Ahrefs Guide: 44.2% of AI Citations Come From First 30% of a Page
Ahrefs researcher Mateusz Makosiewicz published a field-tested playbook on AI search optimization. The numbers inside point to where AI citations actually come from, and which pages lose them fastest.

- 44.2% of AI citations came from the first 30% of a page (Kevin Indig research cited by Ahrefs)
- URLs cited by AI assistants were 25.7% fresher than URLs in Google's organic results
- 57.1% of 'best X' list articles cited by ChatGPT had been updated post-publish; 79.1% were last updated in 2025
- In an Ahrefs test, five major AI systems ignored JSON-LD, hidden Microdata, and hidden RDFa, relying on visible HTML instead
- May 2026: a competitor's updated SEO pricing study overtook Ahrefs' 2024 version in AI citations
By Olivia Hart
A new guide from Ahrefs marketer Mateusz Makosiewicz pulls together field-tested advice on AI search optimization. The most concrete finding inside: in Kevin Indig's research, 44.2% of AI citations came from the first 30% of a page.
That single data point shifts how you should structure any answer-style page.
What does Google's official position look like?
Google says its AI search features still rely on its core search systems. The crawl, index, and ranking pipeline hasn't changed. Microsoft adds that Bing's index is evolving from a list of pages into a source of facts AI systems can pull from to build answers.
The throughline from both: AI search still depends on traditional search, but ranking well is no longer the whole job. Your content also has to feed facts that AI systems can read, trust, and quote.
Takeaway: A page still has to be crawlable, indexable, and useful.
Does structured data help?
Less than you'd think. In an Ahrefs test, five major AI systems ignored JSON-LD, hidden Microdata, and hidden RDFa. They read visible HTML instead. Pricing, product features, or specs that live only in markup, or sit inside images, won't show up in answers.
That single test is a tool signal, not a confirmed Google or Microsoft statement. The safe move is to repeat critical facts in visible text.
Takeaway: If only markup or an image carries the data, an AI likely won't see it.
How important is freshness?
Very. URLs cited by AI assistants were, on average, 25.7% fresher than URLs in Google's organic results, per the Ahrefs study "Do AI Assistants Prefer to Cite Fresh Content?" Separately, in research on "best X" lists, 57.1% of the articles ChatGPT cited had been updated after their original publish date. Another 79.1% were updated in the same year as the study (2025).
The trend hits pricing, integration, and comparison pages hardest.
Ahrefs felt this pressure first-hand. Its cost-of-SEO research went live in 2024 without AEO in mind, and still ranks among the most-cited pages in the topic. In May 2026, a competitor released an updated version using 2026 data. That newer page soon pulled ahead in AI citations.
Takeaway: Stale facts lose citations to fresher posts on the same topic.
What makes a page get cited?
Makosiewicz, an SEO and marketing educator at Ahrefs, argues pages need four things working at once:
- Access. Crawlable pages, no broken routes for AI bots, no WAF or CDN rules blocking them. Cloudflare's AI Crawl Control settings can quietly shut crawlers out.
- Match. A dedicated page for the exact query beats a buried answer on a broader one.
- Retrievable value. Original examples, survey data, benchmarks, workflows, expert judgment. Generic summaries get replaced.
- Atomic content. Self-contained sections, each block should still make sense when lifted out alone. Ahrefs recommends BLUF (Bottom Line Up Front), so the answer lands in the first one or two sentences.
HubSpot recently reported a similar win. It built pages closely matching specific prompts and saw AI visibility rise across additional industries and local markets.
Takeaway: Write for people, write in standalone chunks, and put the answer first.
What can AI agents handle here?
Plenty of the mechanical work, Makosiewicz says. An agent can check robots.txt, status codes, redirects, WAF rules, blocked bot requests, and pages behind logins. It can flag thin sections, vague claims, and missing facts. Tools cited in the piece include Ahrefs Site Audit, Web Analytics, Bot Analytics, and Brand Radar. The author also releases custom audit skills on GitHub for Letaido, Claude Code, ChatGPT Agents, and Codex.
The agent still can't manufacture real expertise, original data, customer insight, or facts about a product. Those have to come from the team. "An AI agent can help check access, review passages, spot inconsistencies, and analyze citation data; you still decide what to publish," Makosiewicz wrote.
Takeaway: Hand the agent the audit. Keep the judgment for yourself.
What's worth monitoring next?
Citation stability. Older pages with strong AI traction are the first to lose ground when competitors publish fresher takes. Open Ahrefs Brand Radar, sort cited pages by age, and check for updates that have already outpaced yours. Set a recurring digest on the URLs you care about most.
That's the watchlist worth setting up this quarter.
via developers.google.com (Original)
More from Olivia Hart
Show full bio
Staff writer covering money, funding and market shifts around Search rankings and algorithm updates, generative engine optimization at SERP Journal.
123 articles