A content site answers a query and is forgotten. A publication has a point of view, editorial standards, relationships with its subjects, and something worth returning to.
The distinction sounds like vanity until you look at what each one is built to do. A content site is built to intercept a question on its way to an answer. Once the answer is given the site has done its job and has no further claim on anybody.
The economics gave us content sites
For fifteen years the reliable way to make money from writing was to publish a page for every question a person might type, and to make each page just good enough to rank. The reward was for coverage, not for judgment. Nobody set out to build something forgettable. The incentives simply never paid for the alternative.
That trade is now closing. Machines write adequate answers faster and more cheaply than any team, and pages that only ever answered a question are being answered over. Whatever survives will be something a person chose to come back to.
A content site is measured by how many questions it answers. A publication is measured by how often somebody returns without one.
What a publication has that a content site does not
A point of view, first. Ours is written down in the manifesto, which commits us to positions that cost us traffic. That cost is the only real test of whether a position is held.
Standards, second. We say who visited, when they visited, and what we did not check. A content site rarely says anything about how it knows, because usually it does not know.
Relationships, third. We go back to the same businesses, which means living with what we wrote about them. That is uncomfortable and it is the most useful discipline we have.
A content site can have none of these and still work, for a while. It does not need to be trusted, only to be found. That is why so many of them read as though nobody in particular wrote them, and why so few of them are missed when they disappear.
Brand is the shorthand for all of it
Brand in our sense is not a logo or a palette. It is the expectation a reader arrives with before they have read a word: that we went to the place, that we will say when something is not good, that we will still be here next year.
That cannot be bought and it cannot be generated. It is the residue of a few hundred decisions to do the slower thing, and it is the only asset in publishing that does not lose its value the moment an algorithm changes.
What it means for how we build
It means four publications rather than forty. Each covers a subject narrow enough that we can hold an opinion about all of it: specialty coffee, waste and repair, better sailing, one stretch of coast.
It means the writing carries a name, so somebody is answerable for it. It means we use software to research at scale and to keep information current, and it decides nothing about what we recommend.
It also means we grow slowly. A publication cannot be launched into an audience; it accumulates one. The first year of any of ours looks like a failure by the measures a content site uses, and looks like the only sensible investment by the measures we use.
And it means we would rather be the second result and the first bookmark. A publication that is returned to does not have to win the query every time, which is fortunate, because nobody wins the query forever.