Does page speed affect whether AI cites my site?

Page speed is not a documented lever for AI citations. OpenAI, Google, Anthropic and Perplexity describe how their crawlers work, and none of them names load time as a reason a page gets cited.

Google is the clearest. Its AI features documentation says a page must be indexed and eligible for a snippet to appear in AI Overviews or AI Mode, and that "there are no additional technical requirements." Whatever role speed plays in classic Google Search, AI features add nothing on top.

The research points the same way. A July 2026 critical survey of 45 GEO studies (arXiv:2607.14035) found that no reviewed technique shows a stable, cross-platform causal effect on discoverability. Speed was not among the levers any study isolated.

What the evidence does support is access. A crawler that receives your content can consider it. A crawler that gets an error, a challenge page or an empty HTML shell has nothing to work with. That is a yes or no question, not a question of milliseconds.

These are the failures with a documented mechanism. Each one removes your page from the candidate pool entirely.

• 403 Forbidden. Bot management, WAF rules and CDN settings can turn away crawlers your robots.txt allows. OpenAI's own guidance warns that these systems "can mistakenly block legitimate crawlers, often returning 403 Forbidden errors."
• 429 Too Many Requests. A rate limit tells the crawler to back off. Until it retries successfully, it holds no copy of the page.
• Challenges and CAPTCHAs. A JavaScript challenge returns a 200 status with no real content. OpenAI notes that because its crawlers are automated, "these checks may block access."
• JavaScript-only content. Vercel's crawler study found that ChatGPT and Claude crawlers fetch JavaScript files but "don't execute them. They can't read client-side rendered content." Googlebot is the exception because it renders.
• Failed or abandoned fetches. A server that errors or never responds delivers nothing. Our evidence base has no published timeout figure for any AI crawler, so treat this as a reliability issue, not a speed target.

The tabs, accordions and load more check covers a common variant: content injected only after a click.

The strongest evidence for any publisher-controlled factor is server-side rendering of the content that matters. Vercel's study, re-confirmed independently for the major AI crawlers as of June 2026, shows those bots read raw HTML only.

The second strong finding is that access is a two-part gate. OpenAI's help centre says eligibility for ChatGPT search means allowing OAI-SearchBot in robots.txt and confirming your host or CDN allows traffic from OpenAI's published IP addresses. Robots.txt alone is not enough. See whether ChatGPT needs OAI-SearchBot to cite you for how the per-bot rules work.

What the evidence does not cover:

• No study compares citation rates for fast and slow pages.
• No operator publishes a load-time threshold for AI retrieval.
• Nobody can see whether your page was retrieved for a given user's question.

So anyone quoting a speed-to-citation number is quoting something our evidence base cannot find.

Speed and access are different questions, but they touch in a few places.

Rate limits triggered by crawl volume

If your server rate-limits aggressively to protect performance, it may answer AI crawlers with 429s. The fix is a crawler-aware rule, not a faster page.

Very large HTML payloads

Google stated in March 2026 that Googlebot fetches up to 2MB per URL (64MB for PDFs) and ignores bytes past that point. Other crawlers that set no limit default to 15MB, per the same post. Google adds that most of the web will never hit this. Check it only if a page is unusually heavy.

CDN defaults you did not set

From 2026-09-15, Cloudflare set Training and Agent crawler categories to blocked by default on ad-displaying pages for new domains, new sites on existing accounts and existing free-tier customers. A site can lose AI crawler access without anyone editing it. Teams sometimes enable bot protection to improve speed under load, which can cause the same result.

Crawlers visit but nothing is cited

Crawler hits in your logs only prove a request arrived. If you see visits but no mentions, read why AI crawlers visit but never cite you.

Work through access before speed. Each step has a clear pass or fail.

1. Request a page as each AI crawler. Record the status code and whether a challenge page comes back. The free AI crawler checker does this per bot, and 99Visibility's AI Crawler Checker shows which ones your firewall turns away.
2. Compare robots.txt with what the server does. A bot allowed in robots.txt but refused at the edge is invisible. The firewall testing playbook walks through it.
3. Diff raw HTML against the rendered page. Anything that appears only after JavaScript runs is missing for most AI crawlers. Run the JavaScript rendering check on your pricing and product pages first.
4. Review CDN bot settings. Look at AI crawler categories, especially if you use Cloudflare.
5. Then work on speed for people. It may help your visitors. It is not a documented AI citation lever.

None of these steps guarantees a citation. They remove the blocks that make one impossible.