DoorDash and its peers turn the moment you're hungry and scrolling into a hundred-billion-dollar ad business — but what happens when an app does your choosing for you?

This explores how marketplace platforms make money from the moment a human scrolls, compares and picks, and what happens to that money when software starts doing the picking. One thing up front: the corpus has nothing on DoorDash's ad business itself, its numbers, or how its ad products work. What it does have is a sharp account of why that kind of business works, and why it may be more fragile than its size suggests.

The core idea is that a platform's ad business is really a toll on human indecision. When you open an app hungry and unsure, the platform controls what you see first, and sellers pay to be there. Retail media is built on that. The clearest way to see it is through what breaks it. Jarc argues that when an agent holds the demand signal, meaning your actual stated criteria, and ranks options against them, paid placement stops mattering because nobody is looking at the page Are AI agents making attention-based ads irrelevant for aggregators?. Tripadvisor's 23% drop in hotel revenue is offered as an early sign. Rustamli makes a finer distinction: agents eat the ad revenue tied to needs ("I want dinner delivered") because they turn discovery into a zero-click transaction, but they can't create new wants, so ad money moves toward social and influencer channels instead of disappearing Do AI shopping agents destroy search ad revenue while leaving influencer marketing intact?. For a delivery app, that's an uncomfortable split, because most of its traffic is need-driven.

The corpus also suggests the attention layer doesn't go away. It moves. One line of research argues that services will compete for an agent's selection the way they now compete for human clicks, rebuilding ranking and promotion infrastructure aimed at machines Will agents compete for attention just like users do?. Jaipuria offers the more hopeful version: if agents optimize for value rather than page position, merchants have a reason to turn ad budgets into lower prices and better terms, and Google's Direct Offers pilot already points that way Could agents redirect merchant ad budgets toward customer savings?. Jaipuria also sketches how an agent's own revenue per transaction could climb from under 1% toward booking-platform margins in stages, which shows that the new intermediary wants the same toll the old one collected How could an AI agent's revenue per transaction grow over time?.

The part you may not have expected to care about: advertising incentives don't disappear when the shopper is a machine. They get harder to see. Hedges warns that shopping agents funded by ads have a built-in pull toward paid placements, however well they're engineered Do advertising-funded shopping agents favor paid product placements?. A simulation found that when a platform tilts its rankings and sellers tilt their product descriptions, each acting alone, the combined harm to consumers is more than double either one, because both are exploiting the same weak spots in how AI shoppers reason Do platforms and sellers exploit AI shoppers together more effectively?. A formal model shows platforms can price how many sellers an agent gets to inspect so that it searches more than it needs to, shifting money from sellers to the platform while consumers end up no better off Does platform pricing of inspection tokens encourage excess search?. Stoller's account of Google tying together personal data, chatbots and payment deals shows the endgame: whoever controls the agent layer inherits the chokepoint that the attention layer used to provide Is Google building a chokepoint across data, chatbots, and payments?.

So the honest answer is that the corpus can't tell you how DoorDash built its ad business. It can tell you that the business rests on humans looking at a ranked list, and that the research expects that list to be read more and more by agents. The open question is whether the same economics get rebuilt for machine readers, or whether ad budgets finally turn into lower prices.