Skip to main content
Checklist

The AI SEO Checklist

Fifty-plus checks in the order worth doing them, including AI crawler access and citation readiness. Crawlability before content, content before schema, schema before links — because each depends on the last.

Published 24 September 2026 9 min read

Order matters more than completeness. Working through section one properly beats doing a scattered handful of items from every section. Pages that cannot be crawled do not benefit from anything else on this list.

1. Crawlability & indexing

Nothing else counts until these pass.

  • Every page you want indexed returns HTTP 200
  • robots.txt exists and does not block important paths
  • No accidental noindex tags on pages you want ranked
  • An XML sitemap exists at /sitemap.xml
  • The sitemap lists only URLs that genuinely exist — no 404s, no redirects
  • robots.txt references the sitemap location
  • Site verified in Google Search Console, sitemap submitted
  • Bing Webmaster Tools set up as well — its index feeds several AI products
  • AI crawlers not blocked in robots.txt: GPTBot, OAI-SearchBot, PerplexityBot, ClaudeBot
  • Google-Extended not disallowed — it governs Gemini and AI Overviews separately from Googlebot
  • No duplicate User-agent: * groups in robots.txt — ambiguous across crawlers
  • Every page has one self-referencing canonical tag
  • HTTPS works, with HTTP redirecting to it
  • One hostname only: either www or bare domain, redirecting the other
  • No orphan pages — everything reachable through internal links

2. Performance & mobile

Fix real problems; do not chase a perfect score.

  • Largest Contentful Paint under roughly 2.5 seconds
  • Cumulative Layout Shift under 0.1 — no content jumping during load
  • Images compressed and served in a modern format
  • width and height set on images to reserve layout space
  • Below-the-fold images lazy-loaded
  • Layout usable on a narrow phone viewport without horizontal scrolling
  • Tap targets large enough to hit comfortably
  • Text readable without zooming

3. On-page fundamentals

Per page, and unglamorous but consistently effective.

  • One <h1> per page describing what the page is about
  • Title tag under about 60 characters, target phrase near the front
  • Meta description around 150–160 characters, written to earn the click
  • URL short, lowercase, hyphenated, no dates or ID strings
  • Headings nest properly: h2 under h1, h3 under h2, none skipped
  • Descriptive alt text on meaningful images, empty alt on decorative ones
  • Internal links using descriptive anchor text, not "click here"
  • No broken internal or external links

4. Content structure for AI

This section is where AI SEO diverges from the traditional checklist. AI systems need to locate and lift a specific answer, not just judge the page overall.

  • Each section answers one clear question
  • The question appears as a heading, with the answer immediately below it
  • Answers lead with the conclusion, then add supporting detail
  • Paragraphs kept short, two or three sentences
  • Lists and tables used where content is genuinely list-shaped
  • Facts stated precisely and self-contained, not dependent on earlier context
  • An FAQ section using questions people actually ask
  • Claims attributed, with dates on anything time-sensitive
  • No important content locked behind tabs or accordions that hide it from crawlers
  • Statistics carry a source and a date — AI systems favour verifiable, attributable claims
  • Entities named explicitly rather than referred to as "it" or "the company"
  • Comparison content uses tables, which extract more cleanly than prose

Optional: llms.txt

A proposed convention for offering language models a clean markdown summary of your site. Adoption is early and it is not an established standard, so treat it as cheap experimentation rather than a requirement. Crawler access and page structure matter far more.

A practical test: take any section of your page in isolation. Does it still answer its question without the rest of the page around it? If not, it is unlikely to be extracted cleanly.

5. Structured data

Helpful, not mandatory — and invalid markup is worse than none.

  • Article schema on blog posts and guides
  • FAQPage schema where you have a genuine question section
  • BreadcrumbList schema matching your visible breadcrumbs
  • Organization schema on the homepage, with logo and URL
  • Course schema if you sell courses
  • Every block validated with the Rich Results Test
  • Schema reflects what is actually visible on the page, not extra claims

6. Authority & links

The slowest section, and the one that decides competitive outcomes.

  • Profile links in place: LinkedIn, X, YouTube, relevant directories
  • Author or organisation clearly identified, with credentials where relevant
  • Contact information genuinely reachable
  • At least one page worth linking to on its own merit
  • No purchased links, PBNs, or mass directory submissions

7. Measurement

Without this you are guessing.

  • Search Console indexing report reviewed for errors
  • Queries report checked for terms you appear for unexpectedly
  • Pages on results page two or three identified — your nearest wins
  • Pages with impressions but no clicks flagged for title and description rewrites

AI visibility, which traditional metrics miss

Rank tracking alone no longer tells you enough. With AI Overviews correlating to a 58% lower click-through rate for the top result (Ahrefs), you can hold position one and lose most of the traffic. Worth tracking separately:

  • Whether an AI Overview appears for your target queries at all
  • Whether you are cited in it, which is independent of your ranking position
  • Referral traffic from chatgpt.com, perplexity.ai, and gemini.google.com in your analytics
  • Brand mentions without links — these still feed entity signals
  • Whether asking the major assistants about your topic surfaces you as a source

Manually querying the assistants for your key topics is crude but genuinely informative, and costs nothing. Dedicated AI visibility tools exist, but start with the manual check before paying for one.

For the reasoning behind section four, what AI SEO actually means covers the extraction model. If you are starting from zero, the 90-day beginner plan sequences this checklist across a realistic timeline.

Frequently asked questions

What order should I work through this in?

Top to bottom. Crawlability and indexing first, since nothing else matters if pages cannot be found. Then on-page structure, then schema, then links. Skipping ahead means optimizing pages search engines may not reach.

Do I need paid tools for this?

No. Google Search Console, PageSpeed Insights, and the Rich Results Test are free and cover nearly every item here. Paid crawlers save time on large sites but are not required.

How often should I run through it?

Fully once at setup, then the technical sections quarterly. Sections three through five apply to each new page you publish rather than on a schedule.

Is structured data required to rank?

Not required, but it helps. Schema makes content easier to parse accurately, improving your odds of being cited correctly. Always validate it — broken markup is worse than none.

Which AI crawlers should I allow in robots.txt?

If you want AI visibility: GPTBot and OAI-SearchBot (OpenAI), Google-Extended (Gemini and AI Overviews), PerplexityBot, and ClaudeBot. Google-Extended is the one most often missed, because it is separate from regular Googlebot, so a site can rank in Google while being excluded from AI Overviews.

How do I know if I am cited in AI answers?

Start by asking the major assistants about your topics and seeing whether you are named. Then check your analytics for referral traffic from chatgpt.com, perplexity.ai and gemini.google.com. Dedicated AI visibility tools exist, but the manual check costs nothing and tells you most of what you need early on.

Go deeper than a checklist

The course covers why each of these items matters, so you can make judgement calls when the list does not fit your situation.

See the curriculum