git:20260720.41d8423 to git:20260728.fac4544

10 added, 11 removed. Audit A to A.

---

name: technical-seo

description: Make sites crawlable, fast, and structured so search finds what deserves finding, without tricks that expire. Use when improving organic traffic or auditing a site's search health.

---

# Technical SEO

Technical SEO removes the obstacles between good content and the

index; it cannot rank content that deserves obscurity. Work in

order: crawlable, then fast, then structured, then measured: and

build nothing that depends on a loophole.

## Method

1. **Verify crawl and index basics first.** Search Console

- as the ground truth: coverage report for what is

- indexed vs excluded and why; robots.txt not blocking

- what should rank; XML sitemap generated from the real

- page set (see docs-as-code: generate, never hand-

- maintain), canonical tags on every page (self-

- referencing by default) so URL variants (params, trailing

- slashes, http/https) consolidate instead of competing;

- 404s for gone content, 301s for moved (chains

- collapsed), and pagination/faceted URLs deliberately

- managed (noindex or canonical) so crawl budget is not

- spent on infinite filter permutations.

+ as the ground truth: coverage report for what is indexed

+ vs excluded and why; robots.txt not blocking what should

+ rank; XML sitemap generated from the real page set (see

+ docs-as-code: generate, never hand-maintain), canonical

+ tags on every page (self-referencing by default) so URL

+ variants (params, trailing slashes, http/https)

+ consolidate instead of competing; 404s for gone content,

+ 301s for moved (chains collapsed), and pagination/faceted

+ URLs deliberately managed (noindex or canonical) so crawl

+ budget is not spent on infinite filter permutations.

2. **Make rendering crawler-friendly.** Server-render or

pre-render the content that must rank (client-only

rendering delays and sometimes loses indexing: see

frontend-build-tooling's SSR options); meaningful

content in the HTML, not behind interactions; one H1,

logical heading tree (see docs-information-architecture,

accessibility's shared interest: the crawler is a

screen reader with a budget).

3. **Hit the speed bars for real users.** Core Web Vitals

(LCP, INP, CLS) green on field data, not just lab (see

web-vitals for the how: image optimization, font

loading, layout stability: see image-optimization,

font-loading); mobile-first because indexing is (see

responsive-design). Speed is a ranking factor, a

conversion factor, and the same work (see

landing-page-strategy step 5).

4. **Mark up what the page actually is.** Structured data

(JSON-LD: Article, Product, FAQ, HowTo, Organization)

where the content genuinely matches the type: rich

results and entity clarity, validated in the testing

tools; honest markup only (marking up fake reviews is

a manual-action magnet). Titles and meta descriptions

as the search snippet's copy: outcome-first, unique

per page, written like the tiny ads they are (see

landing-page-strategy's headline discipline).

5. **Architect internal links as an importance map.**

The pages you most want ranked get the most internal

links, from relevant anchors (see

docs-information-architecture); orphan pages

(reachable only by sitemap) rank like orphans;

hub-and-spoke topic clusters (pillar page linking

detailed children) match how engines model topical

authority: which is content strategy wearing

information architecture's clothes (see

developer-marketing for what earns external links:

the part no tag can fake).

6. **Measure by query cohort, iterate quarterly.**

Search Console queries by page: impressions,

position, click-through (title/description tests move

CTR at constant position); track the money queries'

trajectory, not vanity aggregates (see

product-metrics' vanity warning); after site changes

(migrations especially: see cloud-migration's

rehearsal ethic applied to URL moves: map every URL,

301 everything, monitor coverage daily for a month),

verify nothing deserving fell out.

## Boundaries

- Tricks (bought links, doorway pages, scraped content,

keyword stuffing) work until the next core update and

then subtract; every hour on loopholes is an hour not

spent on content that earns links (see

developer-marketing's trust economics).

- SEO latency is quarters, not sprints; commitments to

leadership should promise process and leading

indicators, not ranking dates (see

status-updates' honesty about uncertainty).

- Content quality is the actual ranking engine; technical

SEO is table stakes plumbing. When plumbing is green

and rankings are not, the answer lives in the writing

(see technical-writing, tutorial-writing).