Skip to content

Foundations

Four chapters on the mechanics underneath every tactic in this book: how a search engine actually turns a URL into a ranked result, how an LLM actually turns a question into a citation, how machines decide who you are, and how you know whether any of it worked. Read these once and the rest of the book stops being a list of tips and becomes a set of moves you can reason about.

How discovery works in 2026 is the map. This part is the terrain.

Why mechanics before tactics

Almost every expensive discoverability failure we've shipped through was a mechanics failure wearing a tactics costume:

  • A local business had immaculate on-page SEO, unique titles, real reviews, and clean structured data — and zero pages indexed, because a security product returned a bot challenge to Googlebot. No tactic fixes that. Understanding the crawl-permission gate does. → Search engines
  • A SaaS site was correctly indexed by Google and still never appeared in ChatGPT answers, because ChatGPT's search leans on Bing and the property's Bing presence was unmanaged. Knowing which index feeds which assistant is worth more than a month of content. → AI retrieval
  • A site emitted two conflicting Organization blocks per page and a provider reference pointing at an entity node that didn't exist. Both "had schema". Neither produced a single confident entity. → Entities and trust
  • A team shipped fifteen pages in a day and could not say whether anything improved, because nobody had written down the numbers from the day before. → Measurement

Tactics are portable; mechanics are what tells you which tactic applies. When a tactic stops working — and they rotate constantly — the mechanics are what's left.

The four chapters

  • :material-spider-web: Search engines — crawl, index, rank The pipeline in four stages, and what gates each one. Why crawl permission, not crawl budget, is the real problem for sites under a few thousand pages. How canonicalization actually picks a winner. What JavaScript rendering does to the schedule. Where classic ranking factors stand as of 2026.

  • :material-robot-outline: How AI finds and cites The three routes by which an assistant knows you exist, which index each major assistant leans on, what a "citation" is mechanically, and the distinction that decides everything: being indexed is not the same as being retrievable at answer time. Includes the evidence base, each claim dated and labelled.

  • :material-account-check: Entities, E-E-A-T, and trust Being a resolvable entity rather than a collection of pages. One identity node, cross-linked by @id. sameAs graphs as identity consolidation. The NAP-consistency evidence. Why E-E-A-T is an impression you earn, not a setting you toggle — and why fabricated content is a discoverability risk, not just an ethics problem.

  • :material-chart-line: Measurement and baselines Record the numbers before you touch anything. The real customdomain.ai baseline (~5 clicks, ~159 impressions, average position ~42) and how it became the yardstick. AI-visibility spot checks. Estimating any site's traffic from open sources with confidence tiers. Fingerprinting a competitor's stack in ten minutes.

How these four interlock

flowchart TD
    P["Crawl permission<br/>(robots, WAF, status codes, TLS)"] --> I[Indexed]
    I --> R["Ranked / retrievable"]
    R --> C["Cited by an answer engine"]
    E["Entity clarity<br/>(one identity, corroborated)"] --> R
    E --> C
    M["Measurement"] -.-> P
    M -.-> I
    M -.-> C

Permission gates indexing. Indexing gates retrieval. Entity clarity decides whether a machine that retrieves you can say who you are with confidence. Measurement is the loop that tells you which link in that chain is broken — and it is the only one of the four that most operators skip.

The reading order that pays

  1. Search engines first, always. Everything downstream inherits its gates.
  2. AI retrieval next — it reuses the crawl model and adds one hard new idea (answer-time retrievability).
  3. Entities and trust once you understand what a machine is trying to resolve you into.
  4. Measurement before you change anything on a live property. Baselines you didn't record are gone forever.

If you only have twenty minutes: read the crawl-permission section of Search engines, then the "indexed is not retrievable" section of AI retrieval, then run the 30-minute AI visibility audit.