← All articles
GEOAI searchSEOChatGPT search

GEO 2026: How to Get Cited in ChatGPT, Perplexity and Google AI

GEO, or SEO for AI search: what helps a small site get cited in ChatGPT, Perplexity, Google AI Overviews and Copilot, how to measure it, and what is myth.

Illustration of a web page sending small source cards into several floating AI answer bubbles

More people now ask ChatGPT, Perplexity, Copilot or Google’s AI Mode instead of scrolling through ten blue links. The answer they get names a handful of sources, and those sources are the ones that can still earn a click. That is what GEO, or Generative Engine Optimization, is about: making your pages easy for AI search tools to find, understand and cite.

There is a lot of noise around GEO right now, from “secret prompts” to paid tools that promise AI rankings. This article is the practical version for small brands, creators and service providers. I set up the basics on this site myself, so I will show what I did and why. One honest note first: alpercengiz.com is new, so I can’t show you citation numbers yet. Everything below comes from the platforms’ own documentation, as of October 2026.

What GEO actually is

The term comes from a 2023 research paper by Pranjal Aggarwal and colleagues, later accepted at the KDD 2024 conference. They tested how content changes affect visibility in AI-generated answers and reported improvements of up to 40%, with results that varied a lot by topic.

In day-to-day work, GEO is not a separate discipline from SEO. Every major AI search tool still has to fetch your page, decide it is relevant and trust it enough to quote it. Google says it plainly in its guide to AI features: there are no additional requirements to appear in AI Overviews or AI Mode, and no special optimisation is needed beyond being indexed and eligible to show a snippet. So think of GEO as SEO with a few new switches and a few new reports.

Step 1: Let the right crawlers in

Each AI company runs separate crawlers for search and for model training. This is the most useful thing to understand, because it means you don’t have to choose between “protect my content” and “be visible”.

AI search Crawler that decides visibility Training crawler (separate) Where you can measure it
ChatGPT search OAI-SearchBot GPTBot Referrals with utm_source=chatgpt.com
Perplexity PerplexityBot None (Perplexity says its bots don’t collect content for AI model training) Referrals from perplexity.ai
Google AI Overviews and AI Mode Googlebot Google-Extended Search Console generative AI reports
Microsoft Copilot and Bing Bingbot None listed Bing Webmaster Tools AI Performance
Claude Claude-SearchBot ClaudeBot Referrals in your analytics

The details from each source:

  • OpenAI says on its crawler page that sites which block OAI-SearchBot won’t appear in ChatGPT search answers (at most as plain navigational links), and that each setting is independent. You can allow OAI-SearchBot and block GPTBot. Changes for search take about 24 hours to apply.
  • Perplexity recommends allowing PerplexityBot to appear in its results. Its Perplexity-User agent fetches pages when a user asks and generally ignores robots.txt.
  • Anthropic describes three bots: ClaudeBot for training, Claude-SearchBot for search quality and Claude-User for user requests. Blocking the last two can reduce your visibility in Claude’s answers.
  • Google uses normal Googlebot for AI Overviews and AI Mode. Google-Extended limits training and grounding in some of Google’s other AI systems. Since 31 August 2026, every site in Search Console also has a Search generative AI control that lets you exclude it from AI Overviews, AI Mode and generative AI features in Discover. Google says the setting is not used as a ranking signal for the rest of Search.

Many sites block these bots without knowing it, through a security plugin, a CDN “block AI bots” toggle or an old robots.txt. Open yoursite.com/robots.txt and check.

On my own site I chose to allow everything, training crawlers included, because I want my work to be findable and a portfolio has nothing to hide. A shop with original product photography or a writer with paid content may decide differently, and that is a fair choice. The point is to decide on purpose.

A small business owner checking website settings on a laptop in a dimly lit home office

Step 2: Write pages an AI can quote

AI answers are built from passages. A page gets cited when it contains a clear, specific passage that answers the question better than other pages do. In practice:

  • Answer first. Put a direct, one- or two-sentence answer under each heading, then explain. A service page that says “I produce 15 to 60 second AI video ads for Reels, TikTok and Shorts” is easier to quote than one that says “we bring your vision to life”.
  • Be specific and factual. Names, dates, prices, steps, limits. Vague marketing text has nothing to cite.
  • Use structure. Bing’s AI Performance announcement recommends clear headings, tables and FAQ sections so key information is easier to reference accurately. It also recommends supporting claims with evidence and keeping text, images and video consistent.
  • Cite your own sources. Link to official documentation when you state a fact. It helps readers check you, and it signals that the content is grounded.
  • Keep it fresh. Bing explicitly recommends keeping content updated so AI systems reference the current version. If a price or feature changes, update the page and its date instead of writing a new one.
  • Show who is behind it. An about section, real contact details and consistent profiles help both people and machines connect the content to a real expert.

None of this is new. It is what good SEO and good writing already asked for. AI search simply rewards it more visibly.

Step 3: Help search engines notice changes

AI search tools mostly rely on search indexes, so getting indexed quickly matters.

  • Sitemap. Submit it in Google Search Console and Bing Webmaster Tools.
  • IndexNow. A free protocol to tell search engines when a URL is added, updated or deleted. According to the IndexNow FAQ, submissions are shared with participating engines, which the FAQ lists as Bing, Amazon, Naver, Seznam.cz, Yandex and Yep. Google is not among them. Bing names IndexNow as a way to help AI systems reference the current version of your content.
  • Don’t ignore Bing. Bing Webmaster Tools is where you see your citations in Microsoft Copilot, and OpenAI’s help page on ChatGPT search for Enterprise and Edu says ChatGPT may share search queries with Bing to return web results. Many small sites only ever set up Google Search Console. Adding Bing Webmaster Tools takes a few minutes, and you can import your site from Google.

Step 4: Structured data, with the right expectations

Structured data describes your page in a format machines read reliably: who wrote it, when, what it is about. It is worth adding, but Google says there is no special schema.org markup needed for its AI features, and the markup must match what is visible on the page.

Two things changed this year. Google stopped showing FAQ rich results in Search from 7 May 2026. And neither Google nor Bing presents schema as a shortcut into AI answers. Use it for accuracy, not as a trick. A FAQ section is still useful on the page itself, because it gives direct answers to real questions.

llms.txt: what it is and what it isn’t

llms.txt is a Markdown file at the root of a site that gives AI tools a short summary and a list of the most important links. Jeremy Howard proposed it in September 2024. It is a community proposal open for input, not a ratified standard, and Google’s AI features guide says you don’t need new machine-readable files or AI text files to appear in its AI features.

I publish one anyway at alpercengiz.com/llms.txt. It took twenty minutes, it gives any agent that reads it an accurate summary of my services, projects and articles, and it forced me to write down clearly what I do. That is the realistic case for llms.txt: cheap, harmless, possibly useful for some agents, and not a ranking factor anyone has confirmed.

How to measure AI visibility

This is where 2026 brought real progress. Three free sources, as of October 2026:

  1. Bing Webmaster Tools, AI Performance. Launched in public preview in February 2026, it shows total citations, cited pages and grounding queries (the searches the AI ran) across Copilot, AI summaries in Bing and select partners. In June, Bing added Intents, Topics, Citation Share and Compare. Citation Share shows your percentage of all citations for a grounding query. On 13 October 2026 Bing added Topics Portfolio in beta, which groups queries into topics and lists queries where other sites are cited but yours is not. Bing calls these numbers directional, and smaller sites may see few topics or none.
  2. Google Search Console, generative AI performance reports. Introduced on 3 June 2026 and rolled out to all sites on 31 August 2026, they show impressions and pages in AI Overviews, AI Mode and generative AI features in Discover, by country, device and date. They report impressions, not clicks; clicks from AI features are still counted in the normal Performance report.
  3. Your analytics. ChatGPT adds utm_source=chatgpt.com to its referral links, so you can filter those visits in Google Analytics or a similar tool. Look for perplexity.ai, copilot.microsoft.com and similar referrers too. Some AI traffic arrives without a referrer, so treat this as a minimum.

Watch trends and the pages that get cited, not single days. Bing itself warns that citation patterns shift with user behaviour, model changes and freshness signals.

A marketer reviewing simple line charts on a monitor in a dark office, with notes on the desk

Myths I would ignore

  • “You need a GEO tool to get into ChatGPT.” No tool can place you in an AI answer. Monitoring tools can be useful later, but the free reports above cover the basics.
  • “llms.txt gets you cited.” None of the official crawler and search documentation cited in this article says llms.txt is used for ranking or citation.
  • “Schema is the secret.” It helps machines read facts correctly. It does not replace content worth quoting.
  • “Blocking AI bots protects you with no cost.” Blocking training crawlers is a fair choice. Blocking the search crawlers removes you from those answers.
  • “GEO replaces SEO.” The AI tools above all lean on search indexes and crawlers. Weak SEO means weak GEO.

What I set up on alpercengiz.com

For transparency, here is the full list on this site: a robots.txt that explicitly allows the search and training crawlers of OpenAI, Anthropic, Perplexity, Google and others; an llms.txt; structured data for the person behind the site, the website, each article (BlogPosting), breadcrumbs and FAQs; a sitemap submitted to Google Search Console and Bing Webmaster Tools; IndexNow pings when I publish; and analytics that can show AI referrals. Every article, including this one, lists its official sources at the end.

I work the same way on content as I did on ad accounts. For 7+ years I’ve worked with Meta advertisers: first advising 150–280 businesses per quarter on campaigns, Pixel, Conversions API and catalog setups, then leading two teams of up to 27 Meta Marketing Pros serving ~4,500 advertiser accounts. The lesson that carries over: set up measurement first, then change one thing at a time. If you want to see how I apply that to ads, read what Meta’s AI labels and policies mean for AI video ads or my guide to Meta’s image-to-video ads.

The short version

Let the search crawlers in, decide on training crawlers on purpose, write pages with direct, specific answers, keep them current, submit your sitemap to Google and Bing, ping IndexNow, and measure with Bing AI Performance, Search Console and your referral data. That covers most of what a small site can control.

If you want help setting this up for your site, or content and AI video that gives AI search something worth citing, take a look at my services or get in touch.

FAQ

What is GEO (Generative Engine Optimization)?

GEO is the work of making your pages easy for AI search tools such as ChatGPT search, Perplexity, Google AI Overviews and AI Mode, and Microsoft Copilot to find, understand and cite as a source. The term comes from a 2023 academic research paper. In practice it builds on normal SEO: crawlable, indexed pages with clear, factual content.

Do I need an llms.txt file to appear in AI answers?

No. llms.txt is a community proposal from 2024, not a standard, and Google says you don’t need AI text files or special markup to appear in AI Overviews or AI Mode. It costs little to publish one, but do not expect it to change whether you are cited.

Should I block GPTBot and other AI crawlers in robots.txt?

Only if you want to keep your content out of model training. OpenAI, Anthropic and Google separate training from search: you can block GPTBot, ClaudeBot or Google-Extended and still allow OAI-SearchBot, Claude-SearchBot and Googlebot. If you block the search crawlers, you remove yourself from those AI search answers.

How can I see if AI search tools cite my website?

Use three free sources. Bing Webmaster Tools has an AI Performance report with citations from Copilot and Bing AI answers. Google Search Console has generative AI performance reports for AI Overviews and AI Mode. In your analytics, look for referral visits from chatgpt.com (ChatGPT adds utm_source=chatgpt.com) and perplexity.ai.

It helps search engines understand a page, as long as it matches the visible content, but Google says there is no special schema you need for its AI features. Google also stopped showing FAQ rich results in May 2026. Clear headings, tables and direct answers on the page matter more than any single markup type.