Work sample · Representative method

Technical Search + Entity Foundation / 2026

Technical Search & AI-Search Visibility Layer

A representative visibility layer for making priority pages easier to crawl, understand, cite responsibly, and measure without pretending AI search is controllable.

Audited

crawl access

Defined

citation rules

Traceable

search measurement

Representative framing

This is a representative field note, not a named-client result. I look for the boring things that decide whether a search system can trust the site: crawl access, canonical discovery, HTML links, visible-content schema, entity consistency, snippet controls, and measurement that survives beyond a dashboard screenshot.

Problem

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.

Approach

I would audit the site like 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. Then I would document preview controls such as noindex, nosnippet, data-nosnippet, max-snippet, and bot-specific access 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 future claims can be checked against first-party data.

Outcome

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.

Implementation proof points

Implementation with evidence.

  • Crawl governance register for robots.txt, sitemaps, canonicals, status codes, blocked URLs, and recrawl notes
  • 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
Research-backed takeaways

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.

Claim boundaries
  • 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.
Related definitions
Selected research sources
Google Search CentralAI features and your website

Supports standard Search eligibility, indexability, snippet eligibility, and preview controls for AI features.

Google Search CentralBuild and submit a sitemap

Supports sitemap best practices, canonical URL inclusion, and submission workflows.

Google Search CentralLink best practices

Supports crawlable HTML links and descriptive anchor text.

Google Search CentralStructured data policies

Supports visible-content matching and warns that valid markup does not guarantee rich-result display.

OpenAI Help CenterPublishers and developers FAQ

Supports public accessibility, OAI-SearchBot access, and ChatGPT referral tracking boundaries.

OpenAI DevelopersOpenAI crawlers

Supports separating search access and training access through bot-specific robots rules.

Anthropic Help CenterAnthropic crawler controls

Supports separate bot access preferences for model development, search, and user-directed retrieval.

llmstxt.orgllms.txt proposal

Supports 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?