The Click Is Fading, So the Ads Move Into the Answer

Share

Search has always worked the way a storefront works. You go there because you want to find something, and the reason the whole thing stays free is that you keep moving through it. A question turns into a page of links and sponsored results, you click through to a handful of sites, and every one of those stops is a chance to sell you something or a chance to charge the person who wanted your attention. That machine quietly funded the open web for the better part of two decades. So it is strange to watch it slow down at the very moment more people are using it than ever.

Google's third quarter was the clearest sign yet that something has shifted. Search revenue was still growing, but the growth slowed to 12 percent from 17 percent the quarter before, and it slowed while the number of searches hit records. The company was open about the cause on its earnings call. People were searching more than ever, executives said, and the company was making less money on each one, in part because its AI Mode, where the model writes the answer, monetizes worse than classic search. The market responded the way markets do when the top of a once unbreakable machine starts to wobble. Alphabet's stock took a hit that erased roughly half a trillion dollars of value as investors read the gap between what AI costs and what it earns, and concluded that the gap is widening, not closing.

The reason is not mysterious, and it is worth spelling out because it is the whole argument. An AI answer ends the search instead of extending it. In the old model, a query produced a page of links and ads, and you clicked through, generating impressions and paid clicks along the way. An AI answer puts the whole thing at the top. It answers the question directly, and the list of sites you would have visited becomes optional reading. The transaction that used to have ten stops now has one. Fewer clicks means fewer chances to earn, and no amount of added search volume gets that revenue back on its own. Every answer that is good enough to satisfy you is, from the company's side, a small revenue loss compared with the query that launched a journey.

So the ads are not going away. They are moving into the answer itself. Google already runs ads in AI Overviews across a dozen markets, and it is testing formats built for the conversational experience, where the ad is not a separate listing but a part of the thread, introduced in the middle of the exchange. There is even a running dispute about whether ads will eventually come to the standalone Gemini app, with Google publicly disputing a report that it had told ad agencies it planned to. I do not need to resolve that specific fight to make the point, because the direction is already clear. When the click disappears, the money has to be collected somewhere else, and there is only one place left, which is inside the answer people are already reading.

There is a quiet irony in all of this. For years the promise of AI search was that it would finally free you from the commercial clutter, from the sponsored results and the ten blue links, and simply tell you the answer. The pitch was a cleaner experience with none of the interruption. What the pitch left out is that somebody has to pay for the answer. The computation is not free, and neither is the engineering that makes the assistant feel effortless and trustworthy. Remove the click and you remove the toll that used to collect the money. So the toll moves closer to you. The free answer is not free at all, and it is funded by putting the moment of decision, and the offer, right in the middle of the answer.

Let me be fair to the counterargument, because it is a real one. This might simply be the honest equilibrium. A genuinely good answer is expensive to produce, and someone should pay for it. Ads inside answers are not inherently worse than ads next to links, and if they are labeled clearly they might even be better, because an offer that arrives in the middle of a real decision can help someone who is already leaning toward buying. I am not going to argue that advertising in AI is a sin, and I am not going to pretend the old search results page was some pure, commercial free zone, because it was the most profitable advertisement machine ever built. The interesting question is not whether the ads show up. It is where they leave the value sitting.

And this is where the story stops being about Google and becomes about everyone standing downstream of the click. A large share of the modern startup economy was built on search traffic. Companies optimized for the query, the listing, the click, the visit, and turned that pipeline into growth and eventually into revenue. That is the pipeline an AI answer quietly reroutes. When the answer is complete, fewer people make the trip, and the ones who do arrive already shaped by what the assistant told them in the first place. For a founder, this is a familiar lesson wearing a new costume. Distribution that you rent from one platform is a lease, not an asset, and the landlord can rewrite the terms whenever the economics demand it.

The history here is older than AI. Radio was financed by advertising. Television was financed by advertising. The open web was financed by advertising. In every one of those transitions, the people who held the direct relationship with the buyer kept the money, and the people who stood one step away from the transaction were the ones squeezed out. The click, for its era, was a toll, and tolls are lovely until someone reroutes the road. What held up across all of those transitions was not who got found. It was who got to finish the sale. The middlemen who only carried attention from the question to the purchase were the first to lose their fee, and I suspect that is exactly what is happening again, just faster.

The merchants who see it coming are already adapting. The emerging play is to let the AI show the goods but keep the checkout for themselves, and to spread their catalogs across many different agents and assistants so that no single one can hold them hostage. That is the smartest instinct in the room. The more capable the answer becomes, the less room there is for a generic middleman who sits between the question and the sale. Your edge stops being that you are the thing that gets found, and becomes that you are the place the transaction happens, the place where the relationship already lives, the place holding the data and the trust that no model can generate out of thin air. I have spent a lot of time arguing that models are commoditizing and that value moves to the layer above them. The search click is a clean example of that same law doing its work.

There is a second consequence worth naming, and it is about the companies paying for all of this. The big AI labs are spending enormous sums with little to show on the revenue side, and the natural pressure is to make the surface where people actually decide into the place that collects the money. That means the platforms that own the interaction, the moment between the question and the answer, gain leverage, and everyone else competes for the scraps of attention that leak out around the edges. This is not a verdict on whether any individual company is good or evil. It is a statement about where the leverage in this market is concentrating, which is with whoever controls the moment of decision, and that is a concentration founders should watch the same way they watch any other gatekeeper.

I do not think this means search is dying, and I am not claiming that ads inside answers are the end of the world. Search will keep being useful, and advertising will keep funding things people want for free, the way it always has. What has changed is the shape of attention, and the money follows attention wherever it goes. The free answer quietly moved the market's clock forward, and a lot of businesses that leaned on the click are about to discover they were renting their front door. If you are building something, the durable question is not whether you show up in the answer. It is whether the person who asked the question is yours to keep.