Skip to content
ODOOVATIONShome

Checklist

AI Search Optimization Checklist

9 min read·Published by ODOOVATIONS, October 2026

Start with access: the pages that matter must be crawlable, indexed and readable as text. Then make each important page state its answer plainly, show evidence and authorship, and keep structured data consistent with what is visible. Make your name, address and services identical across your site and verified profiles, earn genuine reviews and mentions, review crawler settings against each provider’s documentation, and record what AI search says before and after changes. Nothing on this list is a guarantee, and none of it needs special files.

Technical: can it be found and read?

  • Every page that should be found returns a 200 status, is not blocked by robots.txt or a noindex tag, and has one canonical URL.
  • Retired pages redirect permanently to their closest replacement, and there are no redirect chains.
  • Important pages are linked from other pages on the site with descriptive link text.
  • A sitemap lists the canonical URLs, and its last-modified dates change only when the content does.
  • The content that answers questions is present in the HTML a crawler receives, not only after scripts run, and not only in images.
  • Pages load quickly on a phone and the layout does not jump.
  • Structured data (for example Organization, Service, Article, FAQ) is accurate, matches the visible text and uses the same name and URLs everywhere.
  • The site is verified in Google Search Console and Bing Webmaster Tools, and the sitemap is submitted to both.
  • Your CDN, firewall and bot protection do not block search crawlers by accident. Check the logs, not just the settings.

Crawler settings: search and training are separate decisions

ProviderCrawlerWhat the provider says it is for
OpenAIOAI-SearchBotSurfacing sites in ChatGPT search. A site that opts out will not be shown in ChatGPT search answers.
OpenAIGPTBotContent that may be used to train generative AI models. Blocking it does not by itself remove a site from search.
OpenAIChatGPT-UserActions a user asks ChatGPT to take. OpenAI says robots.txt rules may not apply.
AnthropicClaude-SearchBotImproving the quality of search results.
AnthropicClaude-UserFetching pages when a user asks a question that needs the web.
AnthropicClaudeBotCollecting content that may be used for model training.
PerplexityPerplexityBotSurfacing and linking sites in Perplexity search. Perplexity says it is not used for training foundation models.
PerplexityPerplexity-UserUser-initiated fetches. Perplexity says it generally ignores robots.txt because a user requested the fetch.
GoogleGooglebotSearch, including AI Overviews and AI Mode. Limit what is shown with nosnippet, max-snippet or noindex.
GoogleGoogle-ExtendedLimits use of content for AI training and grounding in some other Google systems.

This table summarises each provider’s documentation as read on 30 October 2026. Providers change their crawlers and wording, so check the sources below before changing robots.txt, and keep the choices you have already made about training.

Content: does each page answer?

  • Each service, product and guide page opens with a direct statement of the answer, then the detail.
  • Claims are backed by something a reader can check: a source, a method, a real example. Nothing is invented.
  • Real authorship is shown, and the update date is honest.
  • Comparisons are fair and say when the alternative is the better choice.
  • FAQs are questions buyers actually ask, answered honestly, and visible on the page.
  • Prices, terms and limits are stated plainly, in text.

Brand and entity: does everything agree?

  • One fact sheet: legal name, trading name, address, phone, email, services, what you do not do.
  • Your site, Google Business Profile, LinkedIn page and directory listings all match it.
  • Similar names are disambiguated with a clear description and your own domain.
  • You ask real customers for reviews and never write or buy them.

Measurement: do you know what changed?

  • A written list of buyer questions and the search experiences that matter.
  • A dated baseline recorded before changes are made.
  • The same questions repeated on a schedule, with every answer saved.
  • Mentions, citations, referral traffic and enquiries reported separately.

Not required, and not to be done

  • A special file, special markup or an “AI schema”. Google says none is needed to appear in its AI features. A plain-text site summary is optional.
  • Hidden text or hidden instructions for machines, and text shown only to crawlers.
  • Keyword stuffing, mass-produced near-identical pages, purchased links and invented reviews.
  • Any promise that an assistant will mention you.

Questions

With access: confirm in Search Console that your key pages are indexed, then read what the assistants say about you today. Everything else is prioritised from those two facts.

Want this done for your business?

Ask for a free suitability review, or start with a paid audit. Asking is free and commits you to nothing.