How to Get Cited by AI Search Engines in 2026

You get cited by AI search engines in 2026 by being four things, in this order: fetchable, answer-shaped, fresh, and quotable. Fetchable means the retrieval crawlers can reach you. Answer-shaped means your content answers the question inside the first forty words. Fresh means you touched the page recently. Quotable means your sentences can be lifted verbatim without embarrassing you. Get those four right and you are inside the pool that agentic search engines draw from — whether or not you rank anywhere near the top of Google.

The old playbook — rank top three, earn links, win the featured snippet — still helps, but it is no longer sufficient. The engines that answer questions now run their own retrieval pipelines, their own freshness bias, and their own source preferences. The evidence below comes from real studies (Ahrefs, AirOps, Search Engine Land, MR Research) and from live SERP checks I ran while writing this post. No affiliate links, no sponsorships. Just data.

TL;DR — the 2026 AEO playbook

1. Allow the retrieval bots. OAI-SearchBot, Claude-SearchBot, PerplexityBot, and the -User crawlers must be able to crawl you. Blocking them removes you from live AI answers today.
2. Put the answer in the first 40 words of every section, then prove it below. AI extractors quote the top of the block.
3. Refresh commercial pages every 30 days, everything else every 60-90. AI assistants cite content ~26% fresher than Google organic results on average, and stale pages lose citations at twice the rate.
4. Use FAQPage, Article, and Person schema server-side, matching visible content. Correct schema helps clarity; sparse schema hurts.
5. Treat llms.txt as agent infrastructure, not a citation lever. Adoption is rising fast but almost nobody reads it yet.
6. Watch citations, not just rankings. Only ~7-12% of pages AI engines cite appear in Google’s top 10. The two leaderboards have separated.

That is the whole argument. Everything below is the proof, the configs, and the checklists.

The SERP this targets is thinner than you’d think

Before writing a word, I ran live Google queries through a SERP API and measured what the “get cited by AI” cluster actually looks like. Six queries, US/English, checked August 4, 2026:

The cluster is thin: eight to ten organic results per query, zero answer boxes, zero PAA, across all six. Google has not decided that any of these queries deserves a box yet. That is the window. Thin SERPs with no featured answer are exactly where an answer-first page can take the spot — and the first mover often defines the answer that the extractors and Google’s own box end up using later.

The target query also leaks intent in its related searches: “how to get cited by ai reddit” and “how to get cited by ai apa” both surfaced. That means the query is being asked by two audiences at once — practitioners looking for community tactics and students looking for citation formatting. A page that serves both the mechanical “how does citation work” answer and the operational “what do I change” answer captures the whole demand.

For reference, three representative calls through the SERP proxy took 6.2s, 3.6s, and 3.3s round-trip. Cheap enough that a daily citation watch is a normal operating cost. More on that at the end.

What actually gets cited?

The most important finding in the last year of AEO research is that citation is a two-stage pipeline: retrieval, then synthesis. The engine retrieves a set of candidate pages from its own index, then decides which of those actually make it into the answer as a visible citation. The stages run independently.

The numbers are brutal. Audits of 21,143 citations found the engines filter out roughly 95% of retrieved content before generating an answer, and only about 15% of retrieved pages ever earn a visible citation. Retrieval alone is not enough. You have to survive the synthesis pass too — which is why “answer-shaped” content matters.

Each engine also has a different source personality, and they barely overlap. Only 2.7% of domains are cited by all five major engines; nearly 70% of cited domains appear on exactly one. ChatGPT shares about 10% URL overlap with Google’s organic top 10. Search Engine Land’s analysis of ~8,000 citations across ChatGPT, Gemini, Perplexity, and AI Overviews sketched the personalities:

• ChatGPT is the authority seeker. It leans on Wikipedia (~27% of citations), news (~27%), and blogs (~21%). It cites the fewest sources per answer and rarely touches UGC forums or vendor blogs.
• Gemini is the balanced synthesizer. Blogs (~39%) and news (~26%) dominate; it cites roughly eight brands per answer.
• Perplexity is the expert-and-review curator. It pulls editorial (~38%), news (~23%), and expert-review domains like NerdWallet and Consumer Reports (~9%), and it spreads citations across ~13 brands.
• Google AI Overviews is the broad aggregator. Blogs (~46%) and news (~20%) form the core, and Reddit is its single most-cited domain.

Two structural facts matter more than the personalities. First, 82.5% of AI citations link to deeply nested pages, not homepages — the extractor wants the specific answer page, not your brand front door. Second, the overlap with the old leaderboard is collapsing. AI Overviews pulled 38% of its citations from the organic top 10 in February 2026, down from 76% in July 2025. Ahrefs found that only 12% of URLs cited by AI assistants rank in Google’s top 10. Optimizing only for the top 10 is optimizing for a minority of AI citations.

Read the chart as scarcity. ChatGPT hands out 3-4 citations per answer. Perplexity hands out ~9. If your content is not in the top handful of candidates for a ChatGPT query, you will not be cited — there is no “position 11” for ChatGPT citations. The practical implication: chase the queries where you can plausibly be one of the three best candidates in the entire web, not the queries where you might crack the top twenty.

One more lever worth naming before we get to configs: entity richness. A 2026 audit found pages with rich entity context (named authors, organizations, clear definitions, sameAs links) earn up to 2.67x more citations, and that query-passage similarity predicts citation roughly 7x better than domain authority. Content that literally restates the question’s language, in the right structure, wins.

Does llms.txt matter?

Short answer: not for citations yet — but yes as agent infrastructure. The data is unusually clear here because Ahrefs analyzed server logs across 137,000 domains:

• 28% of domains publish an llms.txt file.
• 97% of those files received zero requests in the study window.
• Of the ~3% that were read, 96% of the requests came from bots, not people.
• Only about 1.1% of the requests came from AI retrieval bots (OAI-SearchBot, PerplexityBot, and friends).

Adoption is exploding even while usage stays near zero. Originality.AI’s tracker counted 4,088 live llms.txt files in June 2025 and 36,120 by May 2026 — an 8.8x jump in a year. Top-10k-domain adoption sits around 5.86%, up from ~0.3% a year earlier. The people reading llms.txt, though, are mostly SEO audit tools (21.7%), generic crawlers (13.1%), and AI coding agents like Claude-Code (10.5%). The actual AI search bots are barely interested.

That gap is why I call it infrastructure, not a lever. If you publish documentation, API references, or structured product data, an llms.txt that points at those pages is a low-cost signal for the agents and coding tools that will matter later. Keep it curated — 10 to 30 links, not 200. Consider llms-full.txt only if you can maintain it, and remember that a file agents are designed to trust is also an injection surface: researchers are already probing llms.txt files for prompt injection. Do not put instructions in it that you would not put in your terms of service.

What goes in robots.txt?

The single most important config decision in AEO is separating training crawlers from retrieval crawlers. Training crawlers harvest content to build future models; blocking them only affects tomorrow’s training data. Retrieval crawlers fetch your page live when someone asks a question; blocking them removes you from today’s answers. They are different bots with different user-agent tokens, and you need to treat them differently.

This is the allow/block matrix, current as of mid-2026:

The default recipe for a publisher that wants AI citations:

User-agent: *
Allow: /
Sitemap: https://example.com/sitemap.xml

If you want to block training bots while keeping retrieval bots (the most common publisher stance — 79% of top news publishers now block training crawlers), name the training bots explicitly. GPTBot is blocked by 62% of them, ClaudeBot by 69%, PerplexityBot by 67%. Note the trap: blocking PerplexityBot is how many sites accidentally killed their Perplexity visibility, because PerplexityBot is also the retrieval bot. The safe split is:

User-agent: GPTBot
Disallow: /

User-agent: ClaudeBot
Disallow: /

User-agent: Google-Extended
Disallow: /

User-agent: CCBot
Disallow: /

User-agent: *
Allow: /

Retrieval bots like OAI-SearchBot, Claude-SearchBot, and Perplexity-User fall through to the permissive catch-all.

Three mistakes kill people here. First, tokens are case-sensitive: gptbot does not match GPTBot. Second, naming only GPTBot misses OAI-SearchBot and ChatGPT-User — the bots that actually drive your ChatGPT visibility. Third, believing robots.txt is a security boundary. It is a polite request, not a lock. And the data says blocking barely protects you anyway: 70.6% of sites that blocked AI crawlers still appeared in citations, while blockers saw a 23.1% traffic decline. Blocking is a values decision about training data, not a citation strategy. If you genuinely want to prevent citation, you need authentication or noindex-level control, and even then live-fetch bots can ignore you.

How much does schema help?

Honest answer: it helps at the margins and hurts when done badly. The vendor studies report dramatic numbers. One audit found FAQPage pages cited 2.4x more often than prose-only Q&A, with the widest gap on Perplexity (3.1x). The same audit found Article schema with complete author and date fields cited at a 58% rate, and Person/Organization schema mainly providing entity context. But the one controlled test — Ahrefs’ 1,885-page experiment — found no causal lift: AI Overview citations fell 4.6% on schema pages relative to controls, ChatGPT rose 2.2%, AI Mode rose 2.4%. All treated as noise. Worse, sparse or generic schema underperformed having no schema at all (41.6% vs 59.8% citation rate).

Here is how I reconcile those: schema does not get you cited, but it does keep you from being misread. The extractor needs to find your answer, your date, and your author. Correct, complete markup makes that reliable; broken or mismatched markup makes it worse than nothing. Google stopped showing FAQ rich results in May 2026, but every major AI crawler still parses FAQPage JSON-LD. The markup outlived the widget.

Priority order, from evidence to vibes:

Two non-negotiable technical rules. Render schema server-side; JSON-LD injected through a tag manager is not reliably parsed by AI crawlers. And match the markup to what is visibly on the page — a FAQPage with answers that do not appear in the body is a mismatched page, and extractors drop mismatched pages.

Freshness is the lever everyone skips

Freshness is the least glamorous and most consistent signal in the entire AEO literature, and it shows up in every major study. Ahrefs analyzed 16.975 million cited URLs across ChatGPT, Perplexity, Gemini, Copilot, and AI Overviews and found AI assistants cite content that is 25.7% fresher than what Google organic serves — 1,064 days average cited-page age versus 1,432. Perplexity, the strongest freshness bias in the audit, cites content updated two hours earlier about 38% more than month-old content. AirOps, looking at 4,000+ ChatGPT-cited pages, found over 70% were updated within the past 12 months, and pages not touched in a year were more than twice as likely to lose citations. Content updated within the last 30 days holds the strongest citation probability across engines. After 60 days without an update, citation rates start to drop.

Cadence, by content type:

The fine print: update dates honestly. Pages with a visible last-updated timestamp earn ~1.8x more citations than pages without one, and AI engines read last-updated more than publish date. But bumping a date without changing content is the one move John Mueller and every AI crawler now punish. Refresh means actually changing the substance, then letting the timestamp reflect it.

Writing for the extractor

Every engine in the pipeline is doing the same thing at the end: pulling language out of your page to put in an answer. So structure matters more than prose style. The audit data points to the same handful of moves:

• Answer in the first 40 words. Queries about a question reward a direct answer at the top of the section, not an intro paragraph.
• Use tables and lists for anything enumerable. Tables get ~2.5x the citation likelihood of prose, per one 21,143-citation audit.
• Add FAQ blocks. FAQ structures are cited 28-40% more often; a FAQ block at the end of a section is the cheapest structured content you can add.
• Write quotable sentences. A single clean declarative sentence with a number in it is what gets lifted. “Fresh content is cited ~26% more than stale content” beats three meandering sentences.
• Restate the query. Query-passage similarity predicts citation ~7x better than domain authority. If the user asked “how to get cited by AI,” your page should contain that phrase in the answer.
• Do not keyword-stuff. Extractors penalize repetition, and sparse structure hurts. Write for one clean answer.
• Keep sentences under 20 words. Short sentences survive extraction better than long clauses.

This post is written to its own advice: every section leads with the answer, the configs are in tables, and the FAQ below is markup-ready.

How to watch your own citations

You cannot manage what you do not measure, and the measurement here is different from classic rank tracking. Rank tracking tells you where you sit in Google. Citation watching tells you whether ChatGPT, Perplexity, and AI Overviews name you as a source — which, per the data above, is now a mostly separate leaderboard.

The cheapest setup is a scheduled set of brand-plus-question queries, run against a SERP API, with the results logged and diffed weekly. Any search API works for this. I run these checks on Keirolabs — its search endpoint costs $0.25 per 1k requests with 500 free credits a month, which makes a daily citation watch effectively free. If you prefer something else, the alternatives in my web search API roundup all do the same job; the point is the loop, not the provider. Ask your questions, note whether your domain appears in the answer or the citations, and re-check after any content refresh.

The questions that matter are your money questions, not your vanity terms. “Best X for Y,” “how to fix Z,” “X vs Y” — the buying-intent queries are the ones with the fewest citations per answer and therefore the highest per-slot value.

The 10-point AEO audit checklist

• 1. Robots.txt allows OAI-SearchBot, Claude-SearchBot, PerplexityBot, and the -User crawlers. Blocking training bots is optional; blocking retrieval bots is not.
• 2. Verify crawl access in server logs — an allow rule you never check is a theory, not a config.
• 3. FAQPage, Article, and Person schema are server-side, complete, and match visible content.
• 4. Every section on your money pages answers the question in its first 40 words.
• 5. Tables or lists, not paragraphs, for anything enumerable.
• 6. A visible, honest last-updated timestamp on every article.
• 7. A refresh calendar: commercial pages at 30 days, hubs at 60, reports at 90.
• 8. An llms.txt published only if you have structured docs or product pages, kept to 10-30 links.
• 9. A monthly citation audit across ChatGPT, Perplexity, and AI Overviews — not just a Google rank report.
• 10. Entity context: named author with Person schema, Organization connected via @id, sameAs links where they exist.

Run the list top to bottom. Numbers one through three are one-time fixes. Numbers four through seven are a content discipline. Numbers eight through ten are the ongoing loop.

FAQ

How do AI search engines decide what to cite?

Through a two-stage pipeline: retrieval, then synthesis. Each engine pulls its top sources from its own index — ChatGPT via Bing, AI Overviews via Google’s core ranking systems, Perplexity via live web search — then filters out roughly 95% of what it retrieved. Only about 15% of retrieved pages earn a visible citation, and each engine has a different source preference: ChatGPT skews to established media, Perplexity to expert-review and editorial sites, AI Overviews to blogs and community content.

Does llms.txt help you get cited by AI?

Not yet. In an Ahrefs study of 137,000 domains, 28% published an llms.txt file but 97% of those files were never fetched, and only about 1.1% of the requests came from AI retrieval bots like OAI-SearchBot or PerplexityBot. Publish it as cheap agent-readiness infrastructure if you have docs or product pages, but do not treat it as a citation lever.

Should I block GPTBot and ClaudeBot in robots.txt?

You can block training crawlers without losing citations, but never block the retrieval crawlers that answer live queries: OAI-SearchBot, ChatGPT-User, Claude-SearchBot, Claude-User, and PerplexityBot. Blocking AI crawlers does not reliably prevent citation anyway — over 70% of sites that block still appeared in citations, while blockers saw a 23% traffic decline.

Does schema markup (JSON-LD) increase AI citations?

Vendor studies report big lifts — FAQPage pages cited 2.4x more than prose-only Q&A in one audit — but Ahrefs’ controlled 1,885-page test found no causal lift and found that sparse or mismatched schema underperforms no schema at all. Use FAQPage, Article, and Person markup correctly and match it to visible content. Treat schema as clarity, not a shortcut.

How often should I update content to stay cited?

Every 30-60 days for competitive topics. In AirOps’ analysis of 4,000+ ChatGPT-cited pages, over 70% were updated within 12 months, and pages untouched for a year were more than twice as likely to lose citations. Add a visible last-updated timestamp — pages with one earn about 1.8x more citations.

What is the fastest way to get cited by ChatGPT?

Match what ChatGPT actually cites: established media and high-authority domains with answer-shaped, fresh content. ChatGPT cites only 3-7 sources per answer, roughly 87% of its citations match Bing’s top results, and only about 7% of the pages it cites appear in Google’s top 10 — so optimize for Bing-indexed, first-40-words answers.

Sources: Ahrefs llms.txt study (137K domains, May 2026); Ahrefs freshness study (16.975M citations); AirOps stale-content report and refresh-cadence guide; Search Engine Land analysis of ~8,000 AI citations; Search Engine Land AI Overviews top-10 citation tracking (76% to 38%); MR Research citation-divergence audit (21,143 citations); originality.ai llms.txt tracker; robots.txt crawler references from SeoMods, Crawlytics, and AmICited; schema audits from GEOlikeaPro, citability.dev, and Ahrefs. SERP checks run live August 4, 2026.