If you have any questions about our product, I’d be happy to chat!
Generative engine optimization (GEO) helps businesses get cited in answers from ChatGPT, Gemini, and Perplexity by making their content easy for AI systems to find, trust, and quote. Instead of competing only for a position in a list of links, GEO aims to get your brand named, quoted, or linked inside the AI-generated answer itself. It builds on SEO rather than replacing it, and Google states that optimizing for its own AI features is still SEO. In practice, GEO comes down to four things: letting the right AI crawlers reach your pages, publishing specific and well-sourced answers, earning genuine mentions on sites the engines already trust, and tracking whether you’re actually cited. This guide explains how each engine chooses sources, which robots.txt settings matter, which tactics have real evidence behind them, and how to measure results with a simple prompt scorecard.
Generative engine optimization is the process of improving how often, and how accurately, AI answer engines mention, quote, or link to your content. Traditional SEO aims for a position in a list of results. GEO aims for a place inside the synthesized answer itself. The term comes from the research paper “GEO: Generative Engine Optimization,” written by researchers from Princeton University and IIT Delhi and presented at the KDD 2024 conference. The researchers tested how different content changes affected visibility in AI-generated responses. Their best methods improved visibility by up to 40% on their benchmark and by up to 37% on Perplexity. Results varied by topic, so treat those figures as an upper bound from controlled tests, not a promise for every site.
GEO and SEO share the same foundation: crawlable, indexable, helpful pages from a credible source. The difference is the finish line.
| Traditional SEO | Generative engine optimization | |
|---|---|---|
| Goal | Rank in a list of results | Be cited or named inside an AI answer |
| Unit of success | Page position and clicks | Mentions, links, and accuracy of how you are described |
| Main surfaces | Google and Bing results pages | ChatGPT, Gemini, Perplexity, AI Overviews, AI Mode |
| Measurement | Rankings, impressions, clicks | Prompt tracking, AI impressions, AI referrals |
| What carries over | Technical health, content quality, authority, topical depth | All of it |
The stakes are real. In a Pew Research Center study of 900 US adults’ browsing in March 2025, Google users clicked a traditional result in 8% of visits when an AI summary appeared, compared with 15% when none did. Only 1% of visits included a click on a source inside the summary. When fewer people click, being named inside the answer matters more. For the closely related discipline of optimizing for direct answers, see our answer engine optimization guide.
All three engines rely on retrieval-augmented generation (RAG). In RAG, the system first retrieves relevant pages from a search index, then writes an answer grounded in those pages and links to the sources it used. Many AI search features also use query fan-out, which means the model runs several related searches at the same time to gather more information for a single question. A question about choosing a CRM, for example, might fan out into searches about pricing, integrations, and reviews. Each engine draws on a different index, so a page cited by one engine may not be cited by another.
ChatGPT’s search features rely on OpenAI’s own crawler, OAI-SearchBot. According to OpenAI’s crawler documentation, sites that opt out of OAI-SearchBot won’t be shown in ChatGPT search answers, though they can still appear as navigational links. OpenAI runs two other agents with different jobs:
What to do: allow OAI-SearchBot, decide separately whether to allow GPTBot, and make sure your firewall or CDN isn’t silently blocking OpenAI’s published IP ranges.
Most GEO advice treats Gemini as one thing. It isn’t. AI Overviews and AI Mode are features of Google Search. They rely on Google’s core ranking and quality systems, so standard Googlebot access and indexing apply. Google’s guide to optimizing for generative AI features, published May 15, 2026 and last updated July 10, 2026, states that optimizing for these features is still SEO. It lists two eligibility requirements:
The Gemini app is a separate product. According to Google’s common crawlers documentation, a robots.txt token called Google-Extended controls whether Google may use your content to train future Gemini models and to ground answers in Gemini Apps and in Vertex AI. Google states that Google-Extended doesn’t affect your inclusion or ranking in Google Search. Blocking it can therefore remove you from Gemini app grounding while leaving AI Overviews untouched.
What to do: keep Googlebot access and snippets enabled for key pages, confirm your Search Console AI features setting, and treat Google-Extended as a separate decision about Gemini app visibility and model training.
Perplexity searches the web and shows numbered sources beside every answer, which makes it the easiest engine to audit. According to Perplexity’s crawler documentation, PerplexityBot surfaces and links websites in Perplexity’s search results and isn’t used to collect content for training AI foundation models. Perplexity recommends allowing it in robots.txt if you want to appear in results. A second agent, Perplexity-User, visits pages when a user’s question needs them. Because a user requested the fetch, Perplexity says this agent generally ignores robots.txt rules. Perplexity also notes that robots.txt changes can take up to 24 hours to take effect.
What to do: allow PerplexityBot if you want to be cited, and use firewall rules rather than robots.txt if you need to restrict user-initiated fetches.
A single blanket rule that blocks “all AI bots” can quietly remove your site from AI answers. Use this matrix to separate search visibility from training use.
| Engine | Agent or token | Job | Effect of blocking it |
|---|---|---|---|
| ChatGPT | OAI-SearchBot | Surfaces sites in ChatGPT search | You won’t appear in ChatGPT search answers |
| ChatGPT | GPTBot | Collects content for model training | Opts you out of training; search visibility unaffected |
| ChatGPT | ChatGPT-User | Visits a page when a user asks | robots.txt rules may not apply; not used for search inclusion |
| Googlebot | Crawls for Search, including AI Overviews and AI Mode | You disappear from Google Search | |
| Gemini app | Google-Extended (token) | Gemini training and grounding in Gemini Apps and Vertex AI | Removed from that use; Google Search unaffected |
| Perplexity | PerplexityBot | Surfaces and links sites in Perplexity results | Much less likely to be cited by Perplexity |
| Perplexity | Perplexity-User | Visits a page on user request | Generally ignores robots.txt |
A common pattern for businesses that want AI visibility without contributing to OpenAI model training:
User-agent: OAI-SearchBot
Allow: /
User-agent: PerplexityBot
Allow: /
User-agent: GPTBot
Disallow: /Crawler names and policies change, so confirm current details in each provider’s documentation before editing your file. OpenAI and Perplexity both note that robots.txt changes can take about 24 hours to take effect in their search systems.
Much GEO advice repeats vendor statistics without sources. This table grades common tactics by the evidence behind them.
| Tactic | Evidence | Verdict |
|---|---|---|
| Add statistics, quotations, and cited sources | Among the strongest methods in the peer-reviewed GEO study | Strong: do it |
| Lead sections with clear, direct answers | Consistent with Google’s advice to organize content for readers with clear headings and sections | Strong: do it |
| Allow AI search crawlers | Platform documentation from OpenAI and Perplexity | Required |
| Earn genuine mentions on trusted third-party sites | Google says its AI features show what’s said about products across the web; industry studies are correlational | Moderate: invest steadily |
| Keep key pages current | Industry studies only; consistent with ranking fundamentals | Moderate |
| Keyword stuffing | Little to no improvement in the GEO study, sometimes worse than no changes | Avoid |
| llms.txt files | Google says Search ignores them; other services may use them | Optional; no effect on Google |
| Special “AI” schema or chunked rewrites | Google says neither is required for its AI features | Skip; keep standard structured data for rich results |
| Seeking inauthentic mentions on forums | Google says its systems focus on quality content and block spam | Avoid |
AI engines quote passages, not whole pages, so write each section to make sense on its own. Google also says there’s no need to break content into tiny pieces for AI, so write naturally for your readers.
The first priority depends on what you sell.
| Business type | What buyers ask AI | First priorities |
|---|---|---|
| Local service business | “Best emergency plumber near Ballard” | Complete Google Business Profile, consistent name, address, and phone everywhere, recent reviews, one page per service and area |
| B2B and SaaS | “Alternatives to X for mid-size teams” | Honest comparison pages, integration and use-case pages, crawlable documentation, original benchmark data, industry press mentions |
| Ecommerce | “Best standing desk under $500” | Complete Merchant Center product data, specification tables, clear shipping and return policies, genuine reviews |
| Professional services | “How do I choose an estate attorney?” | Named, credentialed authors, current regulatory detail, answers to real consultation questions |
Google’s guide notes that its AI responses can include product listings and local business information, and it points to Merchant Center and Google Business Profiles as ways to stay visible. For local businesses, strong local SEO is the foundation of local GEO.
No single dashboard tracks citations across every engine, so combine four sources:
AI answers vary from run to run, so a single check proves little. Use a scored, repeated sample instead.
Share of citation = your total points ÷ maximum possible points × 100 Track the score by engine. A rising share on Perplexity and a flat one on ChatGPT tells you exactly where to focus next month.
| Week | Focus | Actions |
|---|---|---|
| 1 | Access and baseline | Audit robots.txt and firewall rules for each crawler in the matrix, confirm your Search Console AI features setting, check that key pages render in plain HTML, build your prompt list, and record baseline scores |
| 2 | Priority pages | Rewrite your five most important pages to answer first, add dated facts and named sources, and remove filler and keyword repetition |
| 3 | Authority and coverage | Fill gaps in related sub-questions, update outdated figures, and correct inconsistent business details across profiles and directories |
| 4 | Re-test | Re-run the scorecard, compare by engine, and check Search Console and GA4 for changes |
Expect earlier movement on Perplexity, which retrieves live results, than on engines that refresh less often. Treat any change as a signal to keep testing, not as proof.
Many teams can run the first 30 days in-house. Outside help pays off when:
Seattle PPC’s AI search optimization service combines crawler audits, answer-first content, entity cleanup, and prompt tracking, reported against leads rather than vanity metrics. Book a strategy call to see where your brand stands in ChatGPT, Gemini, and Perplexity today.
Disclaimer: AI search platforms frequently update their algorithms, crawler policies, and citation practices. The recommendations and technical details in this guide are based on publicly available information and research available at the time of publication. GEO does not guarantee AI citations, brand mentions, traffic, or rankings, as results depend on factors outside a website owner’s control. Always verify current platform documentation before making technical changes to your website.
Allow OAI-SearchBot in robots.txt, make sure your pages load as readable HTML, and publish specific, well-sourced answers to the questions your buyers ask. Earning genuine mentions on trusted third-party sites also helps. Track progress with repeated prompt tests.
GEO builds on SEO. Both depend on crawlable, helpful, authoritative content. GEO adds a different goal, being cited inside AI answers, and different measurement, such as prompt tracking and AI impressions. For its own AI features, Google says this is still SEO.
Not in Google. Google says Search ignores llms.txt files, so they neither help nor hurt your Google visibility. Other AI services may use them, so adding one is optional. Focus on crawler access and content quality first.
That’s a policy choice. Blocking GPTBot opts your content out of OpenAI model training but doesn’t remove you from ChatGPT search, which is controlled by OAI-SearchBot.
Crawler access changes can take effect within about 24 hours for ChatGPT search and Perplexity. Content and authority improvements take longer to show up in answers. Plan on two to three monthly measurement cycles before judging progress.
You can’t buy a place in organic AI citations. Some platforms, including Google, sell labeled ads near or within AI answers, but those are ads, not citations. Organic visibility comes from access, content quality, and authority.
Yes. AI engines answer specific questions, and a small business with precise, local, first-hand answers can be cited for those queries even against larger brands. Narrow, specific prompts are the best place to start.
We’d love to hear from you. Fill out the form and let’s start building something amazing together.