Picture a subcontractor on the estuary that employs sixty people, holds certifications that took three years to obtain, and has customers who would describe it as excellent. Type the phrase a new buyer would use to find that capability and the company does not appear anywhere. Not on the second page. Not at all. Every client it has was won through a trade show, a recommendation, or a relationship that predates the current sales director.
That is not a rare case in Nantes Métropole. Roughly eight jobs in ten in the metropolis are in services, but the industrial base remains significant: shipbuilding on the estuary, aeronautics, agrifood, materials and mechanical engineering, and an electronics subcontracting fabric that runs up toward Saint-Nazaire. The metropolis names five strategic ecosystems it supports — health, manufacturing, maritime, digital, and the cultural and creative industries. Many of the companies inside them are hard to find for what they actually sell.
The reason given, when you ask, is often the same: search engines are for consumer businesses. Our buyers are professionals, they know the market, they come to us. Part of that is true and getting less true every year. Buyers are people with a browser and a specification. When the person who used to know the supplier list retires, or the purchasing function is centralised into a group two hundred kilometres away, the replacement starts where everyone starts.
It helps to look at how an industrial search actually begins, because it looks nothing like a consumer one. Nobody types the name of your sector. They type a process plus a material plus a constraint: a machining capability with a maximum dimension, a treatment on a specific alloy, a packaging format for a food product, a test bench for a certification, a subcontracting capacity in short runs. These phrases are long, precise, and written by someone who already knows what they need. That is why they tend to convert well.
They are also, individually, tiny. A precise procurement phrase may attract only a couple of dozen searches a month across the whole country. That number is one reason this ground is often open, and it is the reason it is worth taking. An ad budget cannot sensibly be managed at that volume: the management overhead exceeds the value of the clicks, and the moment you stop paying you disappear again. A page written once against the same phrase costs a fixed amount of effort and can then keep working for years. Low volume, high value is exactly the shape that punishes advertising and rewards search.
There is a further asymmetry worth being blunt about. Many local agencies in Nantes sell to restaurants, dentists, tradespeople and estate agents. Those categories are legitimate and they are contested, because they are easy to prospect and easy to explain. Industrial B2B is neither. It requires understanding a process well enough to write about it without embarrassing the client in front of an engineer. The results pages for precise procurement phrases are often thin — a directory listing, an aggregator, a competitor's homepage that happens to mention the term once.
So what does a page for an industrial buyer contain. Not "our expertise" and not "a company on a human scale since 1987". It states a capability in the words a buyer uses: the process, the materials, the dimensional envelope, the tolerances you hold, the run sizes you take, the certifications you carry and their scope, the sectors you already supply, and the lead times you realistically quote. It is closer to a technical datasheet with a sales paragraph than to a marketing page, and that is what makes it work — the reader is qualifying you, not being persuaded.
The objection that arrives at this point is confidentiality. We cannot name our customers. Frequently true, and it changes less than you expect. You can describe a part without naming who it was for. You can state a sector rather than a client. You can publish the tolerance you hold rather than the programme it belongs to. A buyer does not need the logo; they need evidence that you have done something dimensionally and technically like their problem before.
Then there is the site itself, which in industrial firms tends to have a particular shape: built once, around 2014, with a downloadable PDF catalogue, a page called "our know-how", and a contact form nobody has tested since. A buyer arriving on it has very little time to decide whether you are worth an email. In that time they want to know what you make, at what scale, with which certifications, and whether you are likely to answer. A PDF does not answer that, and search engines read the page rather than the brochure inside it.
The technical facts worth putting in plain HTML text, because they are what people search: maximum dimensions and weights you can handle, materials worked, processes performed in-house versus subcontracted, series sizes from prototype to production, certifications with their scope and expiry, standards you work to, quality documentation you supply, and the languages your commercial team operates in. Each of those is a search someone runs.
Export deserves its own decision rather than a default. If you sell outside France, an English version is worth building deliberately — written, not machine-translated, because a translated page reads as translated to the buyer and is not credited as a separate language version unless it is properly implemented. If you sell only within France, do not build one for appearances. It costs maintenance and returns nothing.
Measurement has to change too, and this is where industrial firms are often talked into disappointment. Traffic is a bad metric here. A page that receives thirty visits a month and produces two qualified enquiries a year is not underperforming; in a business where a single new account can run for a decade, it may be the most profitable page on the site. Ask for reporting on enquiries and their source, and treat rankings on the specific procurement phrases as the supporting evidence. Anyone showing you a traffic curve for an industrial site is measuring the wrong thing.
The internal objection that remains is the strongest one: our business comes from trade shows and word of mouth, and it always has. That is true, and it should continue. Nothing here suggests replacing a channel that works. The point is narrower — the buyer who has never heard of you cannot recommend you, cannot meet you at a show they did not attend, and will find someone. Search is one of the few channels that reach the customer who does not already know you exist.
A reasonable starting point is small. Four to six pages, each one written against a phrase a buyer would genuinely type, chosen with your sales team rather than by an agency guessing at your vocabulary. One per major process or capability. One per sector you supply, if the requirements differ meaningfully between them. Then leave them alone for six months and count enquiries, not clicks.
The uncomfortable summary is that the companies with the most to gain from this search ground are often the least inclined to take it. Industrial firms here are conservative about marketing for reasons that were often sound. But the pages cost a fraction of a trade show stand, and unlike the stand they are still working the following year.