Generative Engine Optimization Best Practices: How to Get Cited by AI Search in 2026

New to this topic? Start with our step-by-step guide: How to Set Up Bing Webmaster Tools (and Why It Matters for AI Search).

Generative engine optimization (GEO) best practices come down to one idea: AI search engines cite pages they can find, trust and quote. Google, ChatGPT search and Perplexity all retrieve pages from a search index before they write an answer, so what earns a citation is solid SEO plus content that offers something other pages don’t: first-hand experience, original data and clearly sourced facts.

This guide is for site owners and marketers who want their pages cited in Google AI Overviews, AI Mode, ChatGPT, Perplexity and Copilot. It is built on Google’s own 2026 guide to generative AI features, the crawler documentation from OpenAI and Perplexity, published research, and what we measured on this site in October 2026.

Key takeaways

• AI answers are grounded in search results. A page has to be crawlable, indexed and allowed a snippet before it can be cited.
• Google treats GEO and AEO as the same work as SEO, and says you don’t need llms.txt, “chunked” pages or AI-specific rewrites.
• Each AI platform has its own crawler. Allow OAI-SearchBot for ChatGPT search and PerplexityBot for Perplexity, and check that your CDN or firewall isn’t quietly blocking them.
• Pages get quoted when they answer clearly and back claims with named sources, figures and first-hand experience.
• Measure citations directly: Search Console’s Generative AI report, Bing’s AI Performance report and a fixed set of prompts you re-check every month.

In this guide: What GEO is · How AI search chooses sources · 9 best practices · Myths to skip · GEO vs SEO · Our own data · 30-day plan · FAQ · Sources

What is generative engine optimization (GEO)?

Generative engine optimization is the practice of making your content easy for AI search systems to find, understand and cite when they write an answer. The term comes from a 2023 research paper by Pranjal Aggarwal and colleagues, GEO: Generative Engine Optimization, later published at the KDD 2024 conference. You’ll also see the same work called answer engine optimization (AEO) or LLM optimization.

Despite the acronym, GEO is not geotargeting. Local SEO helps you show up for nearby searches; GEO is about being one of the sources an AI answer links to. If local visibility is your goal, our guide to local SEO for small businesses covers it.

Google’s own position is direct. Its May 2026 guide, Optimizing your website for generative AI features on Google Search, treats AEO and GEO as names for the same work it calls SEO, because AI Overviews and AI Mode are rooted in its core ranking and quality systems. That is useful news: GEO doesn’t replace SEO. It sharpens it.

How do AI search engines choose which pages to cite?

Most AI search answers are produced the same way. The system reads the question, runs several related searches in the background (Google calls this query fan-out), retrieves candidate pages from a search index, and then writes an answer grounded in the pages it judged most useful, linking to them as sources. The technical name is retrieval-augmented generation.

Two things follow from this. First, a page that can’t be retrieved can’t be cited, however good it is. Second, the ranking step still rewards quality and relevance, so AI search inherits most of what makes a page rank. The difference is the last step: the answer quotes passages, so pages with clear, self-contained statements and verifiable facts are easier to use.

9 generative engine optimization best practices

These are ordered by dependency: there is no point polishing passages on a page an AI crawler can’t reach.

1. Make sure the page is indexed and allowed a snippet

Google’s AI features documentation sets the baseline: to appear as a supporting link in AI Overviews or AI Mode, a page must be indexed and eligible to show in Search with a snippet. Its newer guide adds that your site must be included in generative AI features in Search Console; that setting, under Settings > Search generative AI, is set to Include by default.

• Check the page in Search Console’s URL Inspection tool. A page reported as “Crawled, currently not indexed” can’t be cited by Google’s AI features.
• Don’t use nosnippet, max-snippet:0 or data-nosnippet on content you want quoted. Those controls exist to limit what appears.
• If a page isn’t indexed, start with our checklist for a website not showing up on Google.

We learned this one first-hand: before this rewrite, this page was itself listed as “Crawled, currently not indexed”. Thin, off-topic content was the cause, which is a large part of why it was rebuilt.

2. Let the right crawlers in, including at your CDN

Each AI platform fetches pages with its own crawler, and each robots.txt rule is independent. Blocking the wrong one removes you from an answer engine; blocking a training crawler does not.

robots.txt isn’t the only gate. CDN and firewall bot protection can challenge crawlers before they ever reach WordPress. On this site, our CDN’s “Checking your browser” screen was served to an SEO crawler on every one of 282 URLs in a single audit, and an earlier check showed GPTBot being rate-limited with HTTP 429 errors, while human visitors saw normal pages. Check your CDN’s bot settings and allow verified search crawlers. OpenAI and Perplexity both publish IP lists for this, and both say robots.txt changes take up to about 24 hours to be picked up.

3. Publish non-commodity content

“Non-commodity” is Google’s own word for content that isn’t interchangeable with every other page on the topic: a unique point of view, original data, or first-hand or expert experience. A list of ten generic SEO tips is commodity content. A write-up of what happened to your AI citations after you fixed a CDN block is not, because nobody else can write it.

Practical ways to add it: a test you ran, a screenshot of a real result, the price you actually paid, a mistake and how you fixed it, or a comparison you made yourself. Google also warns that creating pages for each fan-out query to manipulate AI answers breaks its scaled content abuse spam policy, so one strong page beats twenty thin variations.

4. Answer the question first, then support it

• Put a direct answer to the page’s main question in its first one or two sentences.
• Start each section with its conclusion, then explain.
• Use numbered lists for steps and tables for comparisons when they genuinely help.
• Write paragraphs that still make sense when read on their own.

This is not “chunking”. Google says there is no need to cut pages into tiny pieces and no ideal page length, because its systems understand pages that cover several topics. Clear structure is for readers first; AI systems benefit from the same clarity.

5. Back claims with named sources, quotes and figures

In the Princeton-led GEO study, adding citations, quotations from relevant sources and statistics raised a source’s visibility in generative engine answers by up to 40% in the authors’ experiments, with results that varied by subject. It was a research setup rather than Google’s or OpenAI’s live systems, but the direction matches common sense: specific, attributable facts are easier to quote than vague claims.

Compare “FAQ rich results stopped appearing in Google Search on 7 May 2026” with “FAQ snippets changed recently”. The first can be quoted and checked; the second can’t. Link the source next to the claim, and date facts that change.

6. Show who is behind the content

Name the author and show relevant experience, keep an honest About page, publish an editorial policy, and show when a page was last updated. These are trust signals for readers and for quality systems. When facts change, update or remove the outdated claim; changing the date alone helps nobody.

7. Use structured data for what it actually does

Google is explicit that structured data is not required for its generative AI features and that there is no special schema to add for them. Keep using it where it earns rich results and describes your business accurately (Organization, Article, Product, LocalBusiness), and make sure it matches what the page visibly shows. Note that FAQ rich results stopped appearing in Google Search on 7 May 2026, so FAQ markup no longer earns a special listing there. You can check your markup with our free schema validator.

8. Earn genuine mentions, not manufactured ones

Google’s guide lists chasing inauthentic mentions as less helpful than it looks: its core systems favour high-quality content and its spam systems discount manipulation. What does help is being talked about for real reasons, such as reviews, listings where your customers already look, original data others cite, and interviews or guest appearances on relevant sites. Keep your brand name, description and profiles consistent everywhere, so any system can tell the mentions refer to you. Our guide to backlinks explains the difference between earned and bought links.

9. Measure AI visibility every month

• Google: clicks from AI Overviews and AI Mode are included in Search Console’s Performance report, and the Generative AI performance report added in 2026 shows visibility in Google’s generative AI features.
• Bing and Copilot: the AI Performance report in Bing Webmaster Tools shows citations, cited pages and the grounding queries behind them.
• ChatGPT, Perplexity and others: there’s no official dashboard, so keep a fixed list of about 20 real customer questions, ask them every month, and record whether you were mentioned and whether you were cited. AI SEO tools can automate this.
• Analytics: track referrals from chatgpt.com, perplexity.ai, copilot.microsoft.com and gemini.google.com in GA4.

Treat third-party “AI visibility scores” as estimates. Google’s guide states plainly that no third-party tool has access to its internal ranking or AI systems.

Which GEO tactics can you skip?

Google’s guide includes a mythbusting section, and it removes a lot of busywork that GEO checklists still recommend.

• An llms.txt file for Google. Google Search doesn’t use it, and Google says keeping one will neither harm nor help your visibility there. Neither OpenAI’s nor Perplexity’s crawler documentation mentions it either.
• Chopping pages into tiny chunks. There is no ideal page length, and Google’s systems understand pages that cover several topics.
• Rewriting everything “for AI”. AI systems understand synonyms and plain meaning, so chasing every long-tail wording isn’t needed.
• Buying or faking mentions. Spam systems are built to discount manipulation.
• Special schema to get into AI answers. Structured data isn’t required for Google’s generative AI features.
• Blocking Google-Extended to leave AI Overviews. Google-Extended covers training and grounding in some of Google’s other AI systems, not Search. To limit what Google shows from a page, use nosnippet, data-nosnippet, max-snippet or noindex.

GEO vs SEO: what is actually different?

The foundations are shared. What changes is where you appear and how you measure it.

For a longer side-by-side comparison, see this GEO vs SEO breakdown. Paid placements are a separate channel again: ChatGPT now sells ads that sit apart from its answers, and our guide to ChatGPT ads explains how buying a placement differs from earning a citation.

What we learned measuring our own AI citations

In late October 2026 we asked Perplexity 21 questions that pages on this site answer, using SEO Packet’s AI visibility check. Soft Learn Hub was cited as a source in 2 of the 21 answers: a question about our brand and its free tools, and a question about a free letterhead maker that exports to Word and PDF. Four lessons stood out.

1. The page that got cited was a working tool. Our letterhead maker has a specific, checkable feature set, which made it the non-commodity answer to a specific question.
2. Platform questions went to the platform first. Answers about Facebook Pages, Google Ads Editor and Figma cited the official help centres. A third-party guide on those topics has to add something the official docs don’t, such as troubleshooting or a comparison.
3. AI referrals were already a bigger channel than Google clicks. Between 30 September and 27 October 2026, GA4 recorded 63 sessions from AI assistants and 29 from organic search on this site.
4. A security setting nearly hid us. The CDN bot challenge described above blocked automated crawlers without ever looking like an error to human visitors.

What we changed: each of those 21 questions now has a direct, sourced answer on the page it belongs to, the FAQ markup on those pages matches what visitors see, and the CDN settings are on our fix list. We’ll re-run the same 21 questions in late November 2026 and update this section with the result.

A 30-day GEO plan for a small website

One focus per week, starting with access, because a page AI search can’t fetch can’t be cited.

1. Week 1, access: check indexing in Search Console, keep Search generative AI set to Include, and allow search crawlers in robots.txt and in your CDN or firewall.
2. Week 2, answers: pick your 10 most important pages, answer each page’s main question in its first two sentences, and add named sources, figures and dates.
3. Week 3, authority: show the author, their experience and an updated date, add one piece of original data or a real example, and work on genuine mentions.
4. Week 4, measurement: run your 20 prompts, check the Generative AI and Bing AI Performance reports, note AI referrals in analytics, and repeat every month.

Frequently Asked Questions

What are GEO SEO best practices to get cited in AI Overviews and ChatGPT?

Start with normal SEO: AI Overviews draw on pages Google can crawl and index, and ChatGPT search finds pages through OpenAI’s OAI-SearchBot, so keep both allowed. Then answer the question in the first sentence, back claims with specific numbers and named sources, keep structured data matched to what the page visibly shows, and show who wrote the page and when it was updated. No markup or file guarantees a citation, so track a fixed set of prompts and check which pages get cited. See Google’s guide to AI features and your website and OpenAI’s crawler documentation.

Does GEO replace SEO?

No. Google describes AEO and GEO as the same work it calls SEO, because its AI features rely on its core ranking and quality systems. GEO adds a few things on top: allowing each AI platform’s crawler, writing quotable passages, and measuring citations as well as clicks.

What is the difference between GEO and AEO?

Mostly the name. Answer engine optimization (AEO) came from optimizing for featured snippets and voice answers, while generative engine optimization (GEO) refers to AI systems that write answers. In practice both mean making pages easy for answer systems to find, trust and quote.

Does structured data help you get cited by AI?

Not directly. Google says structured data isn’t required for its generative AI features and there is no special schema for them. It is still worth using accurately for rich results and to describe your business, as long as it matches the visible page.

Is llms.txt required?

No. Google’s 2026 guide says Google Search doesn’t use llms.txt and that keeping one will neither harm nor help your visibility there. Some tools read it, so it does no damage, but don’t expect it to earn citations.

How do I know if AI engines are citing my site?

Use Search Console’s Generative AI performance report for Google, the AI Performance report in Bing Webmaster Tools for Copilot and Bing, and a fixed set of real customer questions you ask ChatGPT and Perplexity every month. Referrals from chatgpt.com and perplexity.ai in your analytics confirm that citations turn into visits.

How long does GEO take to show results?

It depends on how quickly each system re-reads your pages. OpenAI and Perplexity say robots.txt changes take up to about 24 hours, but citations only change once pages are recrawled and re-evaluated, which can take weeks. We compare the same prompts month to month and judge changes over two to three months.

Sources

• Google Search Central: Optimizing your website for generative AI features on Google Search (published May 2026)
• Google Search Central Blog: A new resource for optimizing for generative AI in Google Search (15 May 2026)
• Google Search Central: AI features and your website
• Google Search Central: Latest documentation updates (FAQ rich result deprecation, May 2026)
• OpenAI: Overview of OpenAI crawlers
• Perplexity: Perplexity crawlers
• Bing Webmaster Blog: Introducing AI Performance in Bing Webmaster Tools
• Aggarwal et al.: GEO: Generative Engine Optimization (KDD 2024)