In 1900 there were roughly 3,000 cars in France. Michelin printed about 35,000 copies of a free guide to driving. It was not advertising. It was an attempt to make driving practical enough to become normal, because more driving meant more worn tires. Publishing built the market the company would later sell into.
That mismatch, 35,000 guides for 3,000 cars, is the part worth stealing. Michelin was not capturing demand. Search did not exist. It was assembling the conditions under which demand could appear at all.
Founders in AI agents, developer tools and other young categories are in the same position. The keywords are thin. Search volume rounds to nothing. Buyers are few and hard to find. The instinct is to wait until the category matures. Michelin is the argument for doing the opposite.
Publishing before the demand exists
The first guide was infrastructure wearing the costume of a booklet. Maps. Fuel stops. Mechanics. Lodging. Instructions for repairing a car at the roadside. None of it was about tires.
Michelin had understood the economics with unusual clarity. More driving meant more wear, and more wear meant more tires. So the useful thing to publish was not a tire catalogue. It was everything standing between a person and a drive.
Most founders publish about their product. Far fewer publish about the work their product makes possible. That is the whole difference, and it shows up in what you choose to write about:
What blocks the first use case?
What confuses a buyer before they are ready to buy?
What makes adoption feel risky?
What language does not exist in this market yet?
Answer those before anyone is searching for them and low search volume stops being a warning sign. Sometimes it is just evidence that the category is still being built.
Utility beats cleverness
The early guide answered plain questions. Where do I refuel. Who can fix this. Where should I stay tonight. How do I plan a route that works.
Durable B2B content has the same shape. It is not trying to be clever. It is trying to be complete enough that someone can act on it.
Chartbeat data reported by Axios in March 2026 points the same way. News and media sites pull the highest raw page views from AI platforms and the lowest engagement per article, which reads like people checking a fact and leaving. Sites built on practical guidance see more page views per article. Someone arriving from a chatbot is not browsing. They want something they can act on.
Michelin was built on answers a century before anyone said "answer engine."
For a technical founder this is the harder version of content marketing. It means writing for a buyer with real constraints rather than a reader who wants a summary. It means documenting the route from problem to resolution, not just naming the problem and stopping.
The 1920 pivot, and what it was really about
By 1920 the guide had been free for twenty years. André Michelin reportedly found a copy propping up a workbench. Whatever the truth of the story, the conclusion drawn from it was serious: a free product, however useful, had become disposable.
So Michelin changed the model. The guide became a paid product at 7 francs. Advertisements came out. Anonymous inspectors went in. The publication started behaving like an editorial institution rather than a promotional insert.
The line associated with the decision is blunt. Man only truly respects what he pays for.
Set the phrasing aside and the mechanism still holds. Free creates reach. Price creates seriousness. Seriousness is what trust is built on. Founders underprice early offers because they confuse accessibility with respect, and a low price attracts attention along with casual buyers, weak commitment and muddled positioning.
Trust is not created by usefulness alone. It is created when usefulness is paired with cost, discipline, and credible judgment.
This does not mean gating your content. It means every asset should have a position on its own value. If everything you publish is free and undifferentiated, the market learns to look past it.
There is a simpler version of the test. Would anyone pay for this? If the answer is no, the work may still be useful, but it will struggle to be treated as authoritative.
That is not an argument for charging. It is an argument for having some mechanism that signals rigor: a published methodology, visible editorial standards, named expertise or transparent curation, evidence instead of slogans. Without one of those, content stays helpful and forgettable.
Stars, and why judgment beats neutrality
Michelin added stars in 1926 and expanded the system in 1931. That is when the guide stopped being a resource and became an authority.
The star system was not neutral. It was opinionated, selective and defensible, and that is precisely what made it citable. Over time the judgment inside it became the reference point for restaurants, travelers and an entire industry.
Neutral summaries rarely become standards. Standards need a point of view, clear enough to be useful and disciplined enough to survive being argued with. Your content should help a buyer decide, not merely inform them that a decision exists.
This matters more now, not less. AI systems favor sources that are explicit, structured and trustworthy. Volume will not get you cited. Recognizable judgment might.
Michelin built that judgment slowly, then made it institutional. Anonymous inspectors were part of the point: the public trusted the process more than any personality inside it. In a market full of founder-led performance, a visible process is a differentiator.
Authority took twenty-six years
The guide launched in 1900. The stars arrived in 1926. Authority did not come with the first issue, or the fiftieth. It accumulated.
This is where most modern content programs die. They expect the first handful of posts to compound, and they stop before the market has had time to form a memory of them.
You cannot manufacture authority in a quarter. You cannot shortcut editorial consistency. You cannot publish vague thought leadership and expect it to be mistaken for expertise. Content that is merely frequent gets forgotten. Content that is merely polished gets ignored. It has to be useful enough to keep and specific enough to trust.
Read that as reassuring rather than discouraging. Patience is the one advantage that is not for sale, which means it is still available to companies too small to buy anything else.
The same mechanism, in an answer-engine market
Distribution has changed repeatedly. Guidebooks gave way to search. Search is giving way to AI answers. The trust mechanism underneath has not moved at all.
People still reward clarity. They still reward usefulness. They still trust whatever reduces their uncertainty.
Which is why the assumption that AI has made content less valuable gets it backwards. An answer engine assembling a response still needs material worth citing. Generic commentary gives it nothing. Practical, well-structured, high-signal writing gives it something to quote and someone to name.
Two questions are worth holding a draft against.
Would someone pay for this if it were not free? And would an AI engine cite this if it had to explain your category to someone?
Two yeses and you are building a trust asset. Two nos and you have content without leverage.
What a two-person company can copy
Not the scale. The logic.
Start with the questions your market asks before it asks about you. Write the guide you wish had existed before a buyer ever became a lead. Work on whatever makes your category easier to enter.
In practice that tends to look like:
A plain-English guide to the decision itself.
A teardown of the mistakes people make first.
A pricing or implementation checklist.
A short benchmark or comparison framework.
A first-customer playbook that removes doubt rather than adding excitement.
What you should not do is treat content as a thin SEO layer laid over a weak offer. Michelin was useful because the company understood the market it served. The publishing supported a commercial logic; it did not substitute for one.
If you want the modern version of that run properly, publishing for the market, in the founder's voice is the work we do.