SEO — Discoverability as a Pipeline, Not a Ranking Theory

• · P1 — You control access and comprehension. You do not control ranking. Directives (`robots.txt`, `noindex`, canonical hints, sitemaps, structured data) are how a site states what it wants understood. Where it then appears is not yours to set. Every rule here is about the first thing; none is about the second.
• · P2 — Ranking-factor speculation is Level 4 and stays out of the pack entirely. Not demoted, not caveated — excluded. "Keyword density", "optimal title length for ranking", "the algorithm rewards X" are blog claims about an unpublished system. Per [source-policy](../../../core/source-policy.md), a Level 4 idea may never be a pack's basis, and an engineering decision made on one is unfalsifiable by construction.
• · P3 — When SEO advice conflicts with accessibility, performance, or honesty toward the user, those win. Every technique that once traded a real user's experience for a crawler's attention has since been penalized. The interests are aligned far more often than folklore suggests; where they appear to diverge, the divergence is usually a sign the advice was folklore.
• · P4 — Indexing is a pipeline — discover, crawl, render, index, serve — and every problem is one stage failing. Naming the failing stage *is* the diagnosis; skipping that step is how teams end up "doing SEO" by adding tags to a page a crawler never fetched.
• · P5 — If it is not in the HTML the crawler ends up with, it does not exist. Content that requires interaction, arrives after an event, or depends on client-side data fetching may never be seen. Rendering strategy is an indexing decision before it is a performance one.

• · Ask which pipeline stage is failing before doing anything. Discover, crawl, render, index, serve. Most "SEO work" is tags added to a page a crawler never fetched.
• · Fetch the page the way a crawler would and read what comes back. Not view-source in a browser that already ran the JavaScript. This single habit catches most render-stage problems.
• · If a stranger should be able to land on it from a search result, put the content in the response. Everything else is a performance decision.
• · Pick one URL form on day one and redirect the rest. Protocol, host, trailing slash, case. Costs minutes at the start and a migration later.
• · Never combine robots.txt disallow with an expectation of de-indexing.

• · Ver anti-patrones específicos del dominio en el pack.

• [ ] Structured data describes only what the page visibly shows. (SEO27)
• [ ] No technique here costs accessibility, honesty to the user, or performance — no hidden text, cloaking, doorway pages, or keyword-stuffed alt attributes. (SEO44)
• [ ] No URL is both disallowed in robots.txt and expected to be de-indexed. (SEO2)
• [ ] Render-critical CSS, JS, and images are fetchable. (SEO3)
• [ ] Non-production environments are kept out of search by auth or noindex, not.
• [ ] Primary content and links exist in the initially served HTML. (SEO9)

• Google Search Essentials
• Google Search Central Documentation
• RFC 9309: Robots Exclusion Protocol
• Schema.org
• Sitemaps XML format

Este pack cita las siguientes fuentes registradas en el índice bibliográfico:

• webGoogle Search Central DocumentationGoogle
• webGoogle Search EssentialsGoogle
• webRFC 9309: Robots Exclusion ProtocolIETF
• webSchema.orgSchema.org
• webSitemaps XML formatsitemaps.org