AI SEO for Your Website: Where to Start in Week One

99Visibility Research · Updated October 21, 2026 · 8 min read

Short answer

AI SEO for your website starts with access, not tricks. In week one, confirm each AI search crawler can reach your pages and put your key facts in the HTML before JavaScript runs. Then check that Google and Bing index those pages, and see what ChatGPT, Perplexity and Google's AI features say about you. These steps remove blocks. None of them guarantees an engine will cite you.

Key takeaways

The levers with real evidence behind them are access levers: crawler permission, network reachability, server-rendered content and index eligibility.
robots.txt rules are per bot, so blocking GPTBot stops training use while OAI-SearchBot controls ChatGPT search.
A firewall or CDN can turn away an AI crawler that robots.txt allows, and no dashboard will tell you.
Check index status engine by engine, because Google and Bing feed different AI surfaces.
Schema, llms.txt and AI-friendly rewrites have no credible evidence of getting a site cited, so they don't belong in week one.

What AI SEO for a website actually involves

An AI answer passes through four stages. Each one has a different owner.

Parametric: what the model knows with no web access. It is frozen at a training cutoff. As of October 2026, OpenAI documents a February 16, 2026 cutoff for its GPT-5.6 models and August 31, 2025 for the Instant path.
Access: whether a crawler is allowed to fetch and store your page at all.
Retrieval: whether your page enters the model's context for a given question.
Generation: which of the retrieved pages the model uses and cites.

Most GEO advice targets the last stage. A July 2026 critical survey of 45 GEO studies (arXiv 2607.14035) found that already-retrieved content can change how it gets cited, but no technique showed a stable, cross-platform effect on being discovered in the first place.

What that means for you: the work with the strongest evidence is removing blocks at the access stage. That's where a first-week plan should start.

What you control and what you don't

Be clear with yourself and your team about where your effort lands.

Stage	Who decides	What you can do
Parametric	The model vendor, at training time	Nothing this quarter
Access	You and your CDN	Fix robots.txt, firewall rules, rendering, noindex and snippet tags
Retrieval	The engine, per query	Be indexed, be topically relevant, be named on sites engines already cite
Generation	The model	Very little, and the tested tricks mostly failed

Retrieval is discretionary on every platform. ChatGPT decides per question whether to search. Google AI Overviews often don't trigger at all. So even a perfectly accessible site appears in some answers and not others.

The next move is to fix everything in the access row, because it is the only row you fully own.

Days 1 and 2: confirm AI crawlers can reach you

Read your robots.txt per bot

Each AI company runs separate bots for training and for search. OpenAI says of its crawlers: "Each setting is independent of the others." GPTBot governs training. OAI-SearchBot governs ChatGPT search. A blanket AI block often removes you from ChatGPT search without anyone noticing.

A site that wants search visibility but no training use might look like this:

User-agent: OAI-SearchBot
Allow: /

User-agent: PerplexityBot
Allow: /

User-agent: Claude-SearchBot
Allow: /

User-agent: GPTBot
Disallow: /

Two points on Google. There is no robots.txt token for AI Overviews or AI Mode, so Googlebot rules are the control there. Google-Extended covers Gemini app training and grounding, not Search inclusion. Meta states that allowing Meta-WebIndexer "helps us cite and link to your content in Meta AI's responses."

Test the firewall, not just the file

OpenAI's help page says eligibility needs both: allow OAI-SearchBot, and confirm your host or CDN allows OpenAI's published searchbot IP addresses. A bot-management rule that returns 403 removes you silently.

This matters more since September 15, 2026. Cloudflare then set Training and Agent crawlers to blocked by default on ad-displaying pages for new domains, new sites on existing accounts and existing free-tier customers. Search crawlers stayed allowed by default. Open your CDN's bot settings and look.

The fastest check is to request a page as each crawler. The AI Crawler Checker does that and shows which bots your firewall turns away.

One limit: user-triggered fetchers such as Perplexity-User generally ignore robots.txt, by Perplexity's own documentation. Treat robots.txt as necessary, not sufficient.

Day 3: server-render the facts that matter

This is the strongest finding in the evidence. Vercel found that ChatGPT and Claude crawlers fetch JavaScript files (11.50% and 23.84% of their requests) but don't execute them. Independent testing in June 2026 found the same for GPTBot, OAI-SearchBot, ClaudeBot, PerplexityBot and others. Googlebot and AppleBot are the exceptions, because they render.

If your pricing table, product names or service list appear only after JavaScript runs, those crawlers see an empty page.

How to check in ten minutes:

Open a key page and view the raw source, not the inspected DOM.
Search the source for your price, your product names and your one-line description.
Repeat on your pricing, about and main service pages.
Note anything inside tabs, accordions or "load more" blocks that loads by fetch.

Anything missing from raw HTML needs server-side rendering or prerendering. The tradeoffs are covered in client-side vs server-side rendering for AI crawlers.

While you're there, make sure specific facts are stated plainly: a price, a number, a one-line definition. That gives an engine a sentence it can quote. It doesn't make the engine choose you.

Day 4: check index status, engine by engine

Google

Google's AI features documentation says a page must be "indexed and eligible to be shown in Google Search with a snippet" to appear as a supporting link in AI Overviews or AI Mode, and that "There are no additional technical requirements." Eligibility is not inclusion, though.

Check these:

Search Console index coverage for your key pages.
Stray noindex, nosnippet or max-snippet:0 tags, often inherited from a template.
data-nosnippet attributes that hide sections from snippets and AI summaries.
The Search Console generative AI control, if your property shows it. Inclusion is the default, but child properties inherit from the parent, so a site can be excluded without anyone choosing it.

Bing

Verify the site in Bing Webmaster Tools. Bing's index backs Microsoft Copilot and Bing's AI summaries.

Why both matter: in Otterly's 2026 experiment, Google deindexed two AI-generated sites (after about 14 and 90 days) and Google's AI surfaces almost stopped citing them. Bing kept them, and Copilot supplied 2,718 of the 3,934 citations Otterly tracked for one site. An engine cites what its index holds. For Google specifics, see how to get cited in AI Overviews and AI Mode.

Day 5: see what the engines actually say

Now look at the output. Pick five to ten questions a buyer would ask before choosing a business like yours. Ask each one on ChatGPT, Perplexity, Gemini and Google, logged out with memory off.

Record four things per answer:

Were you named at all?
Were the facts right (price, location, what you sell)?
Which sources did the engine cite?
Did the answer show citations, which suggests it searched the web?

Wrong facts often trace back to your own site stating the same thing two ways. Run a check for contradicting facts on your own site before blaming the model.

When engines cite review sites, directories or industry roundups instead of you, being named on those pages is the route into the answer. That's an association in the evidence, not a promised effect.

If you'd rather not run this by hand, the free checker on this site tests up to three buyer questions and scores the answers. It is a snapshot, not tracking. The manual version is laid out in this do-it-yourself AI visibility audit.

Common mistakes in the first week

Allowing GPTBot to get into ChatGPT. GPTBot is for training. OAI-SearchBot is the search bot.
Adding schema or FAQ schema to get cited. Google says structured data isn't required for generative AI search. Ahrefs' matched test of 1,885 pages found AI Overviews citations fell 4.6% after schema was added. Existing markup does no harm, so there's no need to remove it. It just isn't a lever.
Publishing an llms.txt. Google says Search ignores it, and no other engine documents reading it.
Rewriting pages in an AI-friendly style or chunking them. Google's optimization guide rejects both and says there's no ideal page length.
Mass-publishing AI content. Google's spam policy calls generating many pages without adding value scaled content abuse. Otterly's sites were removed with no manual action notice.
Hidden text or prompt injection aimed at AI crawlers. These tactics exist and can sway some models in tests. They are illegitimate, they risk penalties, and they can be turned against your brand by others.
Expecting a chatbot to help. An on-site chatbot doesn't change what crawlers fetch. More on whether a website chatbot helps AI search visibility.
Chasing an AI rank. Google's AI report shows impressions only, with no position data to check a rank against.

After week one

Week one fixes the gates. After that, the work shifts to things that move slowly: keeping facts consistent, getting named on sources engines already cite, and re-checking access.

Re-run the crawler and robots.txt checks monthly. CDN defaults change without you editing anything, as Cloudflare's September 2026 change showed. Re-run your buyer questions monthly too, and compare each engine separately rather than blending them into one number.

For a longer schedule, use a 30/60/90 day AI visibility plan. If your domain is new, read whether a brand-new website can get cited to set expectations.

Frequently asked questions

How long does it take for fixes to show up in AI answers?

Nobody publishes a timeline. Access fixes take effect when each crawler next visits and the engine refreshes its index, which varies by engine and page. Answers given without web search depend on the training cutoff, so they won't reflect your changes until a new model ships. Re-check monthly and compare engines separately instead of expecting one date.

Should I block GPTBot if I don't want my content used for training?

You can. OpenAI documents that disallowing GPTBot signals your content shouldn't be used for training, and that this setting is independent of OAI-SearchBot. Blocking GPTBot does not remove you from ChatGPT search. Keep OAI-SearchBot allowed if you want to stay eligible for ChatGPT search answers.

Does my site need to rank in Google's top 10 to appear in AI Overviews?

No. Google requires a page to be indexed and eligible for a snippet, with no additional technical requirements. Ranking data and AI Overview citations overlap only partly, and Google states that indexing and serving aren't guaranteed. Classic SEO still helps you get indexed, but a top-10 position is not a gate.

Does this plan work the same for Google as for ChatGPT?

Mostly, with one big difference. Googlebot renders JavaScript, so client-side content is less of a problem on Google. ChatGPT, Claude and Perplexity crawlers don't execute JavaScript, so server rendering matters most there. Robots.txt rules, firewall reachability and index checks apply to every engine, but each engine uses its own bot and index.

Can I guarantee my business gets mentioned by AI?

No. You can remove the blocks that make you ineligible: crawler access, rendering, indexing and accidental snippet controls. Which pages an engine retrieves for a question, and which it cites, are decided by the engine. Anyone promising a citation, a rank or a score is promising something no one outside the engine controls.

Sources

This page is AI-assisted. It was written from a verified evidence base and checked against our editorial policy. Engine behaviour changes often; facts are as of October 21, 2026.