The Attention Economy Finally Meets the Model Economy
Two business models built on different currencies are now competing for the same user, and one of them is much older.
The dominant business model of the consumer internet monetises attention: content is free, users are the product, and revenue is proportional to engagement.
The emerging model monetises inference: users pay for a task to be completed, and the unit of value is a query rather than a click.
Where the two models collide
- An assistant that answers directly removes the page views that advertising requires.
- An advertising-funded assistant must insert commercial influence into an answer, which users notice.
- Subscription models cap engagement rather than encouraging it.
- Publishers whose content trains and grounds the systems receive no default compensation.
The engineering is easy and the incentives are not
Every incumbent with an advertising business faces the same problem: the product that users want is the product that reduces their revenue. The engineering is straightforward. The incentives are not.
The experimental answers on the table — ads inside generated answers, revenue-sharing with publishers, subscriptions bundled into existing products — all work in the sense that they generate money. The question is whether users accept a slower, more cluttered interface once the novelty fades.
Search displaced directories, mobile displaced desktop
This is not the first time a new interface has threatened an advertising business. Search disrupted directories; mobile disrupted desktop; social disrupted search. In each case the incumbent with the largest existing base adapted more slowly than a new entrant, and in each case the advertising money eventually found a new arrangement. The transition is where the casualties happen.
Impressions are easy to count; tokens are a cost
Advertising is sold on impressions and clicks, both of which are easy to count and easy to inflate. Inference is sold on tokens, which is a cost rather than a value metric. The two require different reporting and reward different behaviour.
A company built on the first will optimise for engagement even when engagement reduces the value delivered. A company built on the second has the opposite temptation, rationing usage in ways that make the product feel stingy.
A subscription, plus ads somewhere else
The arrangement that appears to be emerging combines a subscription for individual users with advertising presented as a distinct, clearly labelled surface rather than embedded in answers. That preserves the advertising business while keeping the assistant’s output free of commercial distortion.
Whether users accept the compromise, and whether advertisers pay enough for a secondary surface, are empirical questions that the next few years will answer.
There is also a question about whether the subscription habit transfers. People pay for professional tools they use daily and resist paying for ones they use weekly. Assistants sit awkwardly between the two, which is why the products that have converted best are the ones embedded in something the user already pays for.
Image credit and licence details for every photograph on this site are listed on the credits page. This article is editorial content; it carries no sponsored material.