The technical work that decides whether anything else you do counts

It is the work that makes sure search engines can reach your pages, understand them, and choose to include them — and that the site is fast and stable enough not to be penalised in the experience it delivers.

It rarely wins a competitive ranking by itself. What it does is remove the ceilings. If a page cannot be crawled, rendered or indexed, no amount of content or authority will help it.

Most audits are a crawl tool export with severity labels attached. They are technically accurate and practically useless, because nobody can tell what to do on Monday.

Ours ends in a prioritised backlog: each item written as a ticket, with the expected impact, the effort, the pages affected and how we will verify it after release. If your engineers cannot pick it up without a translation meeting, we have not finished.

It might, but not in a way anyone can quantify in advance, and it is not the main reason to do it. Google publishes thresholds for good experience — Largest Contentful Paint at or under 2.5 seconds, Interaction to Next Paint under 200 milliseconds, and Cumulative Layout Shift under 0.1 — measured on real users rather than in a lab.

The stronger argument is commercial. Slow, unstable pages lose people before they convert, and mobile is where most of that loss happens.

Yes, and the work starts well before launch. We map every URL that has value, plan redirects, check that internal links, canonicals, structured data and sitemaps all point at the new structure, and set up monitoring so a regression is caught in days rather than at the next quarterly review.

We cannot promise zero movement. Some fluctuation after a move is normal. What is avoidable is the six-month decline that follows a migration where only the redirects were considered.

Not inherently, but it changes what has to be verified. The question is what search engines actually receive: whether important content and links exist in the initial HTML, whether rendering completes reliably, and whether the routes are real URLs rather than fragments.

We test what is genuinely served rather than assuming the framework handles it, because the failure mode is silent — the page looks perfect to you and is largely empty to a crawler.