AI Traffic: How to Get Organic Visits from ChatGPT, Perplexity & AI Search

AI search optimization, also called answer engine optimization (AEO) or generative engine optimization (GEO), is the work of getting your pages found, quoted and linked as sources in the answers that ChatGPT, Google’s AI Overviews and AI Mode, Perplexity, Microsoft Copilot and Claude give. In practice it means letting the AI search crawlers reach your site, ranking in the search indexes those assistants retrieve from, writing pages that answer plainly and cite their sources, earning mentions on the third-party sites the answers lean on, and measuring the citations and visits that come back.

Key Takeaways

• Google says optimizing for generative AI search such as AI Overviews and AI Mode is “still SEO”: a page must be indexed and eligible for a snippet, and llms.txt files, content “chunking” and special schema markup are on its list of things you can ignore, according to its Search Central guide updated in July 2026.
• AI assistants sent 0.32% of all website traffic from January to April 2026, and ChatGPT accounted for 74.78% of those AI referrals, according to SE Ranking’s June 2026 study of 101,574 websites.
• An AI Overview correlated with a 58% lower click-through rate for the top-ranking page in December 2025, according to an Ahrefs study of 300,000 keywords published in February 2026.
• Google searchers who saw an AI summary clicked a regular result in 8% of visits, against 15% when no summary appeared, in Pew Research Center’s analysis of 900 US adults’ browsing in March 2025.
• OpenAI and Google let a site opt out of model training separately from search, so blocking GPTBot or Google-Extended does not remove a site from ChatGPT search or AI Overviews, according to OpenAI’s and Google’s crawler documentation checked in October 2026.

How AI Search Picks the Pages It Cites

An AI answer with sources is built in five steps, and you can influence each one. A crawler fetches your page. A search index stores it. When someone asks a question, the assistant searches that index, often with several related queries at once, and retrieves the most relevant pages. The model then writes its answer grounded in those pages, and links the ones it relied on.

Google describes its own version in its guide to generative AI features on Google Search (last updated July 10, 2026): AI Overviews and AI Mode use retrieval-augmented generation, or grounding, which relies on Google’s core ranking systems to pull relevant pages from the Search index, and “query fan-out”, a set of related searches the model runs to gather more results. Its example: for “how to fix a lawn that’s full of weeds”, fan-out queries might include “best herbicides for lawns” and “how to prevent weeds in lawn”.

ChatGPT search, Perplexity and Claude run their own crawlers for the same job, which is why letting those crawlers in is the first step. Microsoft reports citations in Copilot and in Bing’s AI summaries through Bing Webmaster Tools, so the Bing index matters too.

AEO vs SEO: What Changes and What Doesn’t

Google’s answer to “is AEO different from SEO?” is short: from Google Search’s perspective, optimizing for generative AI search is optimizing for the search experience, “and thus still SEO”. What does change is where you appear, how you are picked and what you can measure.

	Classic SEO	AI search optimization (AEO, GEO)
Where you appear	A ranked list of links	A source link or mention inside a written answer
How a page is picked	Ranking for the query typed	Retrieved for the question and its related fan-out queries, then used to ground the answer
Which engines	Google, Bing	Google AI Overviews and AI Mode, ChatGPT search, Perplexity, Copilot, Claude
Off-site signal	Links to your pages	Links plus what reviews, forums and articles say about you
What you measure	Rankings, clicks, impressions	AI impressions and citations, AI referral visits, self-reported source

Founders who rank well tend to see the same pattern. Buildern, a construction management software company, credits about 95% of its $2M of revenue to SEO, and its founder does almost nothing aimed at chatbots alone:

We kind of doing not much. We actually optimizing for Google and whatever we do, same impact is happening also in LLMs.

Hmayak Tigranyan, CEO and founder of Buildern, in an April 2026 Founderpath interview (watch at 13:24)

If your classic SEO is weak, start with our SaaS SEO strategy guide. The 36 tactics below assume that foundation and add what AI search needs on top, in six steps.

Step 1: Let the AI Search Crawlers In

Every major assistant lets you control its search answers separately from model training, so check which robots.txt token you are blocking before you block it. The table shows the tokens each vendor documents, as of October 30, 2026.

Company	Search crawler (allow it to be cited)	User-triggered fetcher	Training control (blocking it keeps you in search)
OpenAI	OAI-SearchBot (ChatGPT search)	ChatGPT-User	GPTBot
Perplexity	PerplexityBot	Perplexity-User	None listed: Perplexity says PerplexityBot is not used to train foundation models
Anthropic	Claude-SearchBot	Claude-User	ClaudeBot
Google	Googlebot, plus the Search generative AI control in Search Console	Not covered here	Google-Extended, a robots.txt token rather than a separate crawler (Gemini training and grounding; no effect on Google Search inclusion or ranking)

1. Allow OAI-SearchBot. OpenAI’s crawler documentation says sites that opt out of OAI-SearchBot are not shown in ChatGPT search answers, though they can still appear as navigational links. Its settings are independent of GPTBot, so you can allow search and block training. After a robots.txt change, OpenAI says its systems take about 24 hours to adjust.
2. Allow PerplexityBot. Perplexity’s documentation says PerplexityBot surfaces and links websites in Perplexity’s search results and is not used to crawl content for AI foundation models. Perplexity-User only fetches pages when a person asks a question.
3. Allow Claude-SearchBot and Claude-User. Anthropic’s help center says disabling either may reduce your visibility in Claude’s search results. ClaudeBot is the one that collects potential training data.
4. Keep Googlebot crawling and your pages eligible for a snippet. Google’s guide says a page must be indexed and eligible to be shown with a snippet to appear in its generative AI features, so a nosnippet tag or a noindex takes a page out. Blocking Google-Extended does not: Google says it controls training and grounding for Gemini Apps and Vertex AI, and “does not impact a site’s inclusion in Google Search”.
5. Leave Search Console’s Search generative AI control on “include”. Google rolled the control out to all sites by August 31, 2026. It covers AI Overviews, AI Mode and generative AI features in Discover. Include is the default; exclude removes your links and stops your content grounding those answers, generally within a few days, and it doesn’t affect the rest of Search.
6. Check that your firewall or CDN isn’t blocking the bots your robots.txt allows. Bot protection often challenges unfamiliar user agents. OpenAI, Perplexity and Anthropic publish their crawlers’ IP addresses so you can allow them by IP, and Perplexity’s page includes example rules for Cloudflare and AWS WAF.

Step 2: Get Found in the Indexes AI Search Retrieves From

7. Rank in Google for the question. Google says its generative AI features “are rooted in our core Search ranking and quality systems”, so the pages you already rank are the ones it can retrieve. Ranking for the question your buyer asks is the foundation for appearing in the AI answer to it.
8. Get indexed in Bing and use IndexNow. Verify the site in Bing Webmaster Tools and submit your sitemap. Microsoft’s February 2026 announcement of its AI Performance report recommends IndexNow so that AI answers reference the current version of a page.
9. Serve the main content in the HTML. Google says it can process JavaScript as long as it isn’t blocked, but working with JavaScript frameworks is “generally more complex”. Patrick Stox of Ahrefs said on the Ahrefs Podcast that Gemini and Copilot render JavaScript and that the risk is concentrated in sites that hide content until scripts run. Text in the page source is safe for every crawler.
10. Answer the follow-up questions on the page that owns the topic. Fan-out means the assistant also searches the next questions a reader would ask, so cover them as sections and FAQs. Don’t build a separate page for every variation: Google says doing that mainly to manipulate AI responses violates its scaled content abuse policy.
11. Keep one page per intent. Google’s guide asks you to reduce duplicate content, which it says can be a bad user experience and wastes crawling on URLs you don’t care about. For template-driven pages, see our programmatic SEO guide before you scale.
12. Put facts in text, not only in images or PDFs. Tables belong in HTML, and every image needs descriptive alt text that states what it shows. A number that lives only inside a chart image is hard for any system to quote.

Step 3: Write Pages AI Answers Can Quote

13. Answer in the first sentence under each heading. Give the answer most good sources already agree on, in one or two plain sentences, then the detail. That is the passage an assistant can lift without rewriting it.
14. Say plainly what your product does, feature by feature. Patrick Stox described on the Ahrefs Podcast how AI tools kept saying Ahrefs had no social media monitor; Ahrefs created a page for the feature and still had to spread the word before assistants would show it. If a capability has no page that names it, assume assistants will say you don’t have it.
15. Use headings, lists and HTML tables. Google’s guide recommends paragraphs, sections and headings that give a clear structure, and Microsoft’s AI Performance announcement says clear headings, tables and FAQ sections make content easier for AI systems to reference accurately.
16. Cite your sources inline, with dates. Microsoft lists examples, data and cited sources as what builds trust when content is reused in AI answers, and Stox argued that citing sources matters more to these systems than Google has ever said it does for rankings.
17. Add an FAQ built from real questions. Take them from People Also Ask, your Search Console queries and sales calls, and answer each in two or three sentences. Use FAQ markup only if it matches the visible questions; Google says no special schema is required for AI features.
18. Skip the hacks Google says it ignores. Its guide lists what you can ignore for Google Search: llms.txt and other AI text files, “chunking” content into tiny pieces, rewriting content just for AI, and special schema.org markup. It also warns against third-party tools that claim access to internal Google metrics.

Step 4: Add Something AI Can’t Get Anywhere Else

Google’s guide puts this first: content that is “unique, compelling, and useful” will likely influence your presence in generative AI search more than anything else in it. It contrasts commodity content (“7 Tips for First-Time Homebuyers”) with first-hand, expert takes. An assistant that can already answer from common knowledge has no reason to cite you.

19. Publish a statistic only you have. One clear, dated number from your own product data, survey or customer base, in a self-contained sentence. Stox noted that Ahrefs’ own single-number studies, such as its AI Overviews click study, get picked up and repeated.
20. Write from first-hand experience. Google’s example of non-commodity content is a first-hand account (“Why We Waived the Inspection & Saved Money…”), not a summary of what others wrote. Show your own numbers, screenshots and decisions.
21. Build something interactive. A calculator, template or tool gives a reason to click through that an AI summary can’t replace. Our own free tools follow this idea.
22. Publish accurate comparison pages. Stox said comparison content is “highly cited in these systems” and that you almost have to publish it now, even where legal teams resist naming competitors. Keep every claim sourced; our comparison page guide covers the format.
23. Publish video that answers the same questions. Pew Research Center found that Wikipedia, YouTube and Reddit were the three most cited sources in Google’s AI summaries in March 2025, together 15% of the sources listed. Google’s guide also says its AI features can show relevant images and video.
24. Keep text, images and video consistent. Microsoft advises aligning them so they represent the same products and concepts, and refreshing pages regularly so AI systems reference the current version.

StandOut CV, a CV examples site built on search traffic, applied the interactive idea to its core pages:

…add features that you can’t get from ChatGPT. So here, you can click on the CV to enlarge it and have a look at it, and you can also switch to the text version so that you can copy and paste it. So we’re finding that if you’re putting things onto your site that people can’t get in ChatGPT now, that’s helping drive clicks through because, obviously, if they can just get it in ChatGPT, they’ve got no reason to leave.

Andrew Fennell, founder of StandOut CV, in a December 2025 Founderpath interview (watch at 18:50)

Step 5: Earn Mentions on the Sites AI Answers Cite

Assistants describe a brand the way the sources they retrieve describe it, so part of the work happens on pages you don’t own. Club Caddie, a golf course management software company, invests so that when someone asks ChatGPT for the best golf management software, it finds Club Caddie, its founder said. The starting point is reading the answers:

So, ultimately, you do know. It does give you the answers, right? So it tells you right there that Capterra is a huge source of it. It tells you then best country club management software. So you’ll go to these sites, you’ll look at where is AEO searching.

Jason Pearsall, founder and CEO of Club Caddie, in a December 2025 Founderpath interview (watch at 22:52)

25. List the sources each assistant cites for your category. Ask ChatGPT, Perplexity, Gemini and Copilot the questions your buyers ask (“best X software for Y”, “X vs Y”) and note every domain in the citations. Those review sites, directories and articles are your target list.
26. Get listed and reviewed where buyers compare tools. Claim your profiles on the review sites and directories that list showed, keep pricing and features current, and ask real customers for reviews.
27. Take part in the communities the answers quote. Pew’s data puts Reddit among the three most cited sources in Google’s AI summaries. Stox advised having real employees answer questions there rather than spamming. Our community-building guide covers running your own.
28. Pitch your data to journalists and newsletters. A study or dataset gives writers something to cite, which creates both links and the third-party mentions assistants retrieve.
29. Fix what people complain about. If reviews and forums repeat a complaint, assistants repeat it too. Stox called the top tactics business ones: build a better product and service, and address the issues customers complain about, because pages alone can’t outweigh what the rest of the web says.
30. Don’t buy inauthentic mentions. Google’s guide says seeking inauthentic mentions “isn’t as helpful as it might seem”, because its core ranking systems favor high-quality content and its spam systems, which the AI features also depend on, block the rest.

Step 6: Measure AI Visibility and Traffic

No single report shows the whole picture. Four sources together cover the path from being shown, to being cited, to a visit, to a customer.

31. Read Search Console’s Generative AI performance report. Rolled out to all sites by August 31, 2026, it shows how many times links to your site appeared in Google’s generative AI features, by page, country, date and device. It reports impressions, not clicks.
32. Read Bing Webmaster Tools’ AI Performance report. Launched in public preview in February 2026, it counts citations of your pages in Microsoft Copilot, Bing’s AI summaries and some partner integrations, and shows the grounding queries that retrieved them.
33. Track AI referrals in GA4. ChatGPT adds utm_source=chatgpt.com to the links it sends: OpenAI’s publishers FAQ said so (archived October 4, 2026, before OpenAI retired that page), and Founderpath’s own GA4 still recorded chatgpt.com as a UTM source in October 2026. For the other assistants, match referral sources such as perplexity.ai, gemini.google.com, copilot.microsoft.com and claude.ai, and group them in a custom channel group so AI traffic stops hiding inside Referral.
34. Ask every sign-up how they found you. A free-text field on the sign-up form and a lead source field in your CRM catch the buyers who asked an assistant, then typed your URL or searched your name, which analytics files as direct or branded search.
35. Run the same prompt set every month. Answers vary from run to run, so keep a fixed list of a few dozen buyer questions, run each in every assistant and log whether you are cited, which page and which competitors appear.
36. Judge the channel by sign-ups, not visits. AI referrals are still a small share of traffic, so compare sign-ups and pipeline from AI sources with other channels before you move budget.

The self-reported number can be the larger one. Allo, a business phone system, compared the two:

And then LLM, I think it’s very hard. I think the UTM ChatGPT is around 8% on the traffic, but when you talk, think it’s more around 20% traffic coming from LLM today.

Jeremy Goillot, founder and CEO of Allo, in an April 2026 Founderpath interview (watch at 6:10)

Club Caddie took the CRM route: Jason Pearsall said the company had added AI as a lead source and origin source in HubSpot, had only just started tracking it, and had counted a couple dozen leads from it that year.

How Much Traffic Does AI Search Send?

Still a small share, growing year over year, and taking clicks from classic search results. SE Ranking’s study of 101,574 websites with Google Analytics, published in June 2026, found AI platforms sent 0.32% of all website traffic from January to April 2026, up from 0.24% in 2025 and 0.02% in 2024. ChatGPT sent 74.78% of those AI referrals, followed by Gemini (11.56%), Perplexity (7.23%), Copilot (3.51%) and Claude (2.62%).

The bigger effect is on Google’s own results. Ahrefs compared the average desktop click-through rate of the top-ranking page across 300,000 keywords in December 2023 and December 2025, and found an AI Overview correlated with a 58% lower rate, up from the 34.5% it measured in April 2025. Pew Research Center found that Google users who saw an AI summary clicked a traditional result in 8% of visits, against 15% without one, and clicked a link inside the summary in 1% of visits; 18% of the searches it studied produced a summary.

Both studies are correlational and measure Google, not ChatGPT or Perplexity. For a SaaS founder they say the same thing: a page-one ranking now earns fewer clicks when an AI answer sits above it, and links inside the answer drew a click in only 1% of Pew’s visits, so count AI impressions and citations alongside clicks.

Do You Need an llms.txt File?

Not for Google. llms.txt is a proposal, first published in 2024 and now in a second version, for a markdown file at your site root that points AI agents to clean versions of your key pages. Google’s guide says Google Search doesn’t use llms.txt or other AI text files and that creating one will “neither harm nor help” your visibility there, while noting it is fine to maintain one for other services that use it.

We treat it the same way. Founderpath publishes an llms.txt file that is generated automatically, and our pages return a markdown version when an AI agent asks for one, but we spend our AI search effort on the steps above: crawler access, plain answers, sourced numbers and data only we have.

Funding Growth Without Giving Up Equity

Most AI search work is content, data, video and PR that you pay for before it pays back. Founders who grow this way can fund it from revenue rather than sell equity for it.

Founderpath provides non-dilutive capital for bootstrapped software founders. Connect your billing, banking and accounting data, and Founderpath shows you what you qualify for, with every figure disclosed up front. There’s no pitch deck, no equity and no board seats.

Check what you qualify for. It takes less than 5 minutes.

AI Search Optimization FAQ

What is answer engine optimization (AEO)?

Answer engine optimization is the work of getting your pages found, quoted and linked as sources in AI-generated answers from ChatGPT, Google’s AI Overviews and AI Mode, Perplexity, Copilot and Claude. It covers crawler access, ranking in the indexes those assistants retrieve from, writing plain answers with sources, earning third-party mentions and measuring AI citations and referrals.

Is AEO different from SEO?

Mostly not. Google’s Search Central guide says optimizing for generative AI search is optimizing for the search experience and thus still SEO, because AI Overviews and AI Mode retrieve pages through Google’s core ranking systems. What changes is where you appear (inside an answer), how you are picked (retrieval for related questions) and what you measure (AI impressions, citations and referrals).

What is AI search optimization called?

It goes by several names: answer engine optimization (AEO), generative engine optimization (GEO), AI SEO and LLM optimization (LLMO). Google uses AEO and GEO in its own guide and treats both as SEO for its generative AI features.

How do I get my website cited by ChatGPT?

Allow OAI-SearchBot in robots.txt and in your firewall, because OpenAI says sites that opt out of it are not shown in ChatGPT search answers. Then rank for the questions your buyers ask, answer them plainly with sources, publish data only you have, and get mentioned on the review sites and articles ChatGPT cites for your category.

Does blocking GPTBot remove my site from ChatGPT search?

No. OpenAI says GPTBot and OAI-SearchBot are controlled independently: GPTBot collects content that may be used to train its models, while OAI-SearchBot surfaces sites in ChatGPT search. You can block GPTBot and still allow OAI-SearchBot.

Does Google-Extended keep my site out of AI Overviews?

No. Google says Google-Extended controls training and grounding for Gemini Apps and Vertex AI and does not affect inclusion in Google Search. To keep a site out of AI Overviews, AI Mode and AI features in Discover, use the Search generative AI control in Search Console; nosnippet and noindex also remove pages.

Do I need an llms.txt file?

Not for Google: its guide says Google Search ignores llms.txt and other AI text files and that one will neither help nor harm your visibility. llms.txt is a proposal that some AI agents and documentation tools use, so publishing one is harmless, but it is not a substitute for crawlable pages that answer clearly.

How much traffic does AI search send to websites?

A small but growing share. SE Ranking’s study of 101,574 websites found AI platforms sent 0.32% of all website traffic from January to April 2026, up from 0.24% in 2025, and ChatGPT accounted for 74.78% of those referrals.

How do I track traffic from ChatGPT and Perplexity?

In GA4, ChatGPT visits arrive tagged utm_source=chatgpt.com, which OpenAI documented in a publishers FAQ it has since retired; other assistants show up as referral sources such as perplexity.ai, gemini.google.com and claude.ai, which you can group in a custom channel group. Add Search Console’s Generative AI performance report, Bing Webmaster Tools’ AI Performance report and a how-did-you-hear-about-us field on sign-up.

What is the best answer engine optimization tool?

Start with the free first-party reports: Search Console’s Generative AI performance report for Google’s AI features, Bing Webmaster Tools’ AI Performance report for Copilot citations and GA4 for AI referrals. Paid AI visibility trackers can help run prompts at scale, but Google warns that no third-party tool has access to its internal ranking or AI systems.

Sources and Method

Every figure, crawler rule and quote in this guide was checked against its source on October 30, 2026. Platform mechanics come from each vendor’s own documentation on that date: Google’s Search Central guide to generative AI features (updated July 10, 2026), its common crawlers page (updated October 15, 2026) and two Search Console help pages; OpenAI’s crawler page, its publishers FAQ (Wayback Machine copy of October 4, 2026; OpenAI has since retired the page) and our own GA4 data; Perplexity’s and Anthropic’s crawler pages; Microsoft’s Bing Webmaster Blog (February 10, 2026); and llmstxt.org. Traffic and click figures come from SE Ranking (June 2026), Ahrefs (February 2026) and Pew Research Center (July 2025), each linked where it is used. Founder quotes come from Founderpath interviews with Hmayak Tigranyan (Buildern, April 2026), Andrew Fennell (StandOut CV, December 2025), Jason Pearsall (Club Caddie, December 2025) and Jeremy Goillot (Allo, April 2026), each linked to the moment it was said, and Patrick Stox’s views from the Ahrefs Podcast (November 2025). An earlier version of this post focused on featured snippets and listed more than 40 company examples that were not about AI search; they were removed.