The mistake is treating AI visibility as a new surface while the site still has old search hygiene problems. Useful pages can be blocked by robots rules, absent from sitemaps, split across duplicate canonicals, hidden behind JavaScript links, marked up with schema that does not match the page, or too vague about the entity behind the content. Then nobody can say which pages matter, what can be quoted, or whether attention turns into a qualified action.
Technical SEO Audit Example: Evidence & Deliverables
A representative technical SEO audit showing what a buyer should be able to inspect: URL-level crawl and index evidence, canonical and redirect decisions, schema and entity checks, internal-link findings, snippet controls, and a prioritized implementation handoff.
inspectable artifacts
finding evidence
performance claims
This is a representative example, not a named-client result. Use it to evaluate whether an audit gives you evidence you can act on. The standard is URL-level findings, source evidence, an owner and fix, an acceptance check, and a re-test plan—not a PDF of generic warnings.
What this example covers.
- Robots, sitemap, status, canonical, render, and indexability evidence for priority URLs
- Crawlable internal links, visible-content schema, entity references, and snippet controls
- Prioritized implementation handoff with owner, acceptance check, re-test status, and Search Console / analytics measurement readiness
I would audit the site as an evidence trail: robots.txt, sitemap coverage, canonical targets, status codes, URL Inspection signals, crawlable HTML links, structured data, organization/entity markup, and source-page boundaries. Each material finding should leave an implementation handoff: affected URL, evidence, severity, owner, recommended fix, acceptance check, and re-test status. Preview controls such as noindex, nosnippet, data-nosnippet, max-snippet, and bot-specific access are documented where relevant. llms.txt is treated as a curated directory, not a ranking lever. Measurement connects Search Console, GA4, referral tagging, and lead-path events so later performance claims can be checked against first-party data.
The useful outcome is not "more AI citations." It is a site with fewer interpretive gaps: priority URLs can be found, canonical pages are clear, markup describes visible content, source pages are documented, snippet controls are intentional, and the analytics path is ready for review. The claim is improved eligibility, governance, and measurement readiness. Rankings, citations, traffic, revenue, and conversion gains still require separate evidence.
Inspectable work products.
- Prioritized audit register linking each material finding to URL-level evidence, owner, recommended fix, acceptance check, and re-test status
- Priority URL evidence set using URL Inspection, Page Indexing, Crawl Stats, and representative page checks
- Schema and entity manifest with validation status and visible-content matching notes
- Internal link map covering crawlable HTML links, anchor text, hubs, detail pages, and orphan-risk pages
- Snippet and AI access controls matrix for noindex, nosnippet, data-nosnippet, max-snippet, and bot rules
- Measurement workspace tying Search Console, GA4, referral surfaces, and lead-path key events to reviewable evidence
An inspectable deliverable shape.
Sanitized audit register. Statuses are illustrative and are not presented as a client crawl.
| Audit item | Evidence | Decision | Acceptance check |
|---|---|---|---|
| Priority service URL | 200 · indexable · self-canonical · in sitemap | Retain; strengthen discovery and proof | Re-run crawl + URL Inspection after release |
| Legacy URL | 301 to priority URL; stale internal links found | Normalize links to destination | Crawl shows no internal links to legacy URL |
| Preview URL | 200 · noindex · self-canonical · absent from sitemap | Keep excluded from index | Verify noindex + sitemap exclusion before launch |
What the research supports.
AI search readiness starts with crawl reality
I look for whether the important pages can be discovered, rendered, canonicalized, and inspected before treating answer engines as a separate strategy.
The test is boring on purpose
Can a crawler find the priority page through normal HTML links, understand what it is about, and see the same proof a human visitor sees?
Schema has to match the page
Structured data is useful when it describes visible content and entity relationships. It becomes liability when it overstates what the page actually proves.
Access controls do different jobs
Robots.txt, noindex, nosnippet, data-nosnippet, max-snippet, and bot-specific rules are not interchangeable. The useful artifact is a clear control map.
Measurement keeps the claim honest
Search Console, GA4, referral tagging, and lead-path events do not prove lift by themselves, but they create the trail needed before anyone makes a performance claim.
- Do not claim guaranteed rankings, rich results, AI Overviews, ChatGPT citations, or answer-engine inclusion.
- Do not claim robots.txt deindexes content; it mainly governs crawl access, while indexing and snippet controls use other mechanisms.
- Do not claim traffic, leads, conversion, revenue, or market-share impact without first-party analytics and a defined attribution window.
- Do not present llms.txt, schema, or bot allowlists as ranking signals or universal measurement for AI visibility.
Supports standard Search eligibility, indexability, snippet eligibility, and preview controls for AI features.
Google Search CentralBuild and submit a sitemapSupports sitemap best practices, canonical URL inclusion, and submission workflows.
Google Search CentralLink best practicesSupports crawlable HTML links and descriptive anchor text.
Google Search CentralStructured data policiesSupports visible-content matching and warns that valid markup does not guarantee rich-result display.
OpenAI Help CenterChatGPT SearchDocuments ChatGPT search eligibility through OAI-SearchBot and published IP access, possible citations, and no guaranteed placement.
OpenAI DevelopersOpenAI crawlersSupports separating search access and training access through bot-specific robots rules.
Anthropic Help CenterAnthropic crawler controlsSupports separate bot access preferences for model development, search, and user-directed retrieval.
llmstxt.orgllms.txt proposalSupports treating llms.txt as a proposal for inference-time guidance, not a replacement for robots or sitemaps.
Need the AI, website, or visibility map for your own site?