A modern SEO audit keeps the classics — crawlability, indexation, speed, canonicals, links — and adds the AI-consumption layer: whether AI crawlers can reach you, whether answers extract cleanly from your HTML, whether your schema is one consistent graph, and whether you are actually being cited. Audit the machine experience, not just the crawler's.
What still matters from the classic audit?
All of it. Broken links, redirect chains, orphan pages, duplicate titles, missing canonicals, slow templates, thin pages — the fundamentals did not stop mattering because the interface changed. AI engines build on the same index, so a site that fails the classics fails the new layer by default. The mistake is treating the classic audit as sufficient, not treating it as obsolete.
Which AI-era checks should you add?
- AI crawler access — GPTBot, ClaudeBot, PerplexityBot and Google-Extended explicitly allowed, and served the same content as users.
- Render independence — is your key content in the initial HTML? Fetch your page as a crawler and read what comes back.
- Answer extraction — does each important section open with a short, self-contained answer? If a machine lifted one paragraph, would it be the right one?
- Schema graph integrity — one JSON-LD graph per page, no dangling references, named human author, real dates.
- Machine surfaces — sitemap parity, clean robots.txt, and your metadata files fresh rather than hand-written and stale.
- Citation tracking — a fixed monthly prompt panel recording who the engines cite for your target queries.
How often should you re-audit?
Continuously, in the build, for everything mechanical. Broken links, schema validity, title lengths, canonicals and sitemap parity should fail a deploy, not wait for a quarterly review. Reserve human audits for judgement calls: content quality, positioning, and whether your pages still answer what buyers ask. Machines check the machine-readable; people check the meaningful.
Our free GEO audit runs the six AI-era checks on any public URL in seconds — use it as the new layer on top of your existing technical audit, then fix what it flags in priority order.
What should you fix first after the audit?
Fix what blocks machines before what persuades them: crawler access and render independence first, then schema integrity, then answer structure, then measurement. Most teams do it backwards — polishing copy on pages engines cannot properly fetch. Sequence beats effort every time.
Six dimensions, instant score, no email required.