تحسين محركات البحث

Most sites I audit are not held back by anything exotic. They are held back by a handful of ordinary structural problems — pages that cannot be crawled properly, canonicals that contradict the sitemap, private routes in the index, layouts that shift while loading — combined with content that does not quite answer what the searcher asked.

This is the structural half of that work, done during a build where possible and retrofitted where necessary.

An honest note on what SEO work can promise

Nobody can guarantee rankings, and you should be wary of anyone who offers to. Google's own documentation on page experience says plainly that there is no single ranking signal, and that Search still aims to show the most relevant content even when the page experience is sub-par. Good scores in Search Console or a third-party tool do not guarantee a position.

What technical SEO does reliably do is remove the reasons a page cannot rank, and stop you losing visitors who would otherwise have stayed. That is worth doing properly. It is not the same as a promise.

Who this is for

• Site owners whose pages are indexed but not getting traffic.
• Businesses about to launch or relaunch, who want this handled during the build rather than afterwards.
• Sites that lost visibility after a redesign or migration.
• Teams who want a second opinion before committing to an ongoing SEO retainer.

What is included

Crawlability and indexation

• robots.txt that blocks what should be private without accidentally blocking CSS, JavaScript or images.
• An XML sitemap containing only canonical, indexable public URLs.
• noindex on private areas — a Disallow alone only stops crawling, so a URL discovered elsewhere can still be indexed without its content.
• Removing orphan pages, so nothing important is reachable only from the sitemap.

Canonicals and duplication

• Self-referencing canonicals using the exact production URL, consistent on protocol, host and trailing slash.
• Sitemap and canonical tags telling the same story rather than contradicting each other.
• Pagination handled so later pages stay crawlable instead of being canonicalised away along with everything they link to.

Metadata and structure

• Unique titles and descriptions per page, written for the result rather than stuffed.
• One H1 per page and a heading hierarchy that reflects the content.
• Open Graph and Twitter metadata so shared links render properly.
• Structured data that accurately describes the page — and nothing that does not.

Internal linking

• Descriptive anchor text instead of "click here" or "read more".
• Contextual links from body content, not only navigation.
• A structure where important pages are genuinely reachable.

Performance and Core Web Vitals

• Image sizing, formats, lazy loading and explicit dimensions to prevent layout shift.
• Reducing render-blocking resources and unnecessary JavaScript.
• Working to LCP under 2.5 seconds, INP under 200 milliseconds and CLS below 0.1, measured on field data at the 75th percentile rather than a single lab run.

On AI search

A lot of effort is currently being spent on tactics that Google has explicitly said do not apply. Its guidance for generative AI features states there are no special requirements to appear in AI Overviews or AI Mode beyond being indexed, snippet-eligible and technically sound. It also states that Google Search does not use llms.txt files, that there is no special schema markup for generative AI, and that content does not need to be chopped into artificial chunks or rewritten in some machine-friendly dialect.

The practical implication is reassuring: the fundamentals on this page are the work. I would rather spend your budget on crawlability, structure and genuinely useful content than on tactics with no evidence behind them.

What an audit usually turns up

The findings repeat across almost every site I look at, which is encouraging — it means the fixes are known rather than mysterious.

• Pages in the index that were never meant to be public. Admin screens, login pages, staging URLs, authentication callbacks. Usually because a sitemap generator listed every route it found.
• Canonicals that contradict the sitemap, or that point every paginated page back at page one — taking the links to everything on pages two and three with them.
• Titles and descriptions duplicated across templates, so dozens of pages compete with the same snippet.
• Multiple H1 elements per page, usually because heading tags were chosen for their font size rather than their meaning.
• Images with no dimensions, shifting the layout as they load, and no alt text, removing the page from image search and from screen readers.
• Content reachable only from the sitemap — no internal link anywhere on the site actually points at it.

None of these are exotic. All of them are worth more than a month of link building.

Local search

For a business with a physical location or a defined service area, local visibility is often the highest-value channel available. The groundwork is consistency: the same business name, address and phone number on the site and on your Google Business Profile, genuine location context in the page content, and pages that reflect the areas you actually serve.

The approach to avoid is the doorway-page pattern — near-identical pages with the city name swapped. It is a well-documented spam signal and it reliably produces thin pages that rank for nothing.

Measurement

Google Search Console first — it is free, first-party, and the only place showing the queries you actually appear for, which pages are indexed, and what problems Google has found. Analytics tells you what people did next. Anything claiming privileged insight into AI search visibility deserves scepticism; Google's own guidance points site owners at Search Console rather than third-party tools claiming access to internal metrics.

How this fits with a build

Retrofitting is always more expensive than building it in. On a new WordPress build, an Elementor Pro build or a custom Next.js project, this work is part of the build rather than a separate engagement — the structure is decided once, correctly, instead of being unpicked later.