Technical SEO for modern sites — what matters beyond keywords

A site can have the best content in its niche and still never be found. Usually the writing is not the problem — the infrastructure around it is. Technical SEO is the work of making a site trivially easy to crawl, render, understand and trust. Everything else — keywords, editorial voice, publishing cadence — comes after that.

A search engine does four things, in order: it discovers addresses, downloads them, decides whether they are worth rendering, and only then indexes them. Any of those steps can be broken by an architecture decision made months before the first article was written. Technical SEO means removing, one by one, the reasons a step might fail.

It is not a bag of tricks and it is not a setting you tick once. It is a property of the product, exactly like performance or security.

The most expensive modern mistake is content that only exists after JavaScript runs. Crawlers can render, but rendering costs time and resources and is never guaranteed on every visit. If your text, internal links and metadata only appear after a browser-side call, you have bet your visibility on an optional step.

The practical rule: anything you want indexed must be in the HTML the server delivers. Interactivity can stay in the client. I unpacked the trade-offs of that split in the server components architecture piece — it is the same conversation, seen from the code side.

• Check the page source, not what the inspector shows after hydration.
• Content behind a click (tabs, accordions) must exist in the initial DOM.
• Infinite scroll without real links leaves half your catalogue undiscovered.

Every piece of content deserves exactly one URL. In practice duplicates appear through tracking parameters, trailing-slash variants, http versus https, www versus non-www, and filter combinations that multiply without limit.

• Pick one form and permanently redirect the rest.
• Put a self-referencing canonical on every indexable page.
• Stay consistent: internal links, the sitemap and Open Graph must all use that same form.

A canonical is a strong hint, not a command. Contradict it with inconsistent internal links and you weaken it yourself.

The title element and the meta description are not storage for keywords. They are your free advert on a results page crowded with competitors.

A good title says what the page is and who it is for. A good description says why the click is worth it, instead of restating the title in different words.

Write them uniquely per page, keep the title short enough not to be truncated, and put the distinctive information first. The description does not directly influence ranking, but it influences click-through — and that shows up in the data.

Formally, Open Graph belongs to social networks and messaging apps. In practice, a serious share of traffic to any article arrives through a shared link. If the image is missing or broken, you lose clicks before any search algorithm gets a say.

• A 1200x630 image in a format every client can read.
• Open Graph title and description written for sharing, not copied mechanically.
• Test with the official tools, because every platform caches aggressively.

JSON-LD is the cleanest way to state explicitly what a page is. Article for editorial pieces, FAQPage for genuine questions, BreadcrumbList for hierarchy, Organization or Person for the entity behind the site.

Structured data does not guarantee rich results. It makes a page eligible. The decision to display anything stays with the search engine and changes often.

The golden rule: mark up only what is visible on the page. An FAQ that exists solely inside JSON-LD is, at best, ignored.

A sitemap is the list of canonical addresses you want indexed — not a dumping ground. No redirects, no noindex pages, and a modification date that is actually true. The robots file controls crawling, not indexing: a page blocked there can still surface in results, just without a description. If you want something out of the index, use noindex and let the crawler see it.

This is where the most is won for the least effort. Internal links tell the engine what matters, pass authority, and supply context through anchor text.

• Descriptive anchor text beats a bare read more every single time.
• Hub pages gather related topics and give a subject real structure.
• Orphan pages — the ones nothing links to — are effectively invisible.

This article does what it recommends: it points onward to Core Web Vitals and to accessibility as a competitive advantage, because those subjects genuinely continue this one.

hreflang only works when it is reciprocal: every variant lists all of them, including itself, plus an x-default value. The classic mistakes are invented language codes, relative addresses, and hreflang pointing at redirected or non-canonical pages.

For pagination, each page in the series keeps its own canonical rather than one aimed at page one. And thin pages — auto-generated, with no content of their own — do more harm than good: merge them or remove them.

Images need descriptive alternative text, explicit dimensions so they do not shift the layout, and modern formats to keep the weight down. Speed and accessibility are not magic ranking factors, but both reduce friction: a fast page that works from the keyboard is a page people actually read, share and link to.

Search Console is the source of truth: the indexing report tells you what was excluded and why, and field data shows the real user experience rather than a lab simulation. Watch impressions and average position before clicks, because they move first.

The healthy order of work is: fix what blocks indexing, then what blocks understanding, then what blocks the click. Results show up in weeks, not days, and consolidate over months.

If you want that assessment run on your own site rather than a hypothetical one, get in touch.