Technical SEO
Date: 2026-08-16
Making a site crawlable, indexable and unambiguous. It’s architecture work — URLs, rendering, redirects, status codes — and it’s separable from the content and keyword half, which is where most SEO advice lives and most durable value doesn’t.
What it is
Technical SEO is the set of implementation decisions determining whether a search engine can find your pages, render them, understand which version is canonical, and index them efficiently.
The split worth holding:
| Technical SEO — Architecture | Content SEO — Acquisition |
|---|---|
| Can a crawler reach it | Is it worth ranking |
| Can it render the content | Does it match intent |
| Which URL is authoritative | What is it about |
| How is crawl budget spent | Who links to it |
This note is the first half. The second is Search Engine Optimisation in Commerce & Growth.
The durable core
What’s stable regardless of algorithm updates, roughly in the order it matters:
- Discoverability. Real
<a href>links, an XML sitemap, no orphaned sections — Crawling and Indexing - Renderability. Content in the HTML, not behind JavaScript. This got more important, not less — Rendering and SEO
- Unambiguity. One canonical URL per piece of content, consistently signalled — Canonicalisation
- Correct status codes. 200 for real pages, 404 for gone, 301 for moved, and never a 200 on an error page
- Efficient crawling. Don’t spend crawl budget on infinite filter combinations — Faceted Navigation and Crawl Budget
- Stable URLs. Changing them is expensive; changing them without redirects is destructive — URL Structure, Redirects and Link Equity
- Machine-readable meaning where it earns a richer result — Structured Data
Everything above is an engineering property. None of it is a tactic, and none of it changes when an algorithm does.
The version-dependent parts
Deliberately not in this note, because they expire:
- Which rich result types are eligible and what they look like
- Ranking factor weightings, and anything about “the algorithm”
- What proportion of searches produce an AI summary rather than clicks
- Current crawler names and their behaviour
Those belong in a dated landscape note or a platform note. If a claim needs a year attached to be true, it isn’t technical SEO — it’s news. See the version gate.
Where ecommerce sites actually fail
Ranked by how often they’re the real problem:
- Faceted navigation generating millions of crawlable URL combinations, consuming crawl budget on pages that should never be indexed
- Client-side rendered product content, which delays indexing for Google and removes it entirely for AI crawlers
- Duplicate product URLs — same SKU under multiple category paths, with no canonical
- Migrations without redirect mapping — the single most destructive routine event in ecommerce SEO — Site Migrations and SEO
- Out-of-stock handling. Deleting a discontinued product’s URL loses its accumulated authority; leaving it as a 200 with no content is worse. Redirect to the nearest equivalent, or keep the page with alternatives
- Pagination on long category listings, where the pattern has changed several times and most implementations are stuck on an obsolete one
- Soft 404s — an empty search or filter result returning 200 with “no products found”
- Parameter sprawl from tracking, sorting and session identifiers creating endless URL variants
Who owns it
The recurring organisational problem: technical SEO sits between marketing, which is measured on traffic and can’t change the code, and engineering, which can change the code and isn’t measured on traffic.
What works: treat it as an architectural constraint, not a marketing request. URL structure, rendering strategy and redirect handling get decided at design time, with SEO consequences stated then — because after launch every fix is a migration.
Measuring it
- Search Console — coverage report, crawl stats, URL inspection showing the rendered HTML Google actually has
- A crawler (Screaming Frog or similar) run against staging before launch, to catch broken links, redirect chains and missing canonicals
- Log file analysis — which URLs bots actually request, which reveals crawl budget waste that no other method shows
curlto see what a non-rendering crawler receives, which is now a materially different question from what Google sees