A site people find,
and that converts.

We measure your site on a phone in real conditions, not on fibre. We list what is blocking, ranked by effect and not by tool. We fix the technical side first, the structure next, the content last. We measure in enquiries received, not in visits.

What you get: A list of fixes ranked by effect, a site that appears quickly on mobile, a structure that lets you cover a subject, and a monthly report of four indicators that change a decision.

What you do not have to do: Buy links, publish thirty articles a month, or take our word for it: everything we fix is measured before and after.

A fast, properly structured site that answers a real question gets found. A slow, badly marked-up, generic site does not, however many plugins are installed.

We work the three layers in order: the technical side first, because it conditions everything else, the structure next, because it is what lets you cover a subject, the content last, because it only has value on sound foundations.

We measure in rankings and in leads, not in traffic volume. A site that gains ten thousand visitors with no intention to buy has gained nothing.

The practical consequence is that a badly ordered search budget is wasted. Producing content on a slow, badly structured site amounts to filling a leaking bucket: the pages exist, they do not rank, and people conclude that SEO does not work when nothing was measured under good conditions.

The choice of platform weighs heavily in that equation, because it determines what can be fixed and at what cost. We have documented it in our comparisons, in particular Webflow or WordPress and Shopify or PrestaShop for online retail.

Technical, structure, content. That order is not a methodological preference, it is a dependency: each layer conditions the return on the next.

The technical layer conditions everything else

As long as a page displays slowly, loads badly on mobile or is poorly accessible to crawlers, the content it carries cannot produce its effect. It is not a question of penalties, it is mechanical: a page search engines crawl badly is a page they evaluate badly, and a slow page loses some of its visitors before it is even read.

This layer has an advantage that is rare in search work: its effects are fast and verifiable. You fix, you measure, you see. It is the only part of the trade that behaves like engineering rather than like a bet.

The structure decides what you can cover

Once the foundations are sound, the question becomes one of organisation. A site where every page covers a bit of everything ranks for nothing, because search engines do not know which one to offer. A site where each page covers a clearly bounded subject, and where linked pages pass context to each other, behaves quite differently.

It is also at this layer that you prevent cannibalisation, meaning two pages on the same site targeting the same search and depriving each other. It is a problem invisible without analysis and common on sites that have grown by accumulation.

Content is only worth something on sound foundations

It comes last not because it matters less, but because it is the most expensive line and it has to be spent in the right place. An article published in a clear structure, on a fast site, works for years. The same article published in disorder never rises and becomes one more dead page.

What that order implies for the calendar

The first months of an engagement produce almost no content, and that is normal. They produce the conditions in which content will pay. We say so at the start, because it is the main source of misunderstanding: a first-month report that announces no new article is not a wasted month.

An audit that produces a list of two hundred anomalies sorted by tool serves no purpose. What serves is a short list, sorted by effect, with what we fix and what we deliberately leave.

Display speed, measured on mobile and in the field

A site is measured on a mid-range phone on an average connection, not on a desktop computer on fibre. The gap between the two measurements is often a factor of three, and it is the bad one that matches your visitors.

The real causes are almost always the same: images far too heavy for the size at which they display, third-party scripts accumulated over the years that nobody can account for any more, no page caching, and fonts that block the display of text. None of those fixes is spectacular; together they make the difference between a page that appears in one second and a page that appears in four.

Accessibility to crawlers, which is checked and not assumed

• The robots file and the indexing directives, which sometimes block exactly the pages you wanted to rank.
• The sitemap, which must contain every useful page and only those. A sitemap that forgets a whole content type amounts to not having published it.
• Duplicate addresses, which make search engines work on several versions of the same page.
• Worthless pages left indexable, which dilute the overall evaluation of the site.

Markup, which is plumbing and is treated as such

One main heading per page and one only, subheadings that genuinely reflect the plan of the text, page titles distinct from one another, descriptions written rather than generated. None of that lifts a page on its own, but the absence of all of it stops the rest from working.

Redirects, when there is a rebuild

It is where you lose fastest what took years to build. Every address that disappears without a redirect is an abandoned asset. We map old page to new page before the launch, and we check afterwards, because an untested redirection plan is always discovered too late.

The temptation, when you want traffic, is to multiply pages. It is the strategy that fails most regularly, and it fails in a particular way: it does not produce nothing, it produces a slow and general decay.

Why thin pages cost the whole site

A page that adds nothing beyond what can already be found elsewhere does not rank. So far, the cost is limited to producing it. The real cost appears next: quality is assessed at the scale of the site, and fifty weak pages weigh on the assessment of the ten good ones. You have not added potential traffic, you have diluted your own.

It is the reasoning that makes us refuse, on our own projects as on our clients’, grids of pages broken down by city or by keyword with no content of their own. They fill a spreadsheet and languish at the bottom of the results.

The cluster, which is the format that works

A main page covering a subject in depth, surrounded by satellite pages each covering one precise question within that subject, and internal links connecting them both ways. It is what we apply here: our comparisons, such as n8n or Make, point to the service pages, which point back to them.

The point often missed is the both ways part. A set of pages all pointing to a main page without ever receiving a link back passes nothing on: they are dead ends, and their content does not benefit the rest.

Choose subjects on intent, not on volume

A keyword searched a great deal but with no intent to buy brings traffic and zero enquiries. A keyword searched ten times a month by people looking for exactly what you sell can pay a salary. So we prioritise on intent and on real difficulty, never on the volume a tool displays.

It is also what makes us set aside subjects that look obvious. An agency fighting over a generic, highly competitive query spends its budget against sites established for ten years, while the searches that convert stay free.

Search is a market where the promise is easy and the verification slow. A few practices are common and we do not offer them.

A ranking guarantee

Nobody controls a search engine’s ranking, and a guarantee of first place is either a guarantee on a search nobody makes, or a promise that cannot be kept. We commit to what we control: what is fixed, what is produced, and what is measured.

Buying links in volume

Links count, and that is precisely why you do not buy them by the kilo. A link bought on a site created to sell links passes nothing on, and the whole eventually forms a recognisable profile. The real lever is slower and more solid: produce something worth citing, then go and make it known.

Mass content production

Generating a hundred articles in a week has become trivial. That is exactly why it is no longer worth anything: what everybody can produce effortlessly distinguishes nobody. We use these tools as writing aids, never as a substitute for knowing something others do not.

The monthly report that says nothing

A forty-page report full of traffic curves is a report nobody reads and that leads to no decision. We follow four: rankings on the searches that matter, the number of enquiries received, the pages that are actually working, and measured speed. A figure that cannot change a decision has no place in a report.

This page covers the three layers in order: the technical side, then the structure, then the content. The third is broad enough to deserve treatment of its own.

SEO writing

How we choose subjects on intent rather than on volume, what a page that genuinely ranks contains, our position on generation tools, and what we do not sell: a fee per word, a guaranteed monthly volume, a ranking guarantee.

The order stays the one described here: content published on damaged foundations gives back nothing. Writing comes after the technical work and the structure, not instead of them.

And to know where to start

Our SEO audit produces the list of what is blocking, ranked by effect, with an order of treatment. It is the most profitable groundwork when nothing has yet been measured.

Two very different lines sit side by side above: a one-off engagement and continuous support. They do not answer the same need.

The one-off engagement

It includes the technical and semantic audit, the fixes that follow from it, the restructuring and the internal linking, the markup, the redirection plan if there is a rebuild, and a prioritised content plan. It produces a sound site and a marching order. It does not produce rankings on its own, and we say so beforehand rather than afterwards.

Continuous support

It makes sense when search is an acquisition channel and not a bonus, meaning when a ranking gained translates into enquiries. It includes prioritised content production, technical fixes as they arise, rank tracking and a short monthly report.

It makes no sense on a site that almost nobody visits and whose subject nobody searches for. In that case we say so and we point the budget elsewhere, often towards digital strategy or towards a channel that produces faster.

What is costed separately

Subscriptions to measurement tools, which are in your name. Advertising campaigns, which belong to another trade and another budget. And translations, because a multilingual site is a project in itself, not a search option.