AI Source Checklist: Get Picked by Perplexity, ChatGPT and Google AI
How each tool finds sources, side by side (with Claude and Bing as a bonus), then 22 checks, a robots.txt template and a monthly test log. Every claim tagged: confirmed, study, likely or unknown.
You get a side-by-side of what Perplexity, OpenAI and Google say publicly about sources, a checklist that covers all three, and a ready robots.txt block. Use it on your most important page today and measure it with the free reports each company gives you.
For: Founders and small teams who want one page to be cited by every major AI answer tool, not just one.
- How each tool finds sources (what the companies say) · 7
- If you only have 15 minutes · 4
- Bonus: Claude and Microsoft Copilot · 4
- Access checks (do these first) · 9
- Page checks (what cited pages tend to share) · 9
- Copy-paste robots.txt block · 1
- Off-site checks · 4
- The side-by-side test (monthly, 20 minutes) · 5
- Side-by-side log (template + worked example) · 2
- What the test tells you · 5
- Common mistakes · 6
- Short on time? Do the 15-minute quick start first.
- Read the comparison table, then run the access checks. If a crawler is blocked, nothing else matters.
- Run the page checks on your top 5 buyer pages, then the side-by-side test monthly with the log template.
How each tool finds sources (what the companies say)
Tags: CONFIRMED = the company says it in its own docs. LIKELY = widely reported or observed, not stated officially. UNKNOWN = no public answer.
| Perplexity | ChatGPT search | Google AI Overviews / AI Mode | |
|---|---|---|---|
| How it searches | Searches the web for each question (CONFIRMED, Perplexity docs) | Searches live pages when search is used; otherwise answers from training data (CONFIRMED). Uses third-party search providers plus partner content (CONFIRMED, launch post); Bing reported as one provider (LIKELY) | Uses Google's index and ranking systems; may 'fan out' into related searches (CONFIRMED) |
| How it shows sources | Numbered citations linking to each source | Inline citations and a Sources list | Supporting links inside and beside the answer |
| Crawler to allow | PerplexityBot (CONFIRMED) | OAI-SearchBot (CONFIRMED) | Googlebot (CONFIRMED) |
| Training-only token (blocking does not remove you from search) | Perplexity says PerplexityBot is not used to train foundation models | GPTBot | Google-Extended |
| User-triggered fetcher | Perplexity-User (generally ignores robots.txt) | ChatGPT-User (robots.txt may not apply; not used to decide search inclusion) | n/a |
| Free report for site owners | None official (UNKNOWN) | None official (UNKNOWN) | Search Console generative AI performance report: impressions only (CONFIRMED, June 2026; all sites by 31 Aug 2026) |
| How it picks which page to cite | UNKNOWN (not published) | UNKNOWN (not published) | UNKNOWN beyond 'normal search systems' |
If you only have 15 minutes
- Open yourdomain.com/robots.txt (3 min)
Search the page for OAI-SearchBot, PerplexityBot, Claude-SearchBot, Googlebot, Bingbot. Any 'Disallow: /' under them? Remove it, or paste the block from the robots.txt section.
- Check your CDN (3 min)
In Cloudflare or your host, look for an 'AI bots' or 'bot fight' setting. Make sure search crawlers aren't blocked.
- Ask one buyer question in all three tools (6 min)
Write down who gets cited. Note any URL cited by two or more.
- Check Search Console (3 min)
Open the generative AI performance report. Is your main page getting AI impressions? Write the number down.
Bonus: Claude and Microsoft Copilot
| Claude (Anthropic) | Bing / Microsoft AI answers | |
|---|---|---|
| Search crawler | Claude-SearchBot: indexes content for Claude's search results (CONFIRMED, Anthropic support) | Bingbot (CONFIRMED) |
| Training crawler | ClaudeBot: blocking it signals your future content should be excluded from training (CONFIRMED) | n/a |
| User-triggered fetcher | Claude-User: visits pages when a person asks Claude something (CONFIRMED) | n/a |
| Free report | None official | Bing Webmaster Tools AI Performance report (public preview, Feb 2026): citations, cited pages and sample grounding queries (CONFIRMED) |
Access checks (do these first)
- robots.txt allows Googlebot and Bingbot.
CONFIRMED for Google: AI features only use indexed pages. Bing is needed for Microsoft's AI answers, and ChatGPT search is reported to use Bing as one provider (LIKELY).
- robots.txt allows OAI-SearchBot.
CONFIRMED (OpenAI crawler docs): sites that opt out of OAI-SearchBot won't be shown in ChatGPT search answers, though they can still appear as navigational links.
- robots.txt allows PerplexityBot, and your firewall/CDN doesn't block its published IP ranges.
CONFIRMED (Perplexity crawler docs): allow PerplexityBot in robots.txt and its published IP ranges.
- Decide separately on GPTBot and Google-Extended.
These control training (and for Google-Extended, Gemini grounding). Google says Google-Extended doesn't affect inclusion in Google Search.
- Your CDN or bot-protection doesn't challenge AI crawlers by default.
Google lists 'crawling is allowed in robots.txt, and by any CDN or hosting infrastructure'. Check your Cloudflare or host bot settings.
- Key pages are indexed in Google Search Console and Bing Webmaster Tools.
Use URL Inspection. If it isn't indexed, it can't be a Google AI source.
- Important content is in the HTML text, not only in images, PDFs or behind a login.
Google: make sure important content is available in textual form.
- No nosnippet or max-snippet:0 on pages you want quoted.
Google uses these controls to limit what is shown, including in AI features.
- robots.txt allows Claude-SearchBot.
CONFIRMED (Anthropic support): blocking it may reduce your visibility in Claude's search results.
Page checks (what cited pages tend to share)
These are not official ranking factors for any of the three tools. Each is tagged. Use them as the patterns to check when you study which pages get cited.
- The page answers one buyer question, and the heading says that question.
LIKELY. A clear match between question and page makes the page easy to pick for that query.
- The first sentence under the heading is the direct answer.
LIKELY. Easy to lift. Add detail after it.
- Facts are specific: prices, numbers, dates, places, names.
STUDY. The GEO paper (Aggarwal et al., KDD 2024) found adding statistics raised visibility in AI answers in its benchmark, by up to 40% for the best methods. Lab result, not a promise.
- Claims link to their source.
STUDY. Citing sources was one of the strongest methods in the GEO paper. It also helps people trust you.
- Author or business name is visible, with an About page that says who you are and since when.
CONFIRMED as Google guidance ('Is it self-evident who authored your content?'). Effect on AI citation: UNKNOWN.
- Updated date is shown and true.
LIKELY. AI answers about prices and 'best X' are time-sensitive.
- Structured data matches what's visible on the page.
Google lists this in its AI features best practices. Schema doesn't replace visible text.
- Internal links point to the page from your menu or related pages.
Google: make content findable through internal links.
- Comparison and pricing questions have their own pages.
Buyers ask AI these constantly. A page that answers them directly is a candidate source.
Copy-paste robots.txt block
# Search and AI answer crawlers: allow User-agent: Googlebot Allow: / User-agent: Bingbot Allow: / User-agent: OAI-SearchBot Allow: / User-agent: PerplexityBot Allow: / User-agent: Claude-SearchBot Allow: / # Training crawlers: your choice. Delete these lines to allow training. User-agent: GPTBot Disallow: / User-agent: ClaudeBot Disallow: / User-agent: Google-Extended Disallow: / Sitemap: https://yourdomain.com/sitemap.xml
Off-site checks
- Google Business Profile is complete, verified and has recent reviews with replies.
Google says complete info and reviews help local results, and lists 'Business Profile information is up-to-date' in its AI features best practices.
- You appear on the pages that rank for 'best [category]' in your market.
When AI tools search, those list pages are often what they read. Ask for a fair listing or a correction.
- Your brand facts match everywhere: site, profiles, directories.
Conflicting facts give AI tools conflicting answers.
- Buyers discuss you on forums like Reddit, honestly.
Google and OpenAI both have Reddit data partnerships. See the Reddit playbook for the safe way.
The side-by-side test (monthly, 20 minutes)
- Pick 5 buyer questions
Questions a buyer asks before paying, e.g. 'best [category] in [city]', '[your product] vs [competitor]', 'how much does [service] cost'.
- Run each in all three tools
Perplexity, ChatGPT with search on, and Google (look at the AI Overview or open AI Mode). Use fresh, logged-out sessions.
- Screenshot and log the cited URLs
Sheet columns: date, tool, question, brands named, cited URLs, were you cited (Y/N).
- Find overlap
URLs cited by two or three tools are the most valuable. Study them: format, facts, author, date.
- Fix one page
Apply the page checks to the one page that should answer that question. Retest next month.
Side-by-side log (template + worked example)
Date | Tool (Perplexity / ChatGPT / Google AI) | Question | Brands named | Cited URLs | We were cited? (Y/N) | Cited by 2+ tools? | Page we should win with | Fix planned
Question: 'best accounting software for freelancers in India' Perplexity: cites 6 URLs, including 2 comparison blogs and 1 Reddit thread ChatGPT (search on): cites 4 URLs; 1 of the same comparison blogs Google AI Overview: links 3 pages; same comparison blog again Overlap: one comparison blog is cited by all three. That page is your first outreach target: ask for a fair listing or a correction. Our page: not cited. Fix: add a pricing table and a 'who it's for' line at the top.
What the test tells you
| Result | What it means | Next step |
|---|---|---|
| Cited by 0 tools | Access, indexing or clarity problem | Access checks first, then rewrite the page opening |
| Cited by Google only | Indexed and ranking, but other tools read different pages | Check OAI-SearchBot / PerplexityBot access and Bing indexing |
| Cited by ChatGPT/Perplexity, not Google | Page is clear but doesn't rank well enough on Google | Normal SEO: internal links, better match to the query |
| Competitor named, you aren't | Others vouch for them on list pages, reviews, forums | Off-site checks: get on the overlap pages |
| You're named with wrong facts | Old facts somewhere AI reads | Fix your site, then profiles and directories |
Common mistakes
- Blocking GPTBot or ClaudeBot and thinking you've left AI search.
CONFIRMED: those are training crawlers. Search uses OAI-SearchBot and Claude-SearchBot.
- Allowing crawlers in robots.txt but leaving Cloudflare 'block AI bots' on.
The firewall wins. Check your CDN's bot settings.
- Optimising for one tool only.
Our table shows each tool reads different pages. Overlap pages are worth most.
- Testing logged in, with history.
Answers personalise. Use fresh, logged-out sessions where you can.
- Paying for an 'AI ranking' guarantee.
No company publishes how it picks citations (UNKNOWN), so nobody can guarantee one.
- Reading Search Console's generative AI report as clicks.
CONFIRMED: impressions only. Clicks sit inside the normal Performance report.
Unlock the full checklist
Everything above, plus the branded PDF to keep.
Frequently asked questions
How does Perplexity choose its sources?
How does ChatGPT decide which websites to cite?
Does blocking GPTBot remove me from ChatGPT?
Can I see if AI tools cite my site?
Does blocking Google-Extended remove me from AI Overviews?
Sources
- Perplexity: How does Perplexity work?
- Perplexity: Perplexity crawlers
- OpenAI Help: ChatGPT search
- OpenAI: Overview of OpenAI crawlers
- Google Search Central: AI features and your website
- Google: Common crawlers (Google-Extended)
- Aggarwal et al., GEO: Generative Engine Optimization (arXiv 2311.09735)
- Google: Creating helpful, reliable, people-first content
- Google Business Profile Help: Tips to improve your local ranking
- Google Search Central Blog: Search generative AI performance reports in Search Console (June 2026)
- Bing Webmaster Blog: Introducing AI Performance in Bing Webmaster Tools (Feb 2026)
- Anthropic (Claude Help Center): Anthropic's crawlers and how to block them
- Search Engine Land: ChatGPT search officially launches (Bing as a provider)