AI SEO Tips That Hold Up Under the Evidence

Most AI SEO tips in circulation were never tested on the thing you care about: whether an AI engine finds your page and uses it. A July 2026 critical survey of 45 GEO studies (arXiv:2607.14035) concluded that already-retrieved content can change how it gets cited, but no reviewed technique showed a stable, cross-platform effect on discoverability.

That gap matters. The famous GEO experiments placed documents directly into the model's context, then rewrote them. They measured what happens after a page is already in the answer pool. They never measured how a page gets into that pool.

The best-designed of those studies found the rewriting tactics ineffective or harmful. C-SEO Bench, a NeurIPS 2025 benchmark, reported that most conversational SEO methods were largely ineffective and frequently lowered a document's ranking.

So a useful tip has to pass one test: is there a documented mechanism, stated by the engine operator or shown in a controlled study, and can you check it yourself? If you want the longer list of claims that fail this test, see GEO claims that don't hold up.

An AI answer that mentions your brand passes through four stages. Sorting tips by stage tells you which ones you can actually act on.

The access stage is where the evidence is strong. It's also mostly subtractive. There's no markup or phrasing that makes a page more eligible. There are only ways to be ineligible.

Retrieval is also discretionary. As of October 2026, Google AI Overviews often don't trigger, ChatGPT decides per query whether to search, and Claude.ai search is a user toggle. Any AI visibility number blends answers given with and without live retrieval. That's one reason GEO studies disagree with each other.

Server-render the facts that matter

Vercel's crawler study found that ChatGPT and Claude crawlers fetch JavaScript files (11.50% and 23.84% of requests) but don't execute them, so they can't read client-side rendered content. Googlebot renders; most AI crawlers don't.

Check: fetch a key page with JavaScript disabled and compare the text with the rendered page. If your pricing, product names or headings appear only after hydration, most AI crawlers never see them.

Allow each AI search crawler by name

robots.txt rules are per bot. OpenAI's crawler documentation says sites opted out of OAI-SearchBot won't be shown in ChatGPT search answers, and that each setting is independent of the others. GPTBot governs training only.

User-agent: OAI-SearchBot
Allow: /

User-agent: PerplexityBot
Allow: /

User-agent: Claude-SearchBot
Allow: /

Check: read your live robots.txt for each named retrieval bot. The common failure is a blanket AI block that removes search eligibility along with training.

Confirm your firewall agrees with your robots.txt

OpenAI's help centre says eligibility needs both: allow OAI-SearchBot, and confirm your host or CDN allows traffic from OpenAI's published IP ranges. A CDN can return 403 to a crawler that robots.txt allows. Cloudflare announced that from 15 September 2026 it blocks Training and Agent crawlers by default on ad-displaying pages for new domains and existing free-tier customers, so a site can lose access without anyone editing it.

Check: request a page with each AI crawler's user agent and record the status code. The AI Crawler Checker does this per crawler.

Remove accidental suppression

Google and Bing document noindex, nosnippet, max-snippet and data-nosnippet as the controls that keep content out of AI answers. They often arrive by template.

Check: search your HTML and headers for these directives and confirm each one is intentional.

Check index status engine by engine

Google says a page must be indexed and eligible for a snippet to appear as a supporting link in AI Overviews or AI Mode, with no additional technical requirements. Bing's index backs Microsoft Copilot. In Otterly's 2026 experiment, Google deindexed both test sites and its AI surfaces stopped citing them, while Bing kept them and Copilot kept citing them.

Check: open Google Search Console and Bing Webmaster Tools separately. A healthy one tells you nothing about the other.

State specific facts plainly

A price, a number or a one-line definition on a fetchable page gives an engine a sentence it can quote. This is the only positive experimental signal for on-page content, and it's modest.

Get named where engines already look

When engines cite other sites instead of yours, being named on those sites is the route into the answer. The evidence shows a strong association, not a promised effect.

Be clear with your team about this split before anyone promises results.

In your control:

• Whether your important content exists in raw HTML.
• Which crawlers your robots.txt allows.
• Whether your CDN, firewall or bot manager lets those crawlers through.
• Whether your pages carry noindex or snippet restrictions.
• Whether facts on your own site agree with each other across pricing, about and footer pages.

Not in your control:

• What a model learned before its training cutoff.
• Whether an engine decides to search for a given question.
• The queries the engine actually runs. OpenAI says ChatGPT search rewrites your question into one or more targeted queries.
• Where your page sits in the model's context list, which C-SEO Bench found mattered more than any rewrite.
• Brand prominence, which dominates citation patterns across studies and has no shortcut.

Fixing access doesn't guarantee citation. It decides whether your content can be considered at all. If someone asks whether there's a proven way to increase AI citations, the honest answer is no, but there are proven ways to stop blocking yourself.

Work from the cheapest, most decisive checks to the slowest ones.

1. Read robots.txt per bot. Confirm OAI-SearchBot, PerplexityBot, Claude-SearchBot and Googlebot are allowed. Decide on GPTBot, ClaudeBot and Google-Extended as a separate training question.
2. Test network reachability. Request key pages as each crawler. Any 403, 429 or challenge page is a block, whatever robots.txt says.
3. Run the rendering diff. Compare raw HTML with rendered text on your homepage, pricing page and top product pages.
4. Scan for suppression tags. Look for noindex, nosnippet, max-snippet:0 and data-nosnippet.
5. Check index status in Google and Bing. Use Search Console and Bing Webmaster Tools, not a third-party guess.
6. Audit your own facts. Make sure your price, founding year and product names match everywhere on your site.
7. Then work on presence elsewhere. Earn mentions on the sites engines already cite for your topic.

Steps 1 to 4 usually become developer work. Our guide to developer tickets for AI access fixes shows how to write them so they get done. For the longer-term plan around these steps, see building an AI SEO strategy on evidence.

These appear in many tip lists. Each one fails against a primary source.

• "Add schema to get cited." Google says structured data isn't required for generative AI search. Ahrefs' matched test of 1,885 pages found AI Overview citations fell 4.6% after schema was added. Existing markup does no harm; it just isn't an AI lever.
• "Add FAQ schema." Google deprecated FAQ rich results in May 2026. Q&A content can still be useful to readers. See whether you need an FAQ page for AI.
• "Publish an llms.txt." Google's AI optimization guide says Google Search ignores these files. No other engine documents reading one.
• "Chunk your content for LLMs." Google rejects this and says there's no ideal page length. The chunking research measures segmentation done by the retrieval system, not the publisher. See the document chunking study.
• "Write in an AI-friendly style." Google says you don't need to write in a specific way for generative AI search.
• "Improve your E-E-A-T score." Google says E-E-A-T isn't a specific ranking factor. There is no score.
• "Build domain authority." Domain Authority is a Moz metric. No engine computes it.
• "Allow GPTBot so ChatGPT finds you." GPTBot is for training. Search uses OAI-SearchBot.
• "GEO gets you up to 40% more visibility." That figure came from a lab engine with sources already loaded into context.
• "Ping IndexNow and AI will cite you faster." The IndexNow spec never mentions AI, and Google doesn't use it.

• Blocking all AI bots at once. One blanket rule can remove ChatGPT search eligibility when you only meant to opt out of training.
• Trusting robots.txt alone. A firewall rule or CDN default can override it silently. Test the actual response.
• Testing in a logged-in browser. Memory and custom instructions change ChatGPT's search queries. Test logged out, or don't treat results as reproducible.
• Reading one blended score. In Otterly's experiment, Copilot cited the test site thousands of times while ChatGPT and Google's AI surfaces barely did. Measure each engine separately.
• Mass-publishing thin pages. Google's spam policies treat many pages generated to manipulate rankings as scaled content abuse. Otterly's AI-generated test sites were deindexed with no manual action shown.
• Hiding text for AI crawlers. Prompt injection and cloaking exist, and research shows some agents can be manipulated. They're illegitimate and put your whole site at risk.

For the full evidence ranking behind each tip, read AI SEO best practices, sorted by evidence.