For twenty years the arrangement was understood on both sides. A publisher produced material, a search engine indexed it, and in exchange for the indexing the publisher received visitors it could sell advertising against. Nobody signed anything, but the trade was real and it financed a great deal of journalism.
The trade is being renegotiated unilaterally. When a query returns a synthesised answer rather than a list of destinations, the reader's question is resolved without the visit, and the publisher's side of the exchange disappears while its contribution does not.
Building for an audience that does not arrive
The publishers responding coherently have stopped treating referral traffic as a channel to be optimised and started treating it as a source that may not exist in three years. That reframing produces different decisions than a recovery plan does.
The first is a hard shift toward relationships the publisher owns. An email address, an app install, a paying subscriber — each is a connection no intermediary can reprice. This is unglamorous, slower and much smaller in raw numbers than search ever delivered, and it is the only audience that cannot be taken away by a product decision made elsewhere. The economics of that shift are why the newsletter business has moved into consolidation rather than expansion: the durable audiences are worth acquiring.
The second is revenue that does not require a pageview at all. Events, licensing, research and services all convert an audience relationship into money without an impression, and they scale badly, which turns out to be a feature. A revenue line that cannot be arbitraged by a platform is worth having even at lower margin.
The third is negotiating for the material itself. If a model is going to answer using a publisher's reporting, the reporting has a price, and the market for licensed training and retrieval data has matured enough that the negotiation is now commercial rather than theoretical. The leverage is uneven — it accrues to publishers with archives, specialisation or a brand a model builder wants to cite — but it is real leverage where it exists.
What is striking is who is least affected. Publications serving a defined professional readership, which never depended much on general search, are largely unbothered; trade publications have outlasted the consumer magazines that condescended to them for exactly this reason. The same insulation explains why the independent operators building direct relationships from the start look structurally sturdier than organisations many times their size.
The exposed middle is the general-interest publication of moderate scale, which built a cost structure around volume and has neither a specialist audience nor an archive anyone needs to license. That is where the contraction is landing, and it is landing on newsrooms that did nothing wrong except succeed at the thing that was being measured.
None of this is a rescue. The replacement audiences are smaller than what search provided, and the businesses being built on them are smaller businesses. The publishers making the transition have mostly accepted that first, which appears to be the prerequisite for everything else.
Topics mediapublishing
