comparisons, prices and dates
- Article structured data attached
- No invented testimonials, statistics or claims about your business
Plans: Every plan, Free included. Console screen: Blog Studio.

Published where you publish (https://trysearchlift.com/features#publish): On WordPress, Shopify,
Webflow, Wix, Ghost and HubSpot the article is saved as a draft, or published, through the
platform’s own connection. On Notion it becomes a page. Anywhere else, download and paste.
In practice: A Shopify store owner approves the draft and it appears in the store’s blog, ready
to publish, without opening a second tab.
- A draft in your own CMS, or published if you choose
- A download for any other site
Plans: Every plan, Free included. Console screen: Blog Studio.

Articles on a schedule (https://trysearchlift.com/features#auto-blog): Turn on auto-blog and
articles are written on the schedule you choose, each one a draft until you say otherwise.
In practice: A landscaping firm sets one article a fortnight. They arrive as drafts, and the
owner reads each one before it goes out.
- A schedule you set
- Drafts for review, or straight to publish if you prefer
Plans: Every plan, Free included. Console screen: Blog Studio.

### Proof it worked

Every change is measured, every visitor is counted, and the numbers arrive in your inbox without
you logging in.

Your numbers at a glance (https://trysearchlift.com/features#overview): One screen with your
score, how often the AI assistants recommend you, your live Google clicks, and a tile for every
feature with the number that matters this week.
In practice: The owner logs in on a Monday, sees the score went up four points, Google clicks
up, AI share of voice held, and logs out.
- Your search and AI score
- How often AI assistants recommend you
- Google clicks and impressions, live from Search Console
- Every feature at a glance, each with its number for the week
Plans: Every plan, Free included. Console screen: Overview.

Everything that changed, in one list (https://trysearchlift.com/features#change-log): Every fix,
with the value it replaced, who made it (you, autopilot or auto-heal), and an undo button.
In practice: A business partner asks what SearchLift has actually done. The answer is one list,
newest first.
- Every change, with before and after
- What you did and what automation did, side by side
- One-click undo beside each entry
Plans: Every plan, Free included. Console screen: Change Log.

Before and after, measured (https://trysearchlift.com/features#before-after): Each fix is
measured: how often the AI assistants cited you before it went live, against how often they cite
you after, with the date marked on the chart.
In practice: A rewritten services page went live on the 3rd. Citations after are up against
citations before, with the date marked on the chart.
- A before-and-after figure per fix
- The date each fix went live, marked on your visibility line
Plans: Every plan, Free included. Console screen: Change Log.

Your visitors, including the ones AI sent (https://trysearchlift.com/features#analytics): With
the tracking code on your site, see who visits, where they came from, including each AI
assistant, and which pages do best.
In practice: A consultancy sees its best page is a guide it nearly deleted, and that most of its
AI visitors land there.
- Visitors, sources and top pages, in one place
- AI assistants shown as sources in their own right
Plans: Every plan, Free included. Console screen: Analytics.

Which AI bots read your site (https://trysearchlift.com/features#ai-crawlers): See which AI
crawlers visit your site, how often, and what they read, and whether each one is let in or shut
out, so you know the assistants can actually find you.
In practice: A publisher sees OpenAI’s search crawler reading its guides every day and
Anthropic’s never arriving, then finds the second one blocked in robots.txt.
- Every AI crawler that visited, by company and by what it is for
- How often, and which pages
- Whether your robots.txt lets each one in
Plans: Every plan, Free included. Console screen: AI Crawlers.

A weekly email (https://trysearchlift.com/features#weekly-email): Your score, AI share of voice,
visits from AI and the fixes made, every week, in one short email. Switch it off any time.
In practice: The owner never logs in during a busy season and still knows, every Monday, where
things stand.
- One email a week with the numbers that matter
- On or off in Settings, saved automatically
Plans: Every plan, Free included. Console screen: Settings.

A monthly report, as a PDF (https://trysearchlift.com/features#monthly-report): A full AI
visibility report, with your score, share of voice and the fixes applied, emailed as a PDF on
the first of each month.
In practice: An agency forwards the PDF to each client on the first without writing a word.
- Score, share of voice and applied fixes in one document
- Sent only once you switch it on
Plans: Enterprise and Platinum. Console screen: Settings.

### Setup and support

Connecting your site takes minutes, and a real person answers when you get stuck, on every plan.

Connect in one click (https://trysearchlift.com/features#connectors): WordPress, Shopify,
Webflow, Wix, HubSpot, Ghost and Notion connect through their own app stores or connections, so
fixes are applied for you. Next.js and custom sites get a signed webhook.
In practice: A Wix store owner installs SearchLift from the Wix App Market. Minutes later the
first audit is in, and fixes go straight onto the site.
- An app store install or a one-step connection
- Fixes written to the site through the platform’s own connection
- A page per platform saying plainly what it can write and what it can only suggest
Plans: Every plan, Free included. Console screen: Settings.

Any website, by address (https://trysearchlift.com/features#any-url): No connector for your
platform? Paste your web address. You still get the audit, AI visibility tracking, the Search
Console suite and article drafts. Fixes arrive as exact instructions instead of automatic
changes.
In practice: A restaurant on a small local platform pastes its address. The audit, the AI
answers and the Google data all work; the fixes come as a list to hand to whoever edits the
site.
- Everything that reads your site works the same
- Fixes as exact before-and-after instructions
Plans: Every plan, Free included. Console screen: Settings.

Your business, in your words (https://trysearchlift.com/features#business-profile): Your name,
what you do, where you serve, the searches you want to win, and the facts we protect. Everything
we write and check starts from this.
In practice: A removals firm writes "We serve Kent and East Sussex" and lists its three
services. Titles, articles and accuracy checks all use that, not a guess.
- Business name, address, services and target searches
- True facts, one per line, checked against what AI says
- Used by every fix, article and check
Plans: Every plan, Free included. Console screen: Settings.

More than one website (https://trysearchlift.com/features#workspaces): Each website is its own
workspace, with its own audit, settings and plan, under one login.
In practice: An agency runs six client sites from one account and upgrades only the two that
need it.
- A workspace per site, switched from the sidebar
- Each on its own plan, billed on its own
- The card you saved once works for every new workspace
Plans: Every plan, Free included. Console screen: Sidebar.

A real person, any hour (https://trysearchlift.com/features#support): Support runs around the
clock on every plan, including Free. Write from the contact form or from Contact us inside the
console, and a person replies by email, or on WhatsApp if you leave a number.
In practice: A first-time user gets stuck connecting Search Console at 11 pm. They press Contact
us, and the reply comes with their account details already known.
- Contact us in the console sidebar, with your account details attached
- The contact form on this site, for anything before you have an account
- A WhatsApp reply if you leave a number
Plans: Every plan, Free included. Console screen: Contact us.

## Platform

### AI visibility (https://trysearchlift.com/platform/ai-visibility)

Track your AI visibility: what ChatGPT, Claude, Perplexity, Gemini and Grok say about your
business, whether you were named, where, who was named instead, and every source cited.

A rank tracker watches a list of links. When a buyer asks an AI assistant which supplier to use,
there is no list, only one answer with a few names in it. SearchLift measures that answer, on
the questions you choose. That is AI visibility, and it is the first half of ChatGPT SEO.

Asked the way a customer asks: You give SearchLift the questions your buyers really put to an assistant, such as "best coffee
subscription for office teams". It asks all five assistants with web search on, which is how the
real apps answer, not the bare model. What comes back is what a customer would have seen.

Named, missed, and in what position: Every check records, per assistant and per question, whether your brand was named at all, where
it sat in the answer, and whether it was recommended or just mentioned. That becomes your AI
visibility score, tracked over time, with the date of every fix you applied marked on the line.

The sources behind the answer: AI assistants rarely cite a rival homepage. In live accounts they lean on trade blogs,
directories, marketplaces, Reddit and YouTube. SearchLift keeps the exact pages each assistant
used, so when you are missing you can see which third-party page is doing the talking, and
whether your own site was among the sources at all.

Who got recommended instead: For every question you lose, the competitor that won is named. The competitor gap analysis then
fetches the page the assistant cited, compares it with yours, and tells you what an article
would need to contain to win the answer back. That plan feeds straight into the article studio.

Accuracy guard: AI assistants sometimes state things about a business that are simply wrong: an old address, a
discontinued product, a price from two years ago. SearchLift checks every answer against the
facts you hold, flags the mistake on the dashboard, and sends one alert email per new problem,
never a flood.

Proof that it turned into visits: Visits whose referrer or campaign tags point to an AI assistant are counted on their own, so you
can see which assistant is sending people, and, where site analytics are available, what share
of all your traffic that now is. The work stops being a theory.

Questions:
Q: What is AI visibility?
A: AI visibility is how often, and how well, AI assistants name your business when a buyer asks
them a question in your category. SearchLift measures it per assistant and per question: named
or not, the position you held, who was named instead, and the sources each answer used.
Q: How is this different from a rank tracker?
A: A rank tracker records your position in a list of links for a keyword. An AI assistant does not
return a list; it writes one answer with a handful of names and sources in it. SearchLift
records that answer, per assistant, per question, so you can see whether you were in it, where,
and who was chosen instead.
Q: What is ChatGPT SEO?
A: ChatGPT SEO is the work of getting your business named and recommended in ChatGPT answers. This
screen is the measurement half: it shows whether ChatGPT names you today, which pages it used,
and who it named instead. The audit and fixer are the other half, and both cover Claude,
Perplexity, Gemini and Grok too.
Q: Which assistants are tracked?
A: ChatGPT, Claude, Perplexity, Gemini and Grok. The free plan tracks Gemini; every paid plan
tracks all five.
Q: Does it use web search or just the model?
A: Web search, on purpose. The real assistants search the web before answering a buying question,
so asking the bare model would measure the wrong thing. Web search also lets SearchLift keep the
sources each assistant used.
Q: How many questions can I track?
A: One on the free plan, two on Plus, three on Premium, four on Enterprise and six on Platinum.
Each one is put to every assistant on your plan, monthly on the free plan and every two weeks on
paid plans, and every answer is recorded.
Q: Can I see the actual answers?
A: Yes. Each assistant’s own wording is shown on the AI Visibility screen, including what it says a
good choice in your category needs, so you can see the bar it is judging you against.

### Audit and fix (https://trysearchlift.com/platform/seo-audit)

An AI SEO audit that scores every page for Google search and for AI assistants, ranks the issues
by impact, and applies the fix when you approve it. Undo is one click.

Most audit tools hand you a spreadsheet and leave. SearchLift ranks what it finds by what will
move first, writes the fix, and applies it the moment you approve it, keeping the old value so
you can put it back.

Two scores for every page: Pages, products and posts are crawled, and the tags each one really renders are read. Then each
page is scored twice: once the way Google reads it, and once for whether an AI assistant could
pull a clean answer out of it. Missing descriptions, missing structured data, duplicate titles,
thin content and blocked crawlers all show up as ranked, plain-language issues.

Fixes that are written, not described: For each issue the fixer writes the actual replacement: a title that fits the character limit, a
description based on the page, a block of schema.org JSON-LD that validates, a robots rule, an
alt text for an image. You preview it, approve it, and on connected platforms it is written to
the site through the platform’s own API.

Reversible by design: Every change keeps the value it replaced before it goes live. The change log shows each one with
a single undo, and undoing a fix does not spend your monthly allowance. Improving the site never
puts the site at risk.

Autopilot with a ceiling: Turn on autopilot and the automatic audit applies fixes on its own schedule. Automation is
capped at a share of your monthly allowance, with at least one fix always held back for you, so
the product can never spend your whole month before you have clicked anything.

Are the AI crawlers even allowed in?: The audit reads your robots.txt for the crawler names the AI assistants use and flags a site
that is blocking them. This is step one of ChatGPT SEO: if OAI-SearchBot cannot read your pages,
ChatGPT cannot cite them. Where the platform lets the app edit robots.txt, the fixer opens the
crawlers for you; elsewhere it gives you the exact lines to add. It also catches ghost pages,
where missing URLs answer 200 instead of 404, and corrects them where the platform allows.

Hand the work to a developer: If someone else looks after your site, export the whole audit, or only the items you tick, as a
page-by-page work order with exact before and after values. They apply it without ever logging
into SearchLift, and SearchLift notices when the change goes live and measures its effect.

Questions:
Q: What is an AI SEO audit?
A: An audit that checks each page for two readers at once: Google, which ranks pages in a list, and
AI assistants like ChatGPT, which read a page to write an answer. SearchLift scores every page
for both and lists what to fix, in order of impact.
Q: What does the audit check?
A: Titles, meta descriptions, canonical tags, Open Graph tags, structured data, image alt text,
content length, duplicate titles, robots.txt access for AI crawlers, how missing pages respond,
Core Web Vitals, and whether an analytics tag is installed at all.
Q: Does it change my site without asking?
A: Only if you switch on autopilot, and then only within a capped share of your monthly fix
allowance. By default every fix waits for your approval, and every applied fix can be undone in
one click.
Q: What if my platform has no write API?
A: The audit and the fixes are the same; the fix arrives as an exact instruction instead of an
automatic change. You apply it in your own editor, or export the developer work order and hand
it over.
Q: How often does it re-audit?
A: Weekly on Free and Plus, every 3 days on Premium, every 2 days on Enterprise and daily on
Platinum. The automatic sweep only re-reviews pages that changed, so a stable site costs nothing
to keep current.
Q: Is the structured data valid?
A: Each JSON-LD block is generated for the schema.org type of the page and validated before it is
offered as a fix. On pages where the platform gives no head slot, it is delivered as a
suggestion rather than a failed write.

### Search Console suite (https://trysearchlift.com/platform/search-console)

Google Search Console analysis that turns raw data into work: striking-distance queries,
click-through gaps, index coverage, content decay, cannibalization and work orders.

Search Console holds the only honest numbers about your Google traffic, and almost nobody reads
it. SearchLift connects to it and reads the whole thing for you, turning Google’s own data into
a short list of things worth doing this week.

Almost ranking: Real queries where you already sit at positions 8 to 20 with real impressions. Moving one of
those onto page one is the best-value work in SEO, and every row is Google’s own data, never an
estimate. For each, SearchLift reads the live page and drafts a title and description based on
the page’s real top queries.

Pages under-earning their position: A page ranking third that gets the click-through rate of a page ranking ninth does not need a
better ranking; it needs a better title. SearchLift compares every page’s click-through with
what its position should earn and lists the ones losing clicks they have already won.

Index health, page by page: The list of URLs is built from your sitemap plus every page Google has really served, then each
URL is checked against Google’s URL Inspection verdicts in small batches. Pages that quietly
dropped out of the index, and new pages Google has not found yet, are shown as a list rather
than left for you to discover.

Decay and cannibalization: Content decay flags pages whose score fell, whose content is getting old without review, or
whose clicks and impressions are trending down. The cannibalization finder groups your own pages
that compete for the same query, names the one to keep, and writes the steps to merge the rest.
Both are fully deterministic: every flag is computed from the numbers and can be reproduced.

A work order, then a receipt: Any set of findings exports as one self-contained PDF or CSV work order a developer can carry
out without SearchLift access. Each exported item becomes a watch: the nightly check re-reads
the live page, notices when the change was really made, and only then starts a 14-day
before-and-after measurement on that exact URL. Exported, done, proven, with nobody ticking
boxes.

Queries become questions: Query mining turns the searches people already use to find you on Google into tracked AI
visibility questions and article topics, so your Google SEO and your AI SEO work from the same
demand.

Questions:
Q: What do I need to connect?
A: A Google account with access to the Search Console property for your site. SearchLift asks for
the Search Console permissions at connection time and can add the property and submit your
sitemap if they are not there yet.
Q: Are the numbers estimates?
A: No. Every figure on the Search Console screens is read from Google’s Search Analytics or URL
Inspection APIs for your property. Where Google has no data for a page in a window, SearchLift
says so rather than showing a zero.
Q: Can it request indexing or remove pages?
A: Deliberately not. Google’s Indexing API accepts those calls for ordinary pages and discards
them, and the Removals tool has no public API. SearchLift shows you Google’s verdict and gives
you the honest next step instead.
Q: Who is the work order for?
A: Whoever edits your site. It groups findings page by page with exact before and after values, so
an agency or freelance developer can do it in one pass. Work orders are available on the
Enterprise and Platinum plans.
Q: How does Search Console help with AI SEO?
A: The queries people already use to find you on Google are the closest thing to the questions they
ask ChatGPT, which is where ChatGPT SEO starts. SearchLift turns them into tracked AI visibility
questions and article topics, so the two halves of search work from the same evidence.

### Article studio (https://trysearchlift.com/platform/content-studio)

An AI SEO article writer that picks topics from your Search Console queries and the AI answers
you are losing, writes answer-first drafts, and publishes them to your CMS.

Content bots write a post a day and hope. SearchLift writes fewer articles and picks them from
evidence: the queries you nearly rank for, the buyer questions you are losing in AI answers, and
the page a competitor won them with.

Topics from demand, not a keyword list: Suggestions come from three real sources: Search Console queries you already show up for,
tracked buyer questions where an AI assistant recommended someone else, and the competitor gap
analysis of the page that won. Each topic arrives with the reason it was chosen.

Outlined from your own site: Before writing, the studio reads a snapshot of your own content, products and, where the
platform exposes it, your CMS collections. The draft is based on what you really sell and say,
not on a generic view of your industry.

Written to be quoted: Articles are written section by section in an answer-first shape: the direct answer up front,
definitions and comparisons, clear prices and dates where you have them, and Article structured
data attached. Those are the properties that published research found move whether an AI
assistant quotes a page.

Published where you publish: On WordPress, Shopify, Webflow, Wix, Ghost and HubSpot the finished article is created as a
draft, or published, through the platform’s own API. On Notion it lands as a page in the
database you choose. On any other site you download it and paste.

A schedule you control: Auto-blog writes on the schedule you set, with a floor per plan so it can never run faster than
your monthly article allowance. It is the one automation that may use your whole allowance,
because a business that turns it on wants the articles written.

Measured like everything else: A published article becomes a page in the audit, a URL in the Search Console suite and, when it
is cited, a source in your AI visibility record. You find out whether it worked in the same
screens as everything else.

Questions:
Q: How many articles do I get?
A: One a month on the free plan, two on Plus, three on Premium, four on Enterprise and five on
Platinum. Auto-blog cannot schedule faster than that allowance allows.
Q: Will it publish without me seeing it?
A: Only if you turn on auto-blog. Otherwise every article is created as a draft you review, edit
and publish yourself.
Q: Does it write in my brand voice?
A: It writes from your own site content, products and stated business facts, which is where your
voice really lives. It does not invent testimonials, statistics or claims about your business.
Q: Can it publish to my platform?
A: WordPress, Shopify, Webflow, Wix, Ghost, HubSpot and Notion, through their own APIs. Any other
site gets the article to copy or download.
Q: How is this different from a normal AI blog writer?
A: A normal AI blog writer starts from a keyword. This one starts from evidence: the AI answers you
are losing, the queries you nearly rank for, and the page a competitor won with. Then it writes
in the answer-first shape that AI assistants quote, and measures whether the article got cited.
Q: Does it help with ChatGPT SEO?
A: Yes. The studio picks topics from the ChatGPT, Claude, Perplexity, Gemini and Grok answers you
are losing, and writes in the shape those assistants quote. Whether the article got cited shows
up in your AI visibility record.

## Integrations

Eight connectors and a plain URL. Each one says what SearchLift can write through the platform’s
own API, and what it can only hand you as an instruction. AI visibility tracking works the same
on all of them.

### WordPress (https://trysearchlift.com/integrations/wordpress)

Self-hosted and WordPress.com. AI SEO for WordPress with no plugin: connect through the REST
API, audit every page, write titles and descriptions via your SEO plugin, publish posts, and
track what ChatGPT says about you.

How it connects: Self-hosted: paste your site URL, a username and an Application Password from your WordPress
profile. WordPress.com: sign in and approve SearchLift; it works on every plan, Free included.
No plugin to install either way.

What it does:
- Audits the public pages, scoring the tags the site actually renders
- Writes titles and meta descriptions through Rank Math, SEOPress or Jetpack SEO fields
- Writes JSON-LD as a marked block in the post body, removed cleanly on undo
- Creates and publishes posts through the REST API
- Tracks buyer questions across all five assistants
- Exports a developer work order for anything that must be applied by hand
What stays manual: The tracking snippet, pasted once into your theme or tag manager, and any SEO field your plugin
does not open to the REST API, which arrives as a suggestion. Yoast does not accept outside
writes, so on a Yoast site titles and descriptions stay suggestions unless you switch to Rank
Math or SEOPress.

Capabilities: Reads and scores the live site: yes; Writes titles and meta descriptions: Via your SEO plugin; Writes structured data (JSON-LD): In the post body; Writes robots.txt for AI crawlers: no; Publishes articles: yes; Product coverage: From sitemap; Analytics tracker: Paste once; Writes from your business profile: yes; Reads your CMS content to write from: no; Reads platform site analytics: no

Q: Which SEO plugins are supported for writes?
A: Rank Math, SEOPress and Jetpack SEO expose their title and description fields to the REST API,
so SearchLift writes them directly. Yoast keeps its fields closed, so a Yoast site gets
suggestions instead of writes. Every write is verified against the live page before it is
recorded as applied.
Q: Do I need a plugin from SearchLift?
A: No. The connection is WordPress’s own REST API, authorized with an Application Password on
self-hosted sites or a WordPress.com sign-in. Nothing is installed on your site.
Q: Is an Application Password safe?
A: It is WordPress’s own mechanism for API access, revocable from your profile at any time, and it
is stored encrypted. It is not your login password.
Q: WordPress.com or self-hosted?
A: Both. Self-hosted sites and WordPress.com Business plans connect with an Application Password.
Every other WordPress.com plan, Free included, connects by signing in with WordPress.com; on
plans without SEO tools, titles and descriptions stay as suggestions while article publishing
works straight away.

### Shopify (https://trysearchlift.com/integrations/shopify)

App Store app. A Shopify AI SEO app that writes product, page and blog SEO through the Admin
API, installs its tracker for you, and tracks what ChatGPT and other AI assistants recommend.

How it connects: Install from the Shopify App Store. You grant the app its scopes during install and pick a plan
on Shopify’s own pricing screen.

What it does:
- Audits products, store pages and articles from the live storefront
- Writes product, page and article titles and descriptions through the GraphQL Admin API
- Creates and publishes blog articles to your Shopify blog
- Tracks buyer questions across ChatGPT, Claude, Perplexity, Gemini and Grok
- Injects the tracking beacon automatically via ScriptTag; nothing to paste
- Connects Google Search Console for striking-distance queries and index health
What stays manual: robots.txt and structured data. Shopify only exposes robots.txt as a theme template with no API,
and product JSON-LD lives in the theme, so both arrive as exact snippets to paste rather than as
failed writes.

Capabilities: Reads and scores the live site: yes; Writes titles and meta descriptions: yes; Writes structured data (JSON-LD): no; Writes robots.txt for AI crawlers: no; Publishes articles: yes; Product coverage: Native catalog; Analytics tracker: Automatic; Writes from your business profile: yes; Reads your CMS content to write from: no; Reads platform site analytics: no

Q: How is it billed?
A: Through Shopify managed pricing. The plan picker is Shopify’s own screen for the app, and the
charge appears on your Shopify bill.
Q: Does it touch my theme?
A: No. Writes go to product, page and article SEO fields and to blog articles through the Admin
API. The theme-level items, robots.txt and JSON-LD, are left to you with the exact snippet.
Q: Can it write product structured data?
A: It generates the product JSON-LD for you, validated against schema.org, and hands it over as a
snippet for your theme. Shopify gives apps no per-page head slot to write it through the API.

### Webflow (https://trysearchlift.com/integrations/webflow)

Data API connection. AI SEO for Webflow through the Data API: audit every page, write titles and
descriptions, publish CMS articles, and track what ChatGPT, Claude and Perplexity say about you.

How it connects: Sign in with Webflow and choose the site, or paste a Site API token generated in the site’s Apps
and integrations settings with Sites read, Pages read and write, and CMS read and write.

What it does:
- Audits the published pages from the live site
- Writes page titles and meta descriptions through the Data API, then publishes
- Creates and publishes articles as CMS items
- Reads your CMS collections so article drafts use your own facts
- Tracks buyer questions across all five assistants
- Connects Google Search Console for query and index data
What stays manual: Structured data. Webflow has no per-page head injection through the API, so JSON-LD arrives as a
paste-ready suggestion. The tracking snippet is added once in site settings.

Capabilities: Reads and scores the live site: yes; Writes titles and meta descriptions: yes; Writes structured data (JSON-LD): no; Writes robots.txt for AI crawlers: no; Publishes articles: yes; Product coverage: From sitemap; Analytics tracker: Paste once; Writes from your business profile: yes; Reads your CMS content to write from: yes; Reads platform site analytics: no

Q: Why does applying a fix publish my site?
A: Webflow only shows visitors the last published version, so a changed title that is never
published is invisible. SearchLift publishes after each write, and anything else you have staged
goes live with it. Finish or discard unpublished work before turning on auto-heal.
Q: Can it write JSON-LD?
A: Not through the API, because Webflow does not offer a per-page head slot there. Each block is
generated for you to paste into the page settings.
Q: What does the token need?
A: Site API tokens need a paid site plan. SearchLift needs Sites: read, Pages: read and write, and
CMS: read and write. Without Pages: write it can see your SEO settings but not change them.

### Wix (https://trysearchlift.com/integrations/wix)

App Market app. A Wix AI SEO app that audits every page, product and post, writes titles,
descriptions, structured data and robots rules back, and tracks your AI visibility across five
assistants.

How it connects: Install from the Wix App Market. Wix handles the installation and billing; there is nothing to
paste.

What it does:
- Audits pages, products and posts using the tags Wix actually renders
- Writes titles, descriptions and JSON-LD back to pages, products and posts
- Opens robots.txt for AI crawlers and fixes ghost pages that answer 200 instead of 404
- Creates and publishes articles to Wix Blog, with cover images imported to your Media Manager
- Reads Wix Site Analytics to show what share of all your sessions AI assistants now send
- Writes article drafts from your CMS collections and Wix Business Info
- Installs GA4 or Tag Manager correctly through Wix Marketing Tags
- Feeds your real Search Console queries into Wix’s own SEO keyword panel
What stays manual: Nothing. The tracking script is injected automatically, and the plan is managed from your Wix
subscription.

Capabilities: Reads and scores the live site: yes; Writes titles and meta descriptions: yes; Writes structured data (JSON-LD): yes; Writes robots.txt for AI crawlers: yes; Publishes articles: yes; Product coverage: Native catalog; Analytics tracker: Automatic; Writes from your business profile: yes; Reads your CMS content to write from: yes; Reads platform site analytics: yes

Q: Does it need any code?
A: No. Install from the App Market, open the app, run the first audit. The analytics tracker is
embedded automatically through Wix’s embedded-script extension.
Q: How is it billed?
A: Through Wix. Upgrades open the Wix upgrade screen for the app and the plan follows your Wix
subscription.
Q: Will it edit my site without asking?
A: Only with autopilot on, and only within a capped share of your monthly allowance. Every write is
reversible from the change log.

### HubSpot (https://trysearchlift.com/integrations/hubspot)

CMS API connection. AI SEO for HubSpot CMS: write titles, meta descriptions and head HTML on
site pages, landing pages and blog posts, publish articles, and track what ChatGPT and other AI
assistants say.

How it connects: Sign in with HubSpot, or create a private app with the content scope and paste its access token.
Reads and writes use the CMS API.

What it does:
- Audits site pages, landing pages and blog posts
- Writes titles and meta descriptions through the CMS API
- Writes JSON-LD into each page’s head HTML
- Creates and publishes blog posts
- Tracks buyer questions across all five assistants
- Connects Google Search Console for query and index data
What stays manual: The tracking snippet, added once in HubSpot’s site header settings.

Capabilities: Reads and scores the live site: yes; Writes titles and meta descriptions: yes; Writes structured data (JSON-LD): yes; Writes robots.txt for AI crawlers: no; Publishes articles: yes; Product coverage: From sitemap; Analytics tracker: Paste once; Writes from your business profile: yes; Reads your CMS content to write from: no; Reads platform site analytics: no

Q: Which HubSpot plan do I need?
A: One with Content Hub, formerly CMS Hub, because that is where website pages, landing pages and
blog posts live. A Marketing Hub-only account has nothing for SearchLift to write to.
Q: What scope does the private app need?
A: Only content, which grants read and write on CMS pages and posts. A Super Admin creates the app
and can rotate or delete its token at any time.

### Ghost (https://trysearchlift.com/integrations/ghost)

Admin API connection. AI SEO for Ghost CMS through the Admin API: meta titles, descriptions and
per-post JSON-LD written for you, articles published, and your AI visibility tracked across five
assistants.

How it connects: Paste your site URL and the Admin API key from a custom integration in Ghost’s settings. Reads
and writes use the Admin API.

What it does:
- Audits the published site
- Writes meta_title and meta_description on posts and pages
- Writes JSON-LD into each post’s code-injection head slot
- Creates and publishes posts through the Admin API
- Tracks buyer questions across all five assistants
- Connects Google Search Console for query and index data
What stays manual: The tracking snippet, added once in Ghost’s site-wide code injection.

Capabilities: Reads and scores the live site: yes; Writes titles and meta descriptions: yes; Writes structured data (JSON-LD): yes; Writes robots.txt for AI crawlers: no; Publishes articles: yes; Product coverage: From sitemap; Analytics tracker: Paste once; Writes from your business profile: yes; Reads your CMS content to write from: no; Reads platform site analytics: no

Q: Which Ghost versions?
A: Any Ghost site with the Admin API v5, which is every current Ghost(Pro) site and self-hosted
installs on Ghost 5.
Q: Is the Admin key safe to share?
A: Create a custom integration in Ghost for SearchLift so the key is scoped and revocable on its
own. It is stored encrypted, and deleting the integration in Ghost disconnects SearchLift
immediately.

### Notion (https://trysearchlift.com/integrations/notion)

Publishing destination. Notion as knowledge source and publishing destination: your databases
feed the AI SEO article studio, finished articles land as pages, and your live site is audited
and tracked for AI visibility.

How it connects: Sign in with Notion, or create an internal integration and paste its secret, then share the
database articles should be created in with the integration.

What it does:
- Reads the databases you share so article drafts use your own material
- Creates finished articles as pages in the database you choose
- Audits your real website by public crawl, exactly like any URL
- Tracks buyer questions across all five assistants
- Connects Google Search Console for your real site
What stays manual: SEO fixes on your actual website, which arrive as suggestions or as a developer work order,
since Notion is not where the site lives.

Capabilities: Reads and scores the live site: yes; Writes titles and meta descriptions: no; Writes structured data (JSON-LD): no; Writes robots.txt for AI crawlers: no; Publishes articles: To a Notion database; Product coverage: From sitemap; Analytics tracker: Paste once; Writes from your business profile: no; Reads your CMS content to write from: yes; Reads platform site analytics: no

Q: Can it fix my Notion-hosted site?
A: It audits it like any public URL and gives you the fixes as instructions. Notion has no SEO
fields to write through the API.
Q: Why does it say no database was found?
A: An integration only sees pages that have been shared with it. Open the database page in Notion,
choose Connections, pick SearchLift, and it appears on the next test.

### Next.js and custom sites (https://trysearchlift.com/integrations/nextjs)

Signed webhook. AI SEO for Next.js, Astro, Nuxt and headless sites: a public-crawl audit, AI
visibility tracking across five assistants, and approved fixes delivered as signed webhooks you
apply.

How it connects: Enter your site URL. Add a POST endpoint to your app and SearchLift generates a signing secret,
shown once; every approved change is then delivered to that endpoint as a signed payload. Leave
the endpoint blank and you keep audits, monitoring and reports with no write-back.

What it does:
- Audits the public pages from the live site, with products discovered from the sitemap
- Delivers approved fixes and articles as signed outbound webhooks
- Includes robots.txt changes in the same delivery
- Tracks buyer questions across all five assistants
- Connects Google Search Console for query and index data
- Exports a developer work order as an alternative to the webhook
What stays manual: Applying the payload is your code’s job; the write is marked live once your endpoint answers
with a 2xx. The tracking snippet is added to your layout once.

Capabilities: Reads and scores the live site: yes; Writes titles and meta descriptions: By webhook; Writes structured data (JSON-LD): By webhook; Writes robots.txt for AI crawlers: By webhook; Publishes articles: By webhook; Product coverage: From sitemap; Analytics tracker: Paste once; Writes from your business profile: no; Reads your CMS content to write from: no; Reads platform site analytics: no

Q: What is in the webhook?
A: The target URL, the field being changed, the current value SearchLift observed, the new value,
and an HMAC-SHA256 signature over the timestamp and body using your signing secret. Articles
arrive as structured HTML with their metadata. The developer guide in the console includes a
complete Next.js handler.
Q: Which frameworks?
A: Any. Next.js, Astro, Nuxt, Hugo, Eleventy, or a hand-written site. If it serves public HTML over
HTTPS and you can receive a POST, it connects.

### Any website (https://trysearchlift.com/integrations)

The full AI SEO audit from a public crawl, AI visibility tracking, the Search Console suite,
accuracy alerts and article drafts. Fixes arrive as exact instructions or a developer work
order.
How it connects: Paste the URL.
Capabilities: Reads and scores the live site: yes; Writes titles and meta descriptions: no; Writes structured data (JSON-LD): no; Writes robots.txt for AI crawlers: no; Publishes articles: Draft and download; Product coverage: From sitemap; Analytics tracker: Paste once; Writes from your business profile: no; Reads your CMS content to write from: no; Reads platform site analytics: no

## Support

Support runs 24 hours a day, seven days a week, on every plan including Free. Your message goes
straight to the people who build the product.
- Contact us, inside the console: One click in the sidebar. Your account and site details are attached for you, so you only have to describe the problem.
- The contact form on this site: For anything before you have an account: plans, connectors, agencies, or a site you have not connected yet.
- WhatsApp, if you want it: Leave a number with your message and the reply reaches you there as well as by email.

## Frequently asked

Q: What is AI SEO, and how is it different from normal SEO?
A: Normal SEO gets your pages ranked in a list of links on Google. AI SEO gets your business named
and cited when ChatGPT, Claude, Perplexity, Gemini or Grok write the answer instead of showing a
list. Some people call it generative engine optimization, or GEO. The two need different things:
a page can rank first on Google and still be useless to an AI assistant, because nothing on it
states a clear answer. SearchLift measures and fixes both.

Q: What is AI visibility?
A: AI visibility is how often, and how well, AI assistants name your business when a buyer asks
them a question in your category. SearchLift measures it per assistant and per question: named
or not, the position you held, who was named instead, and which sources the answer used. It is
the AI equivalent of a rank tracker, for answers instead of links.

Q: What is ChatGPT SEO?
A: ChatGPT SEO is the work of getting your business named and recommended in ChatGPT answers.
ChatGPT searches the web before it answers a buying question, so it comes down to three things:
whether its crawler can reach your pages, whether those pages state clear facts, and whether the
pages ChatGPT already trusts mention you. SearchLift tracks the answers, finds what is missing
on your pages and fixes it, and does the same for Claude, Perplexity, Gemini and Grok.

Q: How does SearchLift know what ChatGPT says about my business?
A: You tell it the questions your buyers ask. SearchLift puts those questions to ChatGPT, Claude,
Perplexity, Gemini and Grok with web search on, which is how the real apps answer, then records
whether you were named, what position you held, which sources each one used, and what it said
about you.

Q: Do I need to be technical?
A: No. Connected platforms install in one click, the audit is written in plain language, and a fix
is a button. If someone else looks after your site, export the work order and hand them a
page-by-page list with the exact before and after values.

Q: What if I get stuck?
A: Support runs 24/7 on every plan, including Free. Write from the contact form on this site, or
from Contact us inside the console, which already has your account details. A person replies by
email, or on WhatsApp if you leave a number.

Q: Is it safe to let it change my site?
A: Every change keeps the old value first, so undo is one click. Autopilot only runs inside limits
you set, and it is capped so it can never spend your whole monthly allowance before you have
used any of it.

Q: Which platforms does it work with?
A: WordPress, Shopify, Webflow, Wix, Ghost and HubSpot connect through their own APIs or app stores
and take the fixes directly. Notion is a publishing destination for articles. Next.js and
headless sites get a signed webhook. Any other website connects by URL: you get the full AI SEO
audit, AI visibility tracking, the Search Console suite and article drafts, and fixes arrive as
exact instructions instead of automatic changes.

Q: Will I be able to tell whether it worked?
A: Yes. Visits that come from AI assistants are counted on their own, and each fix is measured on
how often you were cited for the 14 days before against the 14 days after. You see the change,
marked with the date the fix went live.

Q: What does the free plan actually include?
A: A real audit of 10 pages and 10 products, one tracked buyer question checked monthly on Gemini,
two AI fixes, one article, a weekly automatic audit and 30 days of history. No card, and no
expiry.

Q: Can I cancel?
A: Any time, from the billing screen. Plans are monthly and there is no contract.

## Blog

Plain writing on AI SEO, ChatGPT SEO and AI visibility: how the AI assistants choose what to
say, what Google actually requires, and what a site can do about either. Every figure is
sourced.
Feed: https://trysearchlift.com/rss.xml

### Does ChatGPT recommend your business? How to find out in ten minutes (https://trysearchlift.com/blog/does-chatgpt-recommend-my-business)
Published September 13, 2026. 8 min read. AI visibility.

A repeatable check that tells you whether ChatGPT names your business when a buyer asks, who it
names instead, and which pages it used. No tool required for the first pass.

Does ChatGPT recommend your business? Probably not, and the way to find out is to ask it the
question your buyer would ask rather than a question about you. Type your own business name into
ChatGPT and it will describe you, because you named the entity. Type "who is the best commercial
roofer in Tulsa" and it has to choose, and whether it chooses you is the thing you actually want
to know. The whole check takes about ten minutes and needs no tool.

Do not ask ChatGPT about your business. Ask it the buying question your customer asks, with no
brand name in it, and record who gets named. That one change is the difference between a
flattering answer and a useful one.

#### Why this stopped being a curiosity in 2026

Use of ChatGPT and similar generative AI tools for local business recommendations rose from 6%
of consumers in 2025 to 45% in 2026, according to the BrightLocal Local Consumer Review Survey
2026, read on 13 September 2026. That makes AI the third most used source of business
recommendations behind Google and Facebook. Over the same period Google’s share as a review
source fell from 83% to 71%.

The same survey found 40% of consumers say they trust AI platforms to provide business
recommendations, against 32% who disagree, and that 82% read AI-generated review summaries with
23% willing to decide on the summary alone. A quarter of the people reading about you through an
assistant never see the underlying sources.

Table: Source of local business recommendations | 2025 | 2026
- Google | 83% | 71%
- Facebook | 40% | 49%
- ChatGPT or other generative AI tools | 6% | 45%
- Apple Maps | 14% | 27%
(Share of consumers using each source, BrightLocal Local Consumer Review Survey 2026, read 13 September 2026. The AI row is the largest single-year movement the survey has recorded in sixteen years.)

#### The ten-minute check: does ChatGPT recommend your business?

Run this as written. The value is in the discipline of the question, not in the tooling.

1. Write down the question a buyer types when they do not know you exist. Not "tell me about Acme
Roofing" but "who is the best commercial roofer in Tulsa". No brand name, no jargon, the buyer’s
words, one geography if geography matters.
2. Open a new ChatGPT conversation with web search on, and make sure you are logged out or in a
temporary chat. A logged-in account carries memory and prior conversations, which will quietly
bias the answer toward businesses you have already discussed.
3. Ask the question exactly as written. Record the businesses named, in order, and note whether you
appear at all.
4. Ask the assistant which sources it used. Record the URLs. These are the pages that decided the
answer, and they are almost never the businesses’ own home pages.
5. Repeat the same question in a second temporary chat. The answers will differ, because the model
samples and the live search results move. Two runs tell you whether your absence is consistent
or incidental.
6. Repeat the whole thing for the other four assistants your buyers use: Claude, Perplexity, Gemini
and Grok. An answer in one is not an answer in all five, and the gaps between them are the most
useful part of the exercise.

Record six fields per run: the assistant, the date, whether you were named, the position you
held in the list, who was named instead, and the sources cited. The method and the recording
sheet are covered in more detail in how to check what ChatGPT, Claude and Gemini say about your
business.

#### Reading the result honestly

One run is an anecdote. The reason to do two runs per assistant is that generative answers vary
between identical prompts, so a single absence proves very little and a single appearance proves
less. What you are looking for is a pattern across runs and across assistants.

- Named in most runs, in most assistants. You are in the consideration set. The work now is
accuracy: check that what they say about you is correct, because an assistant repeating an old
price or a closed location costs more than not appearing.
- Named in one assistant only. Usually a source problem. Look at which sources that assistant
cited and whether the others reach them at all.
- Never named, competitors consistently named. Go and read the pages the assistants cited. The
question is not why they like your competitor, it is what that page states plainly that yours
does not.
- Nobody consistently named. The category may be too broad for the assistant to answer, or your
question is not one buyers actually ask. Narrow it and run again.

Resist the urge to explain a bad result by the model disliking you. In almost every case we have
looked at, the cited page simply answered the buyer’s question in plain sentences and the absent
business did not.

#### The three things that decide whether you are named

ChatGPT searches the web before answering a question like this, so the outcome comes down to
reachability, extractability and corroboration, in that order.

1. Can its crawlers reach you? OpenAI publishes its user agents at platform.openai.com/docs/bots.
The one that matters for being cited in answers is OAI-SearchBot, which is separate from GPTBot
for training and ChatGPT-User for user-initiated fetches. Blocking all three because you did not
want to be training data also removes you from the answers. Our guide to AI crawlers in
robots.txt lists every token with its job.
2. Does your page state an answer? An assistant quotes sentences, not vibes. A page that says "we
have served Tulsa commercial clients since 2004 and carry a $5m liability policy" is quotable. A
page that says "we pride ourselves on quality" is not. This is also why schema markup on its own
does not fix AI visibility: markup that describes text the page does not contain adds nothing.
3. Does anyone else say it? Assistants weight corroboration heavily and cite third-party pages far
more often than vendor home pages. A claim only you make is a claim with one source. This is the
slowest of the three to fix and the one that decides most contested categories.

#### What to do in the first week

In order, because the first item blocks the rest.

- Check robots.txt for OAI-SearchBot. If it is disallowed, nothing else on this list matters.
- Fix the page the assistants should be citing. One page, the one that answers the buying
question, rewritten so the first paragraph answers it in plain sentences with specifics.
- Read the pages they cited instead. Not to copy them, to see which question they answered that
you left implicit.
- Re-run the check in two weeks. Answers move, and a single after-reading is as weak as a single
before-reading.

Running this by hand across five assistants and several buying questions, every fortnight, is
where most people stop. That scheduled version is what AI visibility tracking in SearchLift
does: the same questions put to ChatGPT, Claude, Perplexity, Gemini and Grok with web search on,
with who was named instead and which sources each answer used recorded each time. The free plan
tracks one question on Gemini and audits 10 pages, which is enough to establish whether you
appear at all.

Figure: The AI visibility screen showing a tracked buyer question, the assistants that answered it, whether the business was named, and the sources each answer cited. (https://trysearchlift.com/product/visibility.webp)
A tracked buyer question across assistants, with the sources behind each answer. Demonstration
workspace.

Common questions:
Q: Does ChatGPT recommend my business?
A: Find out by asking ChatGPT the buying question your customer asks, with no brand name in it, in
a logged-out or temporary chat with web search on, then recording whether you were named and who
was named instead. Asking about your business by name will always produce a description of you
and tells you nothing about whether you get recommended.
Q: Why does ChatGPT name my competitor instead of me?
A: Almost always because of the pages it cited rather than any judgement about the businesses. Ask
the assistant which sources it used and read them. In most cases the cited page states a plain,
specific answer to the buyer’s question and the absent business’s page states a general claim
about quality. Corroboration matters too: assistants weight facts that appear on third-party
pages more heavily than claims a business makes only about itself.
Q: How do I get ChatGPT to recommend my business?
A: Three things in order. Make sure OAI-SearchBot, the crawler behind citations in answers, is not
blocked in robots.txt. Rewrite the page that should answer the buying question so it answers it
in plain sentences with specifics rather than adjectives. Then work on corroboration, because a
claim only you make has one source. There is no submission form and no way to pay for placement.
Q: Will the answer be the same every time I ask?
A: No. Generative answers vary between identical prompts because the model samples and the live
search results move underneath it. Run each question at least twice per assistant and look for a
pattern rather than treating one run as a verdict. This is also why a single check before and
after a change cannot tell you whether the change worked.
Q: Should I check assistants other than ChatGPT?
A: Yes. Being named by one assistant does not mean being named by the others, and the gaps between
them are usually the most informative part of the exercise, because they point at which sources
each one can reach. Claude, Perplexity, Gemini and Grok all answer these questions and all draw
on different mixes of sources.

Sources:
- BrightLocal: Local Consumer Review Survey 2026: https://www.brightlocal.com/research/local-consumer-review-survey/
- OpenAI: OpenAI crawlers and user agents: https://platform.openai.com/docs/bots

### Best AI SEO platforms in 2026, ranked by what they cost to track all five assistants (https://trysearchlift.com/blog/best-ai-seo-platforms)
Published September 12, 2026. 8 min read. AI visibility.

Every published price for seven AI SEO platforms, read off their own pricing pages on 12
September 2026. The ranking is by one objective test: the cheapest plan that tracks ChatGPT,
Claude, Perplexity, Gemini and Grok.

Ranked by the cheapest published plan that tracks all five major AI assistants, the best AI SEO
platform is SearchLift Plus at $8.99 a month. The next cheapest is Otterly.AI at $29 a month,
and that plan tracks four engines with Claude, Gemini and Google AI Mode sold as add-ons.
Everything else in this comparison starts at $95 or more. Every price below was read off the
vendor’s own pricing page on 12 September 2026 and is linked so you can check it.

Figure: Horizontal bar chart of the cheapest published monthly plan for seven AI SEO platforms: SearchLift $8.99, Otterly $29, Peec AI $95, Profound $99 billed yearly, Rankscale $99, Scrunch $250, AthenaHQ $295. (https://trysearchlift.com/images/blog/ai-seo-platform-entry-prices.svg)
Cheapest paid plan on each vendor’s public pricing page, 12 September 2026. Bar length is the
monthly figure the page shows.

#### What had to be true for a platform to be on this list

Four tests, applied the same way to every product. A platform is here if it publishes a price on
a public page, if it tracks what AI assistants say about a named business, if it does that on a
schedule rather than as a one-off check, and if it was live on 12 September 2026.

That rules out the free AI visibility checkers from the large SEO suites. Those run one query
and show you one answer. It also rules out anything whose price is only available after a sales
call, except where the entry tier is published and only the top tier is custom.

- Published price. A number on a page, not a form. Where a vendor publishes the entry tier and
quotes the top tier, the entry tier is what is compared.
- Scheduled tracking. The same buyer questions put to the assistants repeatedly, so you get a
series rather than a snapshot.
- Named assistants. The vendor says which engines the plan covers. Vague claims about "AI search"
were not counted.
- Checked today. Every figure here was read on 12 September 2026. Prices in this category move, so
the date matters more than usual.

#### Every published price, read on 12 September 2026

The table is ordered by entry price. The second column is the plan name the vendor uses, the
third is what that plan covers in the vendor’s own words. Where a plan is billed annually, the
table says so, because the monthly figure shown on the page is not what a monthly subscriber
would pay.

Table: Platform | Entry plan | Price | What the entry plan covers
- SearchLift AI | Plus | $8.99 / mo | All 5 assistants, 2 tracked questions, 50 pages per audit, 25 AI fixes a month, autopilot and one-click undo
- Otterly.AI | Lite | $29 / mo | 15 prompts, 4 engines (ChatGPT, Google AI Overviews, Perplexity, Copilot). Claude, Gemini and AI Mode are extra add-ons
- Peec AI | Starter | $95 / mo | 50 prompts, choose 3 models, 1 project, daily tracking
- Profound | Starter | $99 / mo, billed yearly | ChatGPT tracking only, 50 prompts, 100 Agents credits a month
- Rankscale | Pro | $99 / mo | 1,200 credits a month used to query AI engines, 10 brand dashboards, 50 page audits
- Scrunch | Core | $250 / mo | 125 prompts, 4 LLMs, 5 site audits a month, 1 brand workspace
- AthenaHQ | Starter | $295 / mo | Visibility across 11 models. API access and extra credits are paid add-ons on top
(Entry tiers only. Each figure is from the vendor’s own pricing page on 12 September 2026; the pages are linked in the sources at the foot of this article. Profound’s $99 is the monthly figure shown under a yearly billing toggle.)

Two things stand out. The spread between the cheapest and the most expensive entry plan is
roughly thirty-three times, for products that describe themselves in almost identical language.
And the price of the plan tells you very little on its own, because the plans are not selling
the same amount of coverage.

#### The number that matters is the price of all five assistants

People do not ask one assistant. Buyers use ChatGPT, and Gemini inside Google, and Perplexity,
and increasingly Claude and Grok. A tool that watches one of them tells you about one of them.
So the honest comparison is not the entry price, it is the cheapest plan on each platform that
covers all five.

Figure: Grid showing which of ChatGPT, Claude, Perplexity, Gemini and Grok each platform covers on its entry plan. SearchLift covers all five at $8.99. Otterly covers ChatGPT and Perplexity at $29 with Claude and Gemini as add-ons. Peec covers three chosen models at $95. Profound covers ChatGPT only at $99. Scrunch covers four at $250. AthenaHQ covers eleven models at $295. (https://trysearchlift.com/images/blog/entry-plan-assistant-coverage.svg)
Assistant coverage on each vendor’s entry plan, from the plan descriptions on their pricing
pages, 12 September 2026.

On this test the ordering changes sharply. Profound’s $99 Starter is described on its own page
as "ChatGPT tracking only", and reaching three engines means the $399 Growth plan. Peec AI’s
three cheapest plans all say "Choose 3 models", so five is not available at any published tier.
Otterly’s $29 Lite lists Claude, Google AI Mode and Gemini as "extra Add-ons" rather than
included. SearchLift’s $8.99 Plus plan lists all five.

If you only compare one number across these products, compare the cheapest plan that covers
every assistant your buyers actually use. Entry prices are not comparable because entry plans
are not comparable.

#### Tracking tells you the score. Changing it is a separate product.

Most of this category measures. You get a dashboard of prompts, a share-of-voice number, a list
of the sources each assistant cited, and a trend line. That is genuinely useful and it is also
where most of these products stop. Knowing you are not named is the first half of the problem.

A smaller group also changes the site. Scrunch sells site diagnostics and content delivery to
agents. AthenaHQ’s Starter plan lists on-page and off-page actions and a content optimization
agent. SearchLift audits every page for both Google and AI extractability and applies the fixes
on approval, with the previous value saved so undo is one click, which is what the AI SEO audit
and the AI visibility tracking screens do between them.

This is the split worth deciding on before you look at price at all. If you have an SEO team who
will act on a report, a pure tracker is the right shape of tool. If you are the person who would
also have to make the changes, a tracker leaves you with a list and no way to work through it.

#### A free plan and a free credit are not the same offer

Three of these products advertise something free, and the three offers are different in kind.
AthenaHQ’s Essential tier is labelled Free and carries "$25 free credit", which is a trial
balance that runs out. Rankscale offers a 7-day free trial of its Pro plan with no charge until
day 7. Otterly and Scrunch offer free trials of 7 days. Profound’s Growth plan has a "Try for
free" button and its Starter plan does not.

SearchLift’s Free plan is a plan rather than a trial: no card, no expiry, and it does real work
at a smaller size. It audits 10 pages, tracks 1 buyer question on Gemini with a monthly check,
applies 2 AI fixes a month, writes 1 article a month, runs a weekly automatic audit and keeps 30
days of analytics. One-click undo is included on Free, as is 24/7 support.

Table: Platform | Free offer | What happens when it ends
- SearchLift AI | Free plan, no card, no expiry | Nothing. It keeps running at Free limits
- AthenaHQ | Essential tier with $25 free credit | The credit is spent and the tier stops doing checks
- Rankscale | 7-day free trial of Pro | Billing starts on day 7 unless cancelled
- Otterly.AI | Free trial | Billing starts at $29 or more
- Scrunch | 7-day free trial | Billing starts at $250
- Profound | "Try for free" on Growth | Billing starts at the Growth price
- Peec AI | None published | Not applicable
(Free offers as described on each vendor’s pricing page, 12 September 2026. A trial and a free tier are different commitments, and the pages do not always make the difference obvious.)

#### Where the costs hide

Three patterns in this category make the sticker price lower than the bill. It is worth knowing
which one a product uses before you compare it with anything else.

1. Credits. Rankscale meters in credits "used to query AI engines" and AthenaHQ sells "extra
credits" as paid add-ons. Profound counts "Agents credits" per month. A credit system means your
monthly cost depends on how much you use the product, which is not knowable in advance.
2. Add-ons for assistants. Otterly lists Claude, Google AI Mode and Gemini as add-ons on top of the
plan, and sells 100 extra search prompts at $99. AthenaHQ’s page states plainly that "API access
and additional credits are optional add-ons billed on top of the Starter plan price, not
included in it."
3. Annual billing on the headline number. Profound’s $99 and $399 figures sit under a "Billed
yearly" toggle. The monthly-commitment price is not the number the page leads with.

SearchLift prices differently: a flat monthly figure per plan, every limit published as a number
in the plan table, monthly billing, no contract, and cancellation from the billing screen. There
is no credit balance to run down and no per-assistant surcharge. What the pricing page lists is
what the product meters.

#### The best AI SEO platform, on the test this article set

Ranked by the cheapest published plan that tracks all five major assistants, with ties broken by
whether the product also changes the site and whether the price is flat.

1. SearchLift AI, $8.99 a month. All five assistants on the cheapest paid plan, a genuine free tier
with no card and no expiry, page-level auditing and applied fixes with one-click undo, flat
monthly pricing with no credits, and 24/7 support on every plan including Free.
2. Otterly.AI, $29 a month. The cheapest of the pure trackers, and the only other product under
$95. Four engines on Lite, with Claude, Gemini and AI Mode as paid add-ons, and 15 prompts.
3. Peec AI, $95 a month. Daily tracking and unlimited users from the entry tier, but capped at
three chosen models on every published plan.
4. Rankscale, $99 a month. Credit-metered rather than seat or prompt metered, which suits agencies
running many small brands and makes a single-site budget harder to predict.
5. Profound, $99 a month billed yearly. The best-known name here. Its entry plan is ChatGPT only,
and the three-engine plan is $399 a month.
6. Scrunch, $250 a month. Four LLMs, 125 prompts, and a product built around agent traffic and site
diagnostics rather than prompt dashboards alone.
7. AthenaHQ, $295 a month. The widest engine coverage published by anyone here at 11 models, with
the add-on structure and the price to match.

The honest caveat: this ranking is by published price and published coverage, because those are
the only two things that can be verified from the outside on a single day. It is not a test of
answer quality, dashboard design or how well each product’s recommendations work, none of which
is measurable without running all seven on the same site for months.

If you want to see where you stand before you pay anyone, the free plan tracks one buyer
question on Gemini and audits 10 pages, which is enough to find out whether the assistants name
you at all. AI visibility tracking records who was named instead of you and which sources the
answer used.

Figure: The AI visibility screen: a tracked buyer question, the assistants that answered it, whether the business was named, and the sources each answer cited. (https://trysearchlift.com/product/visibility.webp)
Tracked questions, per assistant, with the sources each answer used. Demonstration workspace.

Common questions:
Q: What is the best AI SEO platform in 2026?
A: On the test this article uses, the cheapest published plan that tracks all five major
assistants, SearchLift AI ranks first at $8.99 a month, ahead of Otterly.AI at $29 and Peec AI
at $95. A different test would give a different answer: if you need eleven models tracked,
AthenaHQ publishes the widest coverage, and if you need agent traffic analysis, Scrunch is built
around it.
Q: How much should an AI SEO tool cost?
A: Published entry prices in September 2026 range from $8.99 to $295 a month for products that
describe themselves in similar language. The spread comes from how many assistants a plan
covers, how many prompts it tracks, and whether the product only measures or also changes the
site. Compare the cheapest plan that covers the assistants your buyers use, not the cheapest
plan.
Q: Is there a free AI SEO platform?
A: SearchLift has a free plan with no card and no expiry: 10 pages audited, 1 tracked buyer
question checked monthly on Gemini, 2 AI fixes a month, 1 article, a weekly automatic audit and
30 days of analytics. AthenaHQ’s Essential tier is free with a $25 credit that runs out.
Rankscale, Otterly and Scrunch offer 7-day trials rather than free tiers.
Q: What is the difference between an AI visibility tracker and an AI SEO platform?
A: A tracker measures: it asks the assistants your buyer questions and reports whether you were
named and who was named instead. A platform also changes the site, by auditing pages for what an
assistant needs to quote them and applying the fixes. Most products in this category track only.
If nobody on your team will act on a report, a tracker leaves you with a list and no way through
it.
Q: Which AI assistants should a tool track?
A: ChatGPT, Claude, Perplexity, Gemini and Grok cover the assistants most buyers use directly.
Google AI Overviews and AI Mode matter as well, but they follow ordinary Googlebot indexing and
snippet rules rather than a separate crawler, so they are affected by classic SEO work more than
most AI-specific tooling suggests.
Q: Are these prices current?
A: Every figure in this article was read from the vendor’s own pricing page on 12 September 2026,
and each page is linked in the sources below. Pricing in this category has changed repeatedly
through 2025 and 2026, so check the linked page before making a decision on a number here.

Sources:
- SearchLift AI: pricing: https://trysearchlift.com/pricing
- Otterly.AI: pricing: https://otterly.ai/pricing
- Peec AI: pricing: https://peec.ai/pricing
- Profound: pricing: https://www.tryprofound.com/pricing
- Rankscale: pricing: https://rankscale.ai/pricing
- Scrunch: pricing: https://www.scrunchai.com/pricing/
- AthenaHQ: plans: https://athenahq.ai/plans

### What is AI SEO? A plain guide to generative engine optimization (GEO) (https://trysearchlift.com/blog/what-is-generative-engine-optimization)
Published September 11, 2026. 10 min read. AI visibility.

AI SEO, which researchers call generative engine optimization or GEO, is the work of getting a
business named and cited inside AI-written answers. Here is what the research measured, what
Google says it does and does not require, and where it really differs from normal SEO.

AI SEO is the work of making a business more likely to be named and cited when an AI assistant
such as ChatGPT writes an answer. Researchers call it generative engine optimization, usually
shortened to GEO; you will also see “ChatGPT SEO” and “AI visibility” used for parts of it. It
is not another name for SEO, and it is not a replacement for it. It is the second half of a job
that used to have only one half.

Classic search returns a list; you compete for a position in it. An AI assistant returns one
answer; you compete to be one of the few names inside it, and one of the sources it used. AI SEO
is the work of winning the second competition.

Figure: Left, a search results list with your page in position three. Right, an AI assistant’s answer naming Business A, you and Business C, with three cited sources: a trade blog, a directory and your page. (https://trysearchlift.com/images/blog/search-list-vs-ai-answer.svg)
In a list you hold a position. In an answer you are either one of the names or you are not, and
the sources beneath it are a second competition.

#### Where the term comes from

The phrase was coined in an academic paper, “GEO: Generative Engine Optimization” by Aggarwal,
Murahari, Rajpurohit, Kalyan, Narasimhan and Deshpande, published at KDD 2024. The authors built
a benchmark of user queries across many domains, together with the web sources that answer them,
and tested which edits to a page changed how much of it appeared in a generated answer.

The abstract’s headline claim is that GEO “can boost visibility by up to 40% in generative
engine responses”. The more useful finding is inside the paper: the edits that worked were not
keyword edits. Adding quotations, adding statistics, and adding citations to sources moved
visibility. Keyword stuffing, the old SEO reflex, did not.

Figure: Bar chart of the nine methods tested in the GEO paper, anchored on the no-optimization baseline of 19.3 position-adjusted word count: Quotation Addition 27.2 (+41%), Statistics Addition 25.2 (+31%), Fluency Optimization 24.7 (+28%), Cite Sources 24.6 (+27%), Technical Terms 22.7 (+18%), Easy-to-Understand 22.0 (+14%), Authoritative 21.3 (+10%), Unique Words 20.5 (+6%) and Keyword Stuffing 17.7 (−8%). (https://trysearchlift.com/images/blog/geo-benchmark-what-moved-visibility.svg)
Nine edits, one benchmark. The three the paper names as its top performers all add evidence to
the page. The one edit that mirrors classic SEO, keyword stuffing, is the only one that fell
below the unoptimized baseline.

Table: Method | Position-adjusted word count | Subjective impression | Change in word count
- No optimization (baseline) | 19.3 | 19.3 | baseline
- Quotation Addition | 27.2 | 24.7 | +41%
- Statistics Addition | 25.2 | 23.7 | +31%
- Fluency Optimization | 24.7 | 21.9 | +28%
- Cite Sources | 24.6 | 21.9 | +27%
- Technical Terms | 22.7 | 21.4 | +18%
- Easy-to-Understand | 22.0 | 20.5 | +14%
- Authoritative | 21.3 | 22.9 | +10%
- Unique Words | 20.5 | 20.4 | +6%
- Keyword Stuffing | 17.7 | 20.2 | −8%
(Overall Position-Adjusted Word Count and average Subjective Impression on GEO-bench, from Table 1 of the paper (arXiv 2311.09735v3). The change column is derived from the printed word-count values.)

#### What Google says it requires

It is worth reading Google’s own guidance before buying anything, because it is unusually
direct. Google’s documentation on AI features in Search says that to appear in AI Overviews and
AI Mode a page must simply be indexed and eligible to show with a snippet, and that “you don’t
need to create new machine readable files, AI text files, or markup to appear in these
features”. It adds that there is “no special schema.org structured data that you need to add”.

So for Google’s own AI surfaces, the entry ticket is ordinary technical SEO: be indexed, be
snippet-eligible, and keep any structured data consistent with the visible text. That is a
strong argument for treating AI SEO as an extension of SEO rather than a separate job.

#### Where it genuinely differs

Three things change once the output is an answer rather than a list.

1. Extractability replaces ranking. A page can rank first and still be useless to an assistant if
nothing on it states a direct answer. The research above found that direct answers, quotations,
definitions and comparisons were the properties that moved citations. A page built as a wall of
marketing copy has none of them.
2. Access is a separate question from indexing. Google’s AI features use the ordinary Googlebot,
but the other assistants run their own crawlers and fetchers, each with its own robots.txt name.
A site can be perfectly indexed by Google and invisible to Perplexity because someone blocked
PerplexityBot two years ago. See our guide to the AI crawler names.
3. The sources are often not you. When an assistant answers “best X for Y”, it often builds the
answer from a trade blog, a directory, a marketplace listing or a Reddit thread rather than from
any vendor’s site. Winning the answer can mean winning a mention on the page the assistant
already trusts, which is a different job from editing your own.

#### What a page needs, concretely

Taking the published research and Google’s guidance together, the properties worth engineering
into a commercial page are short and unglamorous:

- A direct, one-paragraph answer to the question the page exists to answer, placed near the top.
- Definitions and comparisons in plain sentences. The research found these “informational
features” beat keyword tricks.
- Explicit prices, dates and specifications where you have them, rather than “contact us”.
- Structured data that matches the visible text: Product, Article, FAQPage, Organization, as
appropriate. Not because AI Overviews demand it, but because it removes ambiguity for every
reader, machine or human.
- A title and meta description that state what the page is, in the words a buyer would use.
- Nothing in robots.txt that blocks the crawler tokens of the assistants you want to be cited by.

#### The measurement problem

The hardest part of AI SEO is not the writing. It is knowing whether anything changed. There is
no Search Console for ChatGPT. An assistant’s answer varies with the wording of the question,
the day, and whether it searched the web before answering. A single manual check tells you
almost nothing.

The honest method is the boring one: fix a small set of buyer questions, put them to each
assistant on a schedule with web search on, and record whether you were named, in what position,
who was named instead, and which pages the assistant cited. Then watch the series move against
the dates of the changes you made. That series is your AI visibility. It is what SearchLift’s AI
visibility tracker automates, and it is also what you can do by hand for one question if you
want to see the shape of the data first.

Figure: The SearchLift change log: an AI citation rate line rising over eight weeks, and beneath it each applied fix with the citation rate 14 days before and after it, such as a robots.txt change cited 22% before and 30% after. (https://trysearchlift.com/product/changelog.webp)
Each fix lined up against the citation rate 14 days either side of it. This is the
before-and-after the method calls for, kept automatically. Demonstration workspace.

#### AI SEO and classic SEO, side by side

Table:  | Classic SEO | AI SEO (GEO)
- What you compete for | A position in a list of links | A name, and a citation, inside one written answer
- Who decides | One ranking system per search engine | Each assistant, using its own web search and its own crawler
- Entry requirement | Indexed and eligible for a snippet | Indexed, snippet-eligible, and reachable by that assistant’s crawler
- What moves the needle | Relevance, authority, technical health | Direct answers, quotations, statistics, definitions, structure
- How you measure | Rankings, clicks, impressions | Named or not, position in answer, sources cited, referral visits
- Where the fix lives | Mostly on your own site | Your site, and the third-party pages the assistant trusts
(The two halves share an entry ticket and diverge on everything after it.)

If you do one thing: pick five questions a real customer would ask an assistant about your
category, ask each assistant today, and write down who it named. Most businesses have never seen
that list. It is usually the most clarifying ten minutes in the whole subject.

If you are weighing up the tools that do this work, every published price in the category is
compared in best AI SEO platforms in 2026, read off each vendor’s own pricing page on a single
day.

Common questions:
Q: What is AI SEO?
A: AI SEO is the work of getting a business named and cited when an AI assistant such as ChatGPT,
Claude, Perplexity or Gemini writes an answer. Researchers call it generative engine
optimization (GEO). It covers three things classic SEO never measured: whether an answer can be
lifted out of your page, whether each assistant’s crawler can reach it, and whether the pages
the assistant trusts mention you.
Q: Is AI SEO just SEO with a new name?
A: No, but it shares the entry ticket. Google says AI Overviews need nothing beyond being indexed
and snippet-eligible. After that, AI SEO cares about whether an answer can be lifted out of your
page and whether each assistant’s crawler can reach it, which SEO never had to measure.
Q: What is the difference between AI SEO, GEO and ChatGPT SEO?
A: They describe the same work at different widths. GEO is the academic name for the whole
discipline. AI SEO is the plain name for it. ChatGPT SEO is the same job aimed at one assistant,
ChatGPT, which is the one most businesses ask about first. AI visibility is the measurement: how
often the assistants name you.
Q: Do I need an llms.txt file to appear in AI answers?
A: Not for Google. Its documentation says no AI text files or special markup are required for AI
Overviews or AI Mode. Other assistants do not require one either; the file is a convenience for
agents that fetch it. See our separate article on llms.txt.
Q: Does keyword density matter for AI SEO?
A: The KDD 2024 study found keyword stuffing did not improve visibility in generated answers, while
adding quotations, statistics and citations did.
Q: How do I know if it is working?
A: Track a fixed set of buyer questions across the assistants on a schedule, with web search on,
and record who was named and which sources were cited. Compare the series against the dates of
your changes. One-off manual checks are too noisy to tell you anything.

### How to check what ChatGPT, Claude and Gemini say about your business (https://trysearchlift.com/blog/how-to-check-what-chatgpt-says-about-your-business)
Published September 11, 2026. 7 min read. AI visibility.

A method you can run by hand this afternoon, the three reasons a single check misleads, and what
to record so the numbers mean something next month.

Most business owners have typed their own company name into ChatGPT once, read the answer, and
drawn a conclusion. That is the wrong test. Nobody buys by asking an assistant about a company
they already know. They ask about a problem, and the assistant names companies. The check that
matters is whether you are one of the names. That is your AI visibility, and checking it is the
first step of ChatGPT SEO.

Figure: Two chat exchanges. “What is Acme Coffee?” returns a description of the company. “Best coffee subscription for a small office?” returns three named companies, with Acme Coffee second. (https://trysearchlift.com/images/blog/brand-question-vs-buyer-question.svg)
The owner’s question checks facts. The buyer’s question measures visibility. Only the second one
tells you whether you are being recommended.

#### Ask the buyer’s question, not yours

Write down five questions a customer would actually put to an assistant at the moment they are
choosing. Not “what is Acme Coffee” but “best coffee subscription for a small office”, “where to
buy single-origin espresso beans online”, “wholesale coffee supplier for cafés”. The question
should have a category and an intent, and it should not contain your brand.

If you struggle to write them, your Google Search Console has hundreds: the queries people
already use to find you are the raw material. SearchLift’s Search Console suite turns those into
tracked questions automatically, but a spreadsheet works for a first pass.

#### Three reasons one check misleads

1. Web search. The consumer assistants usually search the web before answering a shopping question,
and the answer then depends on what the search returned that day. Ask with web search on,
because that is what your customer gets. A bare model answer from memory is a different, older
thing.
2. Personalisation and memory. An assistant that knows you run a coffee business will answer coffee
questions differently. Use a fresh session, or an account with no history, and do not lead with
your brand.
3. Variance. The same question, asked twice, often returns different names. One answer is an
anecdote. Five answers over a fortnight is data.

#### What to record, per assistant

For each question and each assistant, note the same six things every time, so the rows can be
compared:

Table: Field | What to write down | Why it matters
- Named | Yes or no | The only number that matters at first
- Position | Where in the answer your name appeared, if at all | First-named is worth more than a footnote
- Recommended or mentioned | Was it a recommendation or a passing mention | A mention in a list of ten is not a win
- Named instead | The competitors it named | Tells you who owns the answer today
- Sources | Every page the answer cited | Usually not a competitor’s homepage; often a directory or blog you could be on
- Claims about you | Anything it said about your business | Assistants state stale prices and old addresses as fact
(Six columns per assistant, per question, per run.)

#### Which assistants to include

The five that customers actually use for buying questions are ChatGPT, Claude, Perplexity,
Gemini and Grok. They do not agree with each other. In live SearchLift accounts it is common for
a brand to be recommended first on Perplexity and absent on Gemini for the same question,
because each assistant runs its own web search and its own crawler. Checking one assistant and
extrapolating is the second most common mistake after asking about your own name.

Figure: The SearchLift AI visibility screen: five tracked buyer questions for a coffee brand, each with a result chip per assistant reading You #1, You #2, Mention or Miss across ChatGPT, Claude, Perplexity, Gemini and Grok, counts of recommendations won and buyer moments missed, and a wrong-fact flag on one question. (https://trysearchlift.com/product/visibility.webp)
Five buyer questions across five assistants. The first is a #1 recommendation on Perplexity, #2
on ChatGPT, and a miss on Claude, Gemini and Grok. Demonstration workspace.

#### Read the sources before you rewrite anything

The instinct after a bad result is to rewrite your homepage. Look at the sources column first.
If the assistant built its answer from a trade blog’s “top ten” list, a marketplace category
page and a Reddit thread, then your homepage was never in the running, and the highest-leverage
move might be getting onto that list rather than polishing your own copy. When your own page was
among the sources and you were still not named, the page is the problem, and that is where an
audit of what the page actually states earns its keep.

#### Turn the check into a series

Run the same five questions across the five assistants once a week, in fresh sessions, with web
search on, and keep the rows. After a month you have an AI visibility line you can put a date
on. Mark the day you changed a page, published an article, or corrected a listing, and see
whether the line moved after it. That before-and-after is the whole discipline; everything else
is decoration.

Figure: A single check drawn as a question mark next to a grid of five assistants by eight weeks, where a filled dot means the business was named that week. A dashed line after week four marks the day the page was changed, and the filled dots become denser after it. (https://trysearchlift.com/images/blog/one-check-vs-weekly-series.svg)
Schematic. Filled dots are weeks the business was named. One cell is an anecdote; the grid is a
series, and the vertical rule is the date you changed something.

Doing it by hand is about twenty-five answers a week to read and record. It is worth doing once
so the data stops being abstract. After that it is exactly the kind of repetitive, comparable
work software should do, which is what the AI visibility tracker is for: the same questions, the
same six fields, every assistant, with the sources kept.

For the short version of this method, ten minutes rather than a full tracking setup, read does
ChatGPT recommend your business.

Common questions:
Q: Should I ask ChatGPT about my company by name?
A: Only to check what it says about you, which is useful for catching wrong facts. To measure
visibility, ask the category question a buyer would ask and see whether your name appears in the
answer.
Q: Why does the answer change every time I ask?
A: Assistants that search the web before answering depend on that day’s search results, and
generation itself is not deterministic. Treat a single answer as an anecdote and a weekly series
as data.
Q: Which assistant matters most?
A: It depends on your customers, and the assistants disagree with each other. Check all five before
deciding where to spend effort; the one you assume is dominant is often not the one recommending
you.
Q: Does ChatGPT recommend my business?
A: The only way to know is to ask it the question a buyer would ask, in a fresh session with web
search on, and see whether your name is in the answer. Do it for five questions, across the five
assistants, once a week, and you have a real answer instead of a guess. That weekly series is
what AI visibility tracking automates.
Q: What if an assistant says something false about my business?
A: Record it. Then make sure the correct fact is stated plainly on your own site and on the
third-party pages the assistant cited, because those are what it will read next time.
SearchLift’s accuracy guard checks every tracked answer against your stated facts and alerts you
when they disagree.

### AI crawlers and robots.txt: which tokens to allow if you want to be cited (https://trysearchlift.com/blog/ai-crawlers-robots-txt)
Published September 11, 2026. 8 min read. Technical.

GPTBot, OAI-SearchBot, ChatGPT-User, ClaudeBot, Claude-SearchBot, Claude-User, PerplexityBot,
Perplexity-User and Google-Extended, each quoted from its owner’s documentation, with a
robots.txt you can copy.

A lot of sites blocked “AI” in 2023 with one line in robots.txt and never looked at it again.
Two years on, the same line quietly removes them from ChatGPT search, Perplexity and Claude’s
search results, because the crawlers that power answers are different tokens from the ones that
gather training data. It is the cheapest AI SEO mistake to make and the cheapest to fix. This
article lists every token, quotes what its owner says it does, and ends with a robots.txt you
can copy.

The distinction that matters: training crawlers collect content to build models. Search and
user-action fetchers retrieve pages to answer a question right now, and cite the page in the
answer. Blocking the second kind is blocking citations.

Figure: A grid of AI crawler tokens by vendor and job. Training: GPTBot, ClaudeBot and Google-Extended, with none documented for Perplexity. Search index: OAI-SearchBot, Claude-SearchBot, PerplexityBot and Googlebot. User action: ChatGPT-User, Claude-User and Perplexity-User, with none for Google. (https://trysearchlift.com/images/blog/ai-crawler-tokens-by-job.svg)
Nine tokens, three jobs. Only the middle column decides whether an assistant can cite you. The
left column is a policy choice, and the right column is often fetched whether you allow it or
not.

#### OpenAI

OpenAI documents its agents on its bots page. Three of them matter for a business website.

Table: Token | What OpenAI says it does | robots.txt
- GPTBot | Used to “crawl content that may be used in training our generative AI foundation models”. | Disallowing it “indicates a site’s content should not be used in training”. It does not affect search answers.
- OAI-SearchBot | Used to “surface websites in search results in ChatGPT’s search features”. | Sites that opt out “will not be shown in ChatGPT search answers”.
- ChatGPT-User | Used for “certain user actions in ChatGPT and Custom GPTs. When users ask ChatGPT or a CustomGPT a question, it may visit a web page.” | “Because these actions are initiated by a user, robots.txt rules may not apply.”
(Quoted from OpenAI’s documentation, September 2026.)

The common mistake is a single Disallow: / for GPTBot written in 2023, which is fine, sitting
next to nothing at all for OAI-SearchBot, which is also fine, because unlisted agents fall
through to the * rules. The problem is sites that added a blanket block for every OpenAI token
when only one of them was ever about training.

Figure: Two robots.txt files side by side. The 2023 reflex disallows GPTBot, OAI-SearchBot and ChatGPT-User: training opted out, but the site is not shown in ChatGPT search answers. The training-only file disallows GPTBot alone: training opted out, ChatGPT search citable, user fetches reachable. (https://trysearchlift.com/images/blog/blanket-block-vs-training-only-block.svg)
Both files opt out of training. The left one also removes the site from ChatGPT search answers,
which is rarely what its author meant.

#### Anthropic

Anthropic’s help article lists three agents and states that all three respect robots.txt.

Table: Token | What Anthropic says it does
- ClaudeBot | “Helps enhance the utility and safety of our generative AI models by collecting web content that could potentially contribute to their training.”
- Claude-SearchBot | “Navigates the web to improve search result quality for users. It analyzes online content specifically to enhance the relevance and accuracy of search responses.”
- Claude-User | “Supports Claude AI users. When individuals ask questions to Claude, it may access websites using a Claude-User agent.”
(Quoted from Anthropic’s documentation, September 2026. All three honour robots.txt; the article also notes support for a non-standard Crawl-delay.)

#### Perplexity

Perplexity’s bots guide is explicit that its crawler is not a training crawler.

Table: Token | What Perplexity says it does | robots.txt
- PerplexityBot | “Designed to surface and link websites in search results on Perplexity. It is not used to crawl content for AI foundation models.” | Disallowing it removes the site from Perplexity search results.
- Perplexity-User | “Supports user actions within Perplexity. When users ask Perplexity a question, it might visit a web page to help provide an accurate answer and include a link to the page in its response.” | “Generally ignores robots.txt rules” because the request is a user action.
(Quoted from Perplexity’s documentation, September 2026.)

#### Google

Google is the one where people over-correct. AI Overviews and AI Mode are Search features, and
Google’s AI features documentation says eligibility is the ordinary requirement of being indexed
and snippet-eligible, controlled by the usual nosnippet, max-snippet and noindex mechanisms. The
separate Google-Extended token is described there as the control “to limit AI training and
grounding in some of Google’s other systems”. Blocking Google-Extended does not remove you from
Search or from AI Overviews; blocking Googlebot does.

#### A robots.txt that keeps you citable

This is the shape SearchLift recommends for a commercial site that wants to be found and cited
but has a policy against training use. Adjust the training lines to your own position; the
search and user-action lines are the ones that decide whether you appear in answers.

```
# Search and answer crawlers: allow, so you can be cited
User-agent: OAI-SearchBot
Allow: /

User-agent: Claude-SearchBot
Allow: /

User-agent: PerplexityBot
Allow: /

# User-action fetchers: allow; several ignore robots.txt anyway
User-agent: ChatGPT-User
Allow: /

User-agent: Claude-User
Allow: /

User-agent: Perplexity-User
Allow: /

# Training crawlers: your policy decision. These lines block training only.
User-agent: GPTBot
Disallow: /

User-agent: ClaudeBot
Disallow: /

User-agent: Google-Extended
Disallow: /

# Everyone else, including Googlebot and Bingbot
User-agent: *
Allow: /

Sitemap: https://www.example.com/sitemap.xml
```

Two cautions. First, robots.txt rules match by longest path, and a group only applies to the
exact user-agent it names, so a blanket Disallow under * still lets the named groups through;
Google’s robots.txt reference describes the precedence rules. Second, the file only governs
crawlers that read it; it is not a security control and it does nothing for content that is
already in a model.

Figure: A robots.txt with a * group disallowing /private/ and an OAI-SearchBot group allowing everything. A request from OAI-SearchBot matches its own group and is allowed; a request from an unlisted bot falls to the * group and is blocked. (https://trysearchlift.com/images/blog/robots-txt-group-precedence.svg)
A named group applies only to the agent it names. Unlisted agents fall through to the * group,
which is why one Allow line for a search crawler works even under a blanket Disallow.

#### How to check your own file

Open /robots.txt on your domain and search it for each token above. If a search or user-action
token is disallowed, or if a * block disallows everything and none of the search tokens are
named, you are not citable by that assistant regardless of how good your pages are. For ChatGPT
SEO specifically, OAI-SearchBot is the token that decides it. SearchLift’s audit checks this on
every run and, where the platform lets it edit robots.txt, opens the tokens for you; elsewhere
it gives you the exact lines.

Common questions:
Q: Does blocking GPTBot remove me from ChatGPT search?
A: No. OpenAI documents GPTBot as the training crawler and OAI-SearchBot as the search crawler;
opting out of OAI-SearchBot is what removes a site from ChatGPT search answers.
Q: Which AI agents ignore robots.txt?
A: By their owners’ own documentation, user-action fetchers may not honour it: OpenAI says
robots.txt rules “may not apply” to ChatGPT-User, and Perplexity says Perplexity-User “generally
ignores” them, because a person triggered the fetch. Anthropic states all three of its agents
respect robots.txt.
Q: Does Google-Extended affect AI Overviews?
A: Google describes Google-Extended as a control for AI training and grounding in some of its other
systems. Eligibility for AI Overviews and AI Mode follows ordinary Search indexing and snippet
rules, which are governed by Googlebot and the snippet controls.
Q: Is robots.txt enough to stop AI training on my content?
A: It asks crawlers that read it to stay out, and the documented training crawlers say they honour
it. It is not a technical barrier and has no effect on content already collected.

### llms.txt: what the spec actually says, and whether your site needs one (https://trysearchlift.com/blog/llms-txt-explained)
Published September 11, 2026. 7 min read. Technical.

The file format from the proposal itself, Google’s statement that no AI text file is required
for AI Overviews, and the narrow case where writing one is still worth twenty minutes.

llms.txt is a proposed convention for a plain-text file at the root of a website that tells a
language model, in a few hundred words, what the site is and where its important pages are. It
has been enthusiastically over-sold. This article sticks to what the proposal says, what Google
says, and the one situation where it is clearly worth doing.

#### What the proposal says

The format was proposed by Jeremy Howard in September 2024 and is maintained at llmstxt.org. Its
stated problem is that “web pages are built for people”: an HTML page wraps its information in
navigation, ads and scripts, and turning it back into clean text is difficult and imprecise. The
file is meant to be a curated, model-friendly overview, sitting alongside robots.txt and the
sitemap rather than replacing either.

The format is Markdown with a fixed shape. In order:

1. An H1 with the name of the site or project. This is the only required section.
2. A blockquote with a short summary containing the key information needed to understand the rest.
3. Zero or more Markdown sections with more detail, in any form except headings.
4. Zero or more sections delimited by H2 headings, each containing a list of links in the form
[name](url), optionally followed by a colon and a note about that file.

```
# Acme Coffee Roasters

> Specialty coffee roaster in Leeds. Subscriptions for offices,
> single-origin beans, wholesale supply for cafés. Roasted weekly.

Roast dates are printed on every bag. Office subscriptions can be
paused from the account page at any time.

## Products

- [Office subscription](https://www.example.com/office): 250g to 5kg weekly, pause any time
- [Single-origin beans](https://www.example.com/beans): current lots with roast dates

## Company

- [About](https://www.example.com/about): founded 2016, roastery address, opening hours
```

Figure: An example llms.txt for a coffee roaster with four annotated parts: an H1 with the site name, the only required line; a blockquote summary; free text in any form except headings; and H2 sections listing links in the form name, URL and note. (https://trysearchlift.com/images/blog/llms-txt-anatomy.svg)
The four parts, in order. Only the first line is required; everything after it is convention.

#### What Google says

Google used to answer this by implication. As of its AI optimization guide, updated 10 July
2026, it answers by name. The guide says you do not need “new machine readable files, AI text
files, markup, or Markdown to appear in Google Search (including its generative AI
capabilities), as Google Search itself doesn’t use them”, and then goes further:

> It’s completely fine if you decide to create and maintain LLMS.txt files… Doing so will neither harm nor help your site’s visibility or rankings in Google Search, as Google Search ignores them.

Its summary of tactics to skip lists “creating unnecessary AI text files (like llms.txt)”
outright.

So the Google question is settled, in Google’s own words, with the file named. Anyone selling
llms.txt as a Google ranking factor is selling something Google has explicitly said it ignores.

Table: File | Job | Format | Who documents reading it | Google’s position
- robots.txt | Tells crawlers what they may fetch | Plain text, Robots Exclusion Protocol | Every major crawler, in its own documentation | Honoured; the reference documentation is Google’s own
- sitemap.xml | Lists the URLs you want indexed, with change dates | XML, sitemaps.org protocol | Search engines, including Google and Bing | Supported and recommended
- llms.txt | A curated summary of the site for language models | Markdown, llmstxt.org proposal | No assistant vendor documents reading a third-party one | Ignored; neither helps nor harms rankings
(Three files at the root, three different standings. Only the third is a proposal rather than a protocol.)

The second question, whether any other assistant reads it, has a less satisfying answer: there
is no public evidence that it does. OpenAI, Anthropic and Perplexity all publish an llms.txt on
their own documentation sites, which is often cited as proof they consume it. Publishing is not
consuming. None of them documents reading a third-party llms.txt at crawl or inference time, and
we have not found a vendor statement that they do. If that changes we will update this paragraph
and say when.

Table: Vendor | Publishes an llms.txt for its own documentation | Documents reading yours
- OpenAI | Yes, at developers.openai.com/llms.txt | No
- Anthropic | Yes, at platform.claude.com/docs/llms.txt | No
- Perplexity | Yes, at docs.perplexity.ai/llms.txt | No
- Google | No | No; states that Google Search ignores the file
(Checked 12 September 2026. Publishing one is not the same as consuming one.)

#### Where it is still worth twenty minutes

The proposal’s own examples are documentation sites, and that is where the file earns its keep:
an agent asked to use a product or an API can fetch one file instead of crawling forty pages.
For a business website the case is narrower but real. When a user-action fetcher such as
ChatGPT-User or Perplexity-User visits your site because a person asked about you, a clean text
summary of what you sell, where you are and what your prices are is the easiest thing for it to
read. The file costs nothing, cannot hurt you, and states your own facts in your own words,
which is also the cheapest defence against an assistant repeating a stale address it found
elsewhere.

Put differently: it will not get you into an answer. It can make the answer about you more
accurate once you are in one.

Figure: Two lanes. Getting into an answer: indexed, snippet-eligible, reachable by that crawler, named or cited; llms.txt has no documented role here and Google says it ignores the file. Being described accurately once you are in one: someone asks about you, a user-action fetcher visits, and it repeats what your pages state; here llms.txt is a clean statement of your facts, if the fetcher looks for it. (https://trysearchlift.com/images/blog/llms-txt-where-it-helps.svg)
Two different jobs. The file has no documented role in the first and a modest, real one in the
second.

#### Writing one that is honest

- Keep the blockquote to the facts a buyer needs: what you sell, where you are, who it is for.
- Link only to pages that exist and say what the note claims. A dead link in a file meant to help
machines read you is worse than no file.
- State prices, dates and locations plainly where you publish them elsewhere; leave out anything
you would not put on the page.
- Regenerate it when the products or the address change. A stale llms.txt is a stale fact
delivered with confidence.
- Mention it with a <link rel="alternate" type="text/plain"> in the page head so a fetcher that
looks for it finds it.

SearchLift checks a site’s live /llms.txt against the proposal’s format and generates one from
the site’s real pages, products and stated business information; it is one of the lower-drama
items on the audit, and that is the right amount of drama for it.

Common questions:
Q: Does llms.txt improve Google rankings?
A: No. Google’s documentation says no AI text files or special markup are needed for AI Overviews
or AI Mode, and the file has no stated role in ranking.
Q: Do ChatGPT or Perplexity read llms.txt?
A: Neither company documents llms.txt as a requirement. User-action fetchers may retrieve it if it
is present and linked, which is the practical reason to have one that states your facts
correctly.
Q: What is llms-full.txt?
A: A convention some sites use alongside llms.txt for a longer, complete plain-text version of
their content. It is not part of the core proposal’s required format.
Q: Is llms.txt a replacement for robots.txt or a sitemap?
A: No. The proposal positions it as a complement: robots.txt governs access, the sitemap lists
URLs, and llms.txt is a curated summary for models.

### The Search generative AI control: the new setting that decides if Google can cite you (https://trysearchlift.com/blog/search-generative-ai-control-search-console)
Published September 11, 2026. 9 min read. Technical.

Google finished rolling this out to every site on 31 August 2026. It is the first real
eligibility gate for AI Overviews and AI Mode, it lives in Search Console settings, and most
site owners have not looked at it.

For two years the honest answer to “how do I control whether Google’s AI features use my site?”
was that you could not, not separately. Eligibility followed ordinary Search: if you were
indexed and snippet-eligible, you were in. That changed on 31 August 2026, when Google finished
rolling out the Search generative AI control to every site worldwide. It is a setting in Search
Console, it is the first switch that speaks directly to AI Overviews and AI Mode, and because it
was rolled out silently to properties nobody had reason to re-open, most site owners have never
seen it.

The short version. The default is include, so nothing broke while you were not looking. The
control decides whether your pages can appear in, and ground, Google’s generative AI features.
It is not a ranking signal, it does not affect the rest of Search, and it is a different lever
from Google-Extended, which governs training rather than citation.

#### Where it lives

1. Open Google Search Console and select the property you want to check.
2. Go to Settings in the left-hand navigation.
3. Find Search generative AI in the settings list. If you have never touched it, it will show the
inherited default rather than an explicit choice.
4. Read which value is set and whether it is inherited or set on this property. Inheritance is the
part that catches people out, and it is covered below.
5. If you change it, expect one to two days before the change takes effect. Note the date you
changed it so you can line it up against your Generative AI performance report later.

Figure: Breadcrumb: Search Console, Property, Settings, Search generative AI. A panel shows two options, Include (default: pages can be shown and used) and Exclude (removed from the answer and from grounding), a tag reading inherited from parent, and a note that changes take effect in one to two days. (https://trysearchlift.com/images/blog/search-generative-ai-control-where-it-lives.svg)
A schematic of the path, not a screenshot. The value you see is inherited unless someone set it
on this property.

#### What excluding actually does

Excluding is broader than people assume. It does not merely remove the little link cards under
an AI Overview. It removes your content from grounding too, which is Google’s word for the step
where the model reads your page to write its answer. So the choice is not “cited or not cited”.
It is “used or not used”. A site that excludes itself is not quietly summarised without credit;
it is absent from the answer.

What it does not do is equally important. Google’s help documentation is explicit that the
control is not used as a ranking or inclusion signal affecting other parts of Search. Setting it
to exclude does not demote you in the ten blue links, and setting it to include does not promote
you. It is a visibility gate for one surface, not a lever on the rest.

#### It is not the same as Google-Extended

This is the distinction most coverage gets wrong, and it matters because the two controls point
at completely different things.

Table: Control | Governs | Effect on AI Overviews and AI Mode
- Search generative AI control (Search Console) | Whether Google Search may show and ground on your pages in its generative AI features | Direct. Excluding removes you from the answer entirely
- Google-Extended (robots.txt) | Whether crawled content may train Gemini models and ground Gemini Apps and Vertex AI | None. Google states it does not affect inclusion in Google Search and is not a ranking signal
- Googlebot (robots.txt) | Whether Google may crawl the page at all | Total. Blocked pages are not indexed, so nothing downstream applies
- nosnippet, max-snippet, data-nosnippet | Whether and how much text may be shown | Restricts the text available to an AI Overview, at the cost of your ordinary snippets too
(Four controls, four different jobs. Blocking Google-Extended to “stay out of AI Overviews” is the common mistake: it changes training, not citation.)

Figure: A pipeline from Crawl, gated by Googlebot in robots.txt, to Index, to Snippet, gated by nosnippet and max-snippet, branching to the ten blue links and to AI Overviews and AI Mode, which are gated by the Search generative AI control in Search Console with a default of include. A separate dashed branch from Crawl leads to Gemini apps and Vertex AI training and grounding, gated by Google-Extended, which does not affect Search. (https://trysearchlift.com/images/blog/google-ai-controls-pipeline.svg)
Four switches on one pipeline. Google-Extended sits on a branch that never reaches Search, which
is why blocking it changes nothing about AI Overviews.

#### Inheritance, which is where the surprises are

The setting inherits from the closest manually-configured parent property. If you have a domain
property and several URL-prefix properties beneath it, a value set on the domain flows down to
every child that has not been set explicitly. That is convenient until someone sets a value on a
subdomain years ago, forgets, and then wonders why the parent’s setting is not applying there.

Two practical consequences. First, check the property that actually matches the URLs you care
about, not just the domain property. Second, if you manage a site with staging, help-centre or
shop subdomains registered separately, check each of them: the one you never think about is the
one with the stale explicit value.

Figure: A domain property, example.com, set to include, with three URL-prefix children: www and shop inherit include; help.example.com was set explicitly to exclude in 2024 and keeps that value despite the parent. (https://trysearchlift.com/images/blog/search-console-setting-inheritance.svg)
The subdomain nobody has opened since 2024 keeps its explicit value while the domain property
says include. Check the property that matches the URLs you care about.

#### Should you exclude?

Almost certainly not, and the reasoning is short. Excluding does not stop Google competing with
you for the answer; it stops Google using your answer while it does. The visitor still gets an
AI Overview, assembled from whoever did not exclude themselves. You have removed your name from
a surface your competitors remain on.

There are narrow cases where it is defensible: content licensed under terms you cannot square
with generative reuse, or a legal position your counsel has taken. Those are genuine. “I do not
want AI to steal my traffic” is not one of them, because the traffic goes somewhere either way.

#### What to do this week

- Open Settings on every Search Console property you own and record the current value and whether
it is inherited or explicit.
- Reconcile it against your robots.txt. A site that allows Googlebot, blocks Google-Extended and
has the generative AI control set to include is a perfectly coherent position: cite me, do not
train on me.
- Note the date of any change, then compare your Generative AI performance report either side of
it. One to two days for the change, longer for the data to be meaningful.
- If you manage client sites, check this before the client discovers it themselves. An explicit
exclude nobody remembers setting is a real and awkward find.

SearchLift reads this control’s state alongside your Generative AI performance data and your
robots.txt rules in one place, so the three cannot quietly disagree with each other. It is part
of the Search Console suite, and it sits next to the AI visibility tracking that tells you
whether being eligible actually turned into being named. If the crawler side is what you are
reconciling, the AI crawlers and robots.txt article covers every token and what each one really
controls.

Common questions:
Q: What is the Search generative AI control?
A: A setting in Google Search Console that decides whether your site can appear in and ground
Google’s generative AI features, meaning AI Overviews and AI Mode. Google completed its rollout
to all sites worldwide on 31 August 2026. The default is to include the site.
Q: Does excluding my site from AI Overviews hurt my rankings?
A: No. Google states the control is not used as a ranking or inclusion signal affecting other parts
of Search. It governs the generative AI surface only.
Q: Is this the same as blocking Google-Extended?
A: No, and confusing the two is the most common error. Google-Extended governs whether crawled
content may train Gemini models and ground Gemini Apps and Vertex AI. Google states it does not
affect inclusion in Google Search, which is where AI Overviews and AI Mode live. The Search
Console control is the one that governs citation.
Q: How long does a change take to apply?
A: Google indicates one to two days for the setting itself. Seeing the effect in your Generative AI
performance report takes longer, because you need enough post-change data to compare against.
Q: Which property does the setting apply to?
A: It inherits from the closest manually-configured parent property. A value set explicitly on a
subdomain overrides the domain property, so check the property that matches the URLs you care
about rather than assuming the domain-level value applies.
Q: Does excluding stop Google summarising my content without a link?
A: Yes. Excluding removes the content from grounding as well as from the link cards, so the page is
not used to compose the answer at all. That is the argument against excluding: the answer still
gets written, just without you in it.

Sources:
- Google Search Central: AI optimization guide: https://developers.google.com/search/docs/fundamentals/ai-optimization-guide
- Google Search Console Help: Search generative AI control: https://support.google.com/webmasters/answer/16908024

### Do AI crawlers execute JavaScript? What GPTBot and ClaudeBot actually see (https://trysearchlift.com/blog/do-ai-crawlers-execute-javascript)
Published September 11, 2026. 8 min read. Technical.

If your content arrives via JavaScript, the assistant quoting your competitor may simply never
have seen yours. Here is how to test what each crawler receives, in ten minutes, without a tool.

There is a specific, quiet failure that costs sites their place in AI answers, and it has
nothing to do with content quality. The page reads perfectly in a browser. It scores well. It is
indexed by Google. And when an assistant is asked the question that page answers, it cites
somebody else, because what the assistant fetched was a nearly empty HTML shell with a <div
id="root"> in it.

Googlebot renders JavaScript. It has for years, and Google documents the process. The AI
crawlers are a different population with different budgets, and the safest working assumption is
that most of them do not render at all.

Figure: Left, a product page as a browser renders it: a title, a price of $1,249, an in-stock badge, specifications and reviews. Right, the raw HTML a non-rendering fetcher receives for the same URL: a head with a script tag and a body containing only an empty div with the id root. (https://trysearchlift.com/images/blog/rendered-page-vs-raw-html.svg)
The same URL, fetched two ways. Everything an assistant would want to quote is on the left, and
none of it is on the right.

The short version. Do not trust anyone’s table of which bot renders what, including ours. Vendor
behaviour changes without announcement. Test your own URL against the documented user agents,
diff raw HTML against rendered HTML, and make the decision on what you measure today.

#### Why this is not the same question as “is my site indexed?”

Google runs a two-pass process: crawl the HTML, then queue the page for rendering when resources
allow, then index what the rendered page contained. That second pass is expensive, and Google
built it because the open web broke without it.

An AI search crawler has a different job. It is assembling a corpus to retrieve from, at
enormous scale, with no obligation to reproduce a browser. Nothing in OpenAI’s, Anthropic’s or
Perplexity’s crawler documentation promises rendering. Apple is the interesting exception: its
documentation states that Applebot may render page content within a browser, which is one reason
Applebot behaves more like Googlebot than like the others.

So “Google indexes my React app fine” is true and irrelevant. It tells you Googlebot rendered
it. It tells you nothing about what ClaudeBot received.

Figure: Googlebot: fetch HTML, then a render queue that runs JavaScript when resources allow, then index the rendered page. AI search crawlers as documented by OpenAI, Anthropic and Perplexity: fetch HTML, no documented render step, index what the response contained. (https://trysearchlift.com/images/blog/googlebot-vs-ai-crawler-pipeline.svg)
Google documents a second pass. Nothing in the OpenAI, Anthropic or Perplexity crawler
documentation promises one; Applebot is the documented exception.

#### Test it yourself in ten minutes

This is the whole method. It needs curl and a browser, and it produces a number you can act on.

1. Pick one URL that matters commercially: a product page, a pricing page, or the article you most
want cited.
2. Fetch it with a documented AI crawler user agent and save the response: curl -A "GPTBot/1.2
(+https://openai.com/gptbot)" -s https://example.com/page > raw.html. Repeat for ClaudeBot and
PerplexityBot using the exact strings in each vendor’s docs.
3. Strip the tags and count the words: sed -e "s/<[^>]*>//g" raw.html | tr -s "[:space:]" "\n" |
grep -c .. This is roughly what a non-rendering fetcher has to work with.
4. Open the same URL in a browser, view the fully rendered DOM, and count the words the reader
actually sees.
5. Compare the two numbers. If raw is within about 80% of rendered, you are fine. If raw is a tenth
of rendered, your content does not exist for anything that does not run JavaScript.
6. Repeat for one page per template (a product page, a listing page, an article), because
frameworks often render one route type on the server and another on the client.

Two honest caveats. Setting a user-agent string does not make you that crawler; it tells you
what your server and CDN return to that string, which is exactly the variable you control. And a
vendor that does not render today may render tomorrow. Date your findings.

#### What the numbers usually mean

Table: Raw vs rendered word count | What it means | What to do
- Raw ≈ rendered | Server-rendered or static. Every crawler sees your content | Nothing. Move on to the next problem
- Raw ~50–80% of rendered | Core content is server-rendered; tabs, reviews or specs hydrate client-side | Server-render the parts that carry the facts you want quoted
- Raw <20% of rendered | Client-rendered shell. Non-rendering crawlers get navigation and nothing else | This is the finding. Prerender, server-render, or accept that you are invisible to most AI crawlers
- Raw is an error page | Your CDN or WAF is blocking the agent before rendering is even a question | Different problem entirely, and a more common one than people expect
(The fourth row catches more sites than the third. A bot-management rule you did not write can return 403 to an agent you meant to welcome.)

Figure: A scale from 0% to 100% of raw words as a share of rendered words. Under 20%: a client-rendered shell, where non-rendering crawlers get navigation and nothing else. 50 to 80%: the core is server-rendered but tabs, reviews or specs hydrate client-side. About 100%: fine. A separate case: the raw fetch is an error page, meaning the CDN or WAF is blocking the agent. (https://trysearchlift.com/images/blog/raw-to-rendered-word-count-scale.svg)
The table as a scale. The dashed case at the bottom is the one that catches more sites than the
red band.

#### What to do if the answer is bad

- Server-render the commercially important routes first. You rarely need to convert the whole
application. The pages that answer buyer questions are the ones that need to exist in raw HTML.
- Put the facts in the first response, not in a lazy-loaded panel. Prices, specifications, hours
and availability are exactly what an assistant is asked for, and exactly what tends to hydrate
late.
- Do not solve it with a JSON-LD block alone. Google’s guidance is that structured data should
match the visible page. Markup describing content a non-rendering fetcher cannot see is a
mismatch waiting to be caught.
- Re-test after you ship. The word count is the metric. If raw did not move, the fix did not work,
whatever the framework docs promised.

This is the single most common finding on a first SearchLift audit, and it is why the audit
fetches every page the way a non-rendering crawler would rather than the way a browser does. If
the fetch is failing before rendering is even reached, the AI crawlers and robots.txt article
covers the access layer, and checking what the assistants currently say about you tells you
whether the invisibility is already costing you answers.

Figure: The SearchLift SEO audit screen: a banner reading Some AI engines cannot crawl your site with an Allow AI crawlers button, an SEO score of 61 out of 100, and a list of pages and products each with a score, the issues found on it and a Fix with Sam button. (https://trysearchlift.com/product/audit.webp)
Every page is fetched the way a non-rendering crawler fetches it, and scored on what came back.
Demonstration workspace.

Common questions:
Q: Do AI crawlers execute JavaScript?
A: Mostly no. Googlebot renders JavaScript as a documented second pass, and Apple states Applebot
may render pages in a browser. OpenAI, Anthropic and Perplexity do not document rendering for
their crawlers, so the safe assumption is that a client-rendered page reaches them as an empty
shell. Test your own URLs rather than relying on any published table, including this one.
Q: My React site is indexed by Google. Am I fine?
A: Not necessarily. Google indexing your site proves Googlebot rendered it. It says nothing about
GPTBot, ClaudeBot or PerplexityBot, which are separate crawlers with separate budgets and no
documented rendering step.
Q: How do I check what an AI crawler sees on my page?
A: Fetch the URL with the crawler’s documented user-agent string using curl, strip the HTML tags,
and count the remaining words. Compare that against the word count of the rendered page in a
browser. A large gap means your content depends on JavaScript that the crawler probably never
runs.
Q: Does server-side rendering help AI search visibility?
A: It helps in the narrow, mechanical sense that your content exists in the first HTTP response
instead of only after JavaScript runs. It is not a ranking factor. It is the difference between
the content being available to be cited and not being available at all.
Q: Will adding schema markup fix a client-rendered page?
A: No, and it can make things worse. Google asks that structured data match the visible content of
the page. Markup asserting facts a non-rendering fetcher cannot see is a mismatch, not a
workaround.

Sources:
- OpenAI: bots and crawlers: https://developers.openai.com/api/docs/bots
- Google Search Central: JavaScript SEO basics: https://developers.google.com/search/docs/crawling-indexing/javascript/javascript-seo-basics
- Apple: About Applebot: https://support.apple.com/en-us/119829

### Does schema markup help you get cited in AI search? What Google actually says (https://trysearchlift.com/blog/does-schema-markup-help-ai-search)
Published September 11, 2026. 7 min read. AI visibility.

The confident answer everywhere is yes. Google’s own documentation says there is no special
markup you need. Both can be true, and the difference is where the money is.

Search for whether schema markup helps you get cited by AI and you will find a great deal of
confidence and very little sourcing. The honest position is more interesting than either the yes
or the no, and Google has written it down.

The short version. There is no schema.org type that makes an assistant cite you, and Google says
so in plain words. Structured data still earns its place, for a reason that has nothing to do
with AI: it disambiguates what your page is about, and it is the cheapest way to state facts
unambiguously. Add it because it is correct, not because it is a lever.

#### What Google actually wrote

From Google’s AI optimization guide, in a section listing tactics to stop worrying about:

> Overfocusing on structured data: Structured data isn’t required for generative AI search, and there’s no special schema.org markup you need to add. However, it’s a good idea to continue using it as part of your overall SEO strategy.

And, separately: “You don’t need to create new machine readable files, AI text files, markup, or
Markdown to appear in Google Search (including its generative AI capabilities), as Google Search
itself doesn’t use them.”

Read those together and the claim being denied is specific. There is no AIOptimized type, no
markup that buys citation. The claim not being denied is that structured data is worthless;
Google explicitly says keep using it.

#### The rule that matters more than the type

The instruction worth internalising is the one about agreement: make sure your structured data
matches the visible content of the page. This is not a formality. A page whose markup claims a
price, a rating or an availability the reader cannot see is making two contradictory statements
about the same fact, and a system trying to decide whether to repeat your claim now has a reason
not to.

That is the practical mechanism by which schema affects AI citation, and it is defensive rather
than offensive. Correct, corroborated markup does not lift you. Incorrect markup gives a machine
a reason to distrust the page.

Figure: Left, a pricing page showing $49 a month while its JSON-LD says price 39: two statements about one fact. Right, the same page with JSON-LD saying 49: one fact, stated twice. (https://trysearchlift.com/images/blog/schema-markup-and-visible-text-agreement.svg)
The failure mode is not missing markup. It is markup that says something the page does not.

Table: Claim you will read online | Status | What is actually true
- Adding FAQPage schema gets you into AI Overviews | False | Google states no special schema is required. FAQ markup has a narrow, separate rich-result role
- Schema is how LLMs understand your page | Mostly false | Assistants read the page. Markup is one corroborating signal, not the channel
- Structured data is pointless now | False | Google explicitly advises continuing to use it as part of ordinary SEO
- Mismatched markup is harmless | False | Google asks that structured data match visible content. Mismatch is a quality problem, and for some types a manual-action risk
- More schema types is better | False | Types that do not describe the page add noise and more surface area to contradict yourself on
(The pattern: schema is a correctness instrument, not a visibility lever. Most of the confident advice inverts that.)

Figure: Your page, with prose as the thick channel to the answer, labelled what gets quoted, and JSON-LD as a thin dashed corroborating signal. A crossed-out chip reads AIOptimized schema: no such type exists in schema.org. (https://trysearchlift.com/images/blog/what-an-assistant-reads-prose-vs-markup.svg)
The prose is what gets quoted. Structured data corroborates who and what the page is. There is
no type that does the first job.

#### What to do instead

- Mark up what the page genuinely is: Organization, Product, Article, BreadcrumbList. Then stop.
Accuracy beats coverage.
- Audit for disagreement. Every price, rating, date and availability in your JSON-LD should be
visible on the page. This is the check almost nobody runs and the one with a real downside.
- Put the facts in the prose. An assistant quoting you is quoting sentences. A price that exists
only in JSON-LD is a price it cannot comfortably repeat.
- Keep the entity consistent. Same organisation name, same URL, same identifiers across every
page. This is where structured data genuinely helps a machine resolve who you are.
- Ignore anything selling a schema type as an AI ranking factor. Google named this pattern
specifically as a thing to stop doing.

Table: Type | What it should state | What to check
- Organization | Who you are: name, logo, the profiles that are yours | The same name and URL on every page
- Product | Name, price, currency, availability | Every value visible on the page, and current
- Article or BlogPosting | Headline, author, published and modified dates | Dates that match what the page shows
- BreadcrumbList | Where the page sits in the site | Matches the visible breadcrumb
- FAQPage | Questions and answers that appear on the page | A narrow rich-result role in ordinary Search; never a question the page does not show
- LocalBusiness | Address, hours, phone | The same as your listings elsewhere
(Six types that describe what most commercial pages actually are. Accuracy beats coverage: a type that does not describe the page is surface area to contradict yourself on.)

SearchLift’s audit checks structured data against the rendered page rather than in isolation,
because a valid JSON-LD block that disagrees with the visible text passes every validator and
still fails the test that matters. If the page is client-rendered, read what AI crawlers
actually see first: markup describing content the fetcher never received is the worst version of
this problem. The same principle applies to llms.txt, which Google has also said it ignores.

Common questions:
Q: Does schema markup help you get cited in AI search?
A: Not directly. Google states that structured data is not required for generative AI search and
that there is no special schema.org markup to add. Structured data remains worth maintaining as
part of ordinary SEO, and accurate markup that agrees with the visible page avoids giving a
machine a reason to distrust your claims.
Q: Does FAQPage schema get me into AI Overviews?
A: No. Google says no special schema is needed to appear in its generative AI features. FAQ markup
has its own narrow rich-result eligibility in ordinary Search, which is a separate matter.
Q: Should I remove my structured data then?
A: No. Google advises continuing to use structured data as part of your overall SEO strategy. The
advice is to stop treating it as an AI visibility lever, not to delete it.
Q: What happens if my schema does not match my page?
A: Google asks that structured data match the visible content. A mismatch is a quality problem: for
some rich-result types it risks a manual action, and more generally it means your page states a
fact two different ways, which is a poor position from which to be quoted.
Q: Is there a schema type specifically for AI or LLMs?
A: No. No such type exists in schema.org, and Google has explicitly said there is no special markup
required for its generative AI features. Anything marketed as AI schema is not a recognised
standard.

Sources:
- Google Search Central: AI optimization guide: https://developers.google.com/search/docs/fundamentals/ai-optimization-guide
- Google Search Central: AI features and your website: https://developers.google.com/search/docs/appearance/ai-features

### Find keyword cannibalization in Search Console, without buying a tool (https://trysearchlift.com/blog/keyword-cannibalization-search-console)
Published September 11, 2026. 8 min read. Search Console.

Two of your pages are competing for the same query and Google is picking a different winner each
week. The data to prove it is already in Search Console. Here is the exact procedure.

Cannibalization is one of the few SEO problems where the diagnosis is genuinely free. You do not
need a crawler licence or a rank tracker. Search Console already records which of your URLs
Google showed for which query, and that is the entire dataset the expensive tools are inferring
from.

What the tools sell is the filtering. That part is a pivot table.

The short version. A query is cannibalized when two or more of your URLs both take meaningful
impressions on it and neither is clearly winning. Rotation is the tell: Google swapping which
page it shows week to week means it cannot decide, and an undecided Google ranks both worse than
one confident page.

Figure: Two grids of eight weeks for two URLs. Cannibalized: the URL that leads the query swaps between /guide-a and /guide-b almost every week. Healthy: /guide-a leads every week and /guide-b never does. (https://trysearchlift.com/images/blog/keyword-cannibalization-rotation.svg)
Schematic. The filled cell is the URL Google showed most that week. Rotation means Google cannot
decide, and an undecided Google ranks two competing pages worse than one confident one.

#### The procedure

1. Open Search Console, go to Performance → Search results, and set the date range to the last 3
months. Shorter windows are too noisy to judge rotation.
2. Open the Queries tab and sort by impressions. Pick a query with real volume that matters
commercially, not a brand term, since two of your pages ranking for your own brand name is
normal and harmless.
3. Click that query to filter the whole report by it, then switch to the Pages tab. You are now
looking at every URL Google showed for that one query.
4. Read the impression split. One URL with 95% of impressions is healthy. Two URLs each holding 20%
or more of impressions on the same query is the signal.
5. Add the Position column and compare. Two pages both sitting between roughly position 5 and 20 on
the same query is the classic pattern: neither is strong enough to hold the top slot alone.
6. Confirm rotation before acting. With the query filter still on, use Compare to set two adjacent
date ranges. If the leading URL swaps between them, Google is undecided, and that is
cannibalization rather than two pages that happen to be relevant.

#### Thresholds worth writing down

Most advice stops at “look for multiple pages ranking”, which flags half your site. These are
the filters that keep the list short enough to act on.

Table: Condition | Threshold | Why
- Query volume floor | 100+ impressions in 3 months | Below this, the split is noise and the fix is not worth the risk
- Secondary URL share | ≥20% of the query’s impressions | A 2% straggler is not competing, it is incidental
- Position band | Both URLs between 5 and 20 | Two pages at 40 are not cannibalizing, they are both simply losing
- Rotation | Leading URL changes between adjacent periods | The strongest single signal that Google cannot choose
- Intent match | Both pages answer the same question | A product page and a support article on one term is correct, not cannibalization
(Apply all five. The last one is the most commonly skipped and the most commonly fatal to a merge decision.)

Figure: A funnel of five filters narrowing every query in the report to a shortlist: query volume of 100 or more impressions in three months; a secondary URL taking 20% or more of the impressions; both URLs between position 5 and 20; the leading URL rotating between adjacent periods; both pages answering the same question. (https://trysearchlift.com/images/blog/keyword-cannibalization-five-filters.svg)
Apply them in order and the list shrinks to something you can act on. The last filter is the one
most often skipped.

#### What to do about it

Merging is the reflex and it is the most destructive option. Work through these in order.

- Do nothing. If the two pages genuinely serve different intents, leave them. Two URLs on one
query is not automatically a fault.
- Differentiate. Cheapest real fix. Retitle and re-angle the weaker page toward the intent it
should own, so the two stop overlapping.
- Consolidate internal links. Often the cause. Count your internal anchors: if half point at one
page and half at the other with the same anchor text, you built the confusion yourself. Point
them at the page you want to win.
- Merge and redirect. Only when the pages are substantially the same content. Fold the best of the
loser into the winner, 301 it, and update the internal links.
- Canonicalize. For near-duplicates you must keep for other reasons, such as a print or regional
variant.

Whatever you change, record the date. Cannibalization fixes take weeks to settle, and without a
change date the before-and-after is unreadable.

Figure: A decision flow. Different intents? Yes: do nothing. Substantially the same content? Yes: merge and 301. Must both exist? Yes: canonicalize. Otherwise differentiate the weaker page and point internal links at the winner. Record the date of any change. (https://trysearchlift.com/images/blog/keyword-cannibalization-what-to-do.svg)
Merge last. Most cannibalization is fixed by differentiating the weaker page and pointing
internal links at the winner.

#### When the free method runs out

This procedure is a query-at-a-time pivot, which is fine for a site with a few hundred pages and
impossible at a few thousand. The point at which it breaks is not sophistication, it is
arithmetic: you cannot click through ten thousand queries. That is when the Search Console API
or a BigQuery bulk export becomes worth the setup, and it is what the Search Console suite in
SearchLift automates: same data, same thresholds, run across every query instead of the ones you
thought to check.

Figure: The SearchLift Search Console screen: a connected property with 1,840 clicks, 41,200 impressions and a 4.5% average click-through rate, a clicks and impressions chart for the last 28 days, and a top queries table with clicks, impressions, CTR and position for each query. (https://trysearchlift.com/product/search.webp)
The same Search Console data inside SearchLift, where the query-to-page split runs across every
query instead of the ones you thought to click. Demonstration workspace.

Cannibalization also has a sibling worth knowing about: pages that are not competing with
anything and are simply fading. That is content decay, and the two are often confused because
both show as declining clicks.

Common questions:
Q: How do I find keyword cannibalization in Google Search Console?
A: Open Performance, set a 3-month range, pick a high-impression non-brand query from the Queries
tab, click it to filter, then open the Pages tab. If two or more URLs each hold 20% or more of
that query’s impressions and both sit between roughly position 5 and 20, you have a candidate.
Confirm it by comparing adjacent date ranges to see whether the leading URL rotates.
Q: Is two pages ranking for one keyword always bad?
A: No. If the pages serve genuinely different intents, such as a product page and a support
article, two URLs on one query is correct. It is only cannibalization when both answer the same
question and Google cannot decide which to show.
Q: Should I merge cannibalized pages?
A: Usually not first. Try differentiating the weaker page and fixing your internal anchor text
before merging. Merge and redirect only when the two pages are substantially the same content,
because a merge is hard to reverse.
Q: How long does a cannibalization fix take to show?
A: Weeks rather than days, since Google needs to recrawl both pages and re-evaluate. Record the
date you made the change so the before-and-after comparison is readable later.
Q: Do I need a paid tool to find cannibalization?
A: No. Search Console holds the underlying data and the method above is free. Paid tools automate
the filtering, which matters once you have more queries than you can reasonably click through.

Sources:
- Google Search Console Help: Performance report: https://support.google.com/webmasters/answer/7576553

### Content decay: finding the pages quietly dying in Search Console (https://trysearchlift.com/blog/content-decay-search-console)
Published September 11, 2026. 8 min read. Search Console.

Total traffic looks flat, so nothing looks wrong. Underneath, a third of your articles are
sliding and a few new ones are covering for them. Here is how to see it, and how to tell decay
from AI Overview absorption.

Content decay is hard to notice because the number you check every week is the one that hides
it. Sitewide clicks stay roughly flat while the composition underneath churns: older articles
slide, newer ones cover the gap, and by the time the total moves you have lost a year of
compounding.

The fix is not to publish more. It is to look at pages rather than totals, with a defined signal
instead of a vibe.

Figure: A stacked area chart over twelve months, schematic: older articles shrink from most of the total to about a third, steady pages hold, and new articles grow to cover the gap, so the sitewide clicks line on top stays flat. (https://trysearchlift.com/images/blog/content-decay-flat-total-moving-parts.svg)
Schematic. The total is flat by construction; the composition is not. Look at pages, not totals.

The short version. Compare each page’s trailing 28 days against the same 28 days a year earlier,
filter out anything too small to matter, and sort by absolute clicks lost. Then, and this is the
2026 part, check whether impressions fell too. If impressions held and clicks dropped, you did
not lose the ranking. Something took the click.

#### The procedure

1. Open Performance → Search results and switch to the Pages tab.
2. Use Compare and set the trailing 28 days against the same 28-day window one year earlier.
Year-on-year, not month-on-month: it cancels seasonality, which is what makes most decay
analysis wrong.
3. Export to a sheet. The UI will not sort by difference, and difference is the entire point.
4. Add a column for click change and one for percentage change. Filter out any page with fewer than
50 clicks in the earlier period; below that, percentages are meaningless.
5. Sort by absolute clicks lost, not percentage. A page down 30% from 2,000 clicks matters more
than one down 90% from 60.
6. For the top 20, add the impression change. This is the diagnostic column, and it splits the list
into two genuinely different problems.

#### Reading the impressions column

Clicks falling tells you something is wrong. Whether impressions fell with them tells you what.

Table: Clicks | Impressions | Diagnosis | Action
- Down | Down | Genuine ranking loss. You slipped, or someone overtook you | Refresh, or reassess whether you can still compete
- Down | Flat or up | You still rank. Something is absorbing the click: an AI Overview, a featured snippet, or a better competitor title | Check the SERP by hand. Often a snippet problem, not a content problem
- Flat | Up | Widening into weaker queries. Not decay | Leave it alone
- Down sharply | Down sharply | Possible deindexing or technical fault | Inspect the URL before touching the content
(Row two is the one that has grown since AI Overviews. Rewriting a page that still ranks fine is the most common wasted refresh.)

For row two, Google’s Search Console now reports generative AI performance separately, which
gives you somewhere to check rather than guess. Be honest about the limit of the inference: a
flat-impressions, falling-clicks page is consistent with AI absorption, and also with a
competitor writing a better title. Look at the SERP before concluding.

Figure: A decision path. Did clicks fall? No, and impressions rose: widening into weaker queries, leave it. Yes, and impressions fell too: genuine ranking loss, refresh or reassess, and inspect for a technical fault if both fell sharply. Yes, but impressions are flat or up: you still rank and something is taking the click, so check the SERP before rewriting. (https://trysearchlift.com/images/blog/content-decay-clicks-vs-impressions.svg)
The table as the three questions it answers. The last branch is the one that has grown since AI
Overviews, and the one that wastes the most rewrites.

#### Triage: refresh, consolidate, or retire

- Refresh when the page still targets a query with demand and the content is simply stale: dated
figures, superseded screenshots, a year in the title. This is the highest-return work in SEO and
the least glamorous.
- Consolidate when several thin pages circle one topic and none is winning. Merge into one strong
page and redirect. Check first that you are not looking at cannibalization, which is the same
symptom with a different cause.
- Retire when the query has genuinely gone away. Redirect to the nearest useful page. Not every
page deserves resuscitation, and pretending otherwise is how you end up refreshing content
nobody searches for.
- Leave alone when impressions are steady and the decline is a snippet problem. Fix the title and
description instead, and re-measure in 28 days.

Figure: Four cards. Refresh when the query still has demand and the content is stale. Consolidate when several thin pages circle one topic and none is winning. Retire when the query has gone away, redirecting to the nearest useful page. Leave alone when impressions are steady and the decline is a snippet problem: fix the title and description and re-measure in 28 days. (https://trysearchlift.com/images/blog/content-decay-triage.svg)
Four outcomes, one condition each. Refreshing content nobody searches for is the failure on the
retire side; rewriting a page that still ranks is the failure on the leave-alone side.

#### Why this beats a content calendar

A decayed page already has history, internal links and whatever authority it accumulated.
Bringing it back is cheaper than earning the same clicks from a new URL, and the feedback loop
is measured in weeks rather than quarters. The reason it rarely gets done is that nobody owns
it: refreshing is invisible work, and publishing is visible work.

SearchLift runs this comparison continuously and turns the output into a ranked refresh queue,
which is the same procedure without the spreadsheet. It sits in the Search Console suite, and
the pages it surfaces feed the article studio as rewrite briefs rather than blank topics. If the
decline turns out to be AI absorption rather than ranking loss, checking what the assistants say
about you is the next question, and the Search generative AI control is worth confirming while
you are in Search Console anyway.

Figure: The SearchLift analytics screen: an SEO and AI score of 61, up 3 since the last audit, counts of issues resolved, new issues, pages improved and pages regressed, and charts of the score and AI visibility over the last eight weeks. (https://trysearchlift.com/product/analytics.webp)
Pages that regressed since the last audit, listed next to the ones that improved, with the score
and AI visibility over eight weeks. Demonstration workspace.

Common questions:
Q: What is content decay?
A: The gradual loss of organic traffic to a page that once performed, usually from a mix of
staleness, competitors publishing better answers, and shifts in how results are presented. It is
hard to spot because sitewide totals stay flat while individual pages slide.
Q: How do I find decaying pages in Search Console?
A: In the Performance report, open the Pages tab and compare the trailing 28 days against the same
28 days a year earlier. Export, filter to pages with at least 50 clicks in the earlier period,
and sort by absolute clicks lost rather than percentage.
Q: Why compare year-on-year instead of month-on-month?
A: Year-on-year cancels seasonality. A month-on-month comparison will tell you that your ski
content declined in April, which is true and useless.
Q: My clicks fell but impressions did not. What does that mean?
A: You still rank; something is taking the click before the user reaches you. That is usually an AI
Overview or a featured snippet, but it can also be a competitor with a better title. Check the
live SERP and your Generative AI performance data before rewriting anything.
Q: Should I refresh every declining page?
A: No. Refresh pages whose query still has demand and whose content is merely stale. Consolidate
thin pages circling one topic, and retire pages whose query has genuinely disappeared.
Refreshing content nobody searches for is wasted effort.

Sources:
- Google Search Console Help: Performance report: https://support.google.com/webmasters/answer/7576553
- Google Search Central: AI optimization guide: https://developers.google.com/search/docs/fundamentals/ai-optimization-guide

## Company

About (https://trysearchlift.com/about): SearchLift AI is an AI SEO platform built by Dude Lemon
LLC, a US software company founded in 2017. It started as one SEO app and became a platform for
both halves of search: Google rankings and AI visibility.
Contact (https://trysearchlift.com/contact): the contact form, answered by the team that builds
the product. Support runs 24/7 on every plan, Free included.
Create a free account: https://searchlift.dudelemon.cloud/register
Short index: https://trysearchlift.com/llms.txt