Chatbots are becoming ad channels, and shoppers are biting

If you sell anything online, the place your customers ask questions is about to become a place your competitors can buy their way into. Amazon's shopping assistant and OpenAI's chatbot are converging on the same business model: answer the question, and quietly attach a paid brand to the answer.

What the 20% figure actually tells us

According to Marketplace Pulse, Amazon told investors that roughly one in five shoppers who hit a sponsored brand prompt inside Rufus stay in the conversation about that brand. That is the whole ballgame. Display ads on a product page are easy to ignore; a suggestion woven into an assistant's reply is not. When a fifth of exposed users keep engaging, the placement is working as intended, and Amazon now has the internal proof it needs to sell more of them.

The mechanism matters. Rufus doesn't just show a banner. It shapes the shopping conversation, and a sponsored prompt nudges that conversation toward a paying brand. Multiply that across millions of queries and you have a new inventory type that didn't exist two years ago.

Why ChatGPT copying this is the real story

Amazon monetizing its own store is expected. The bigger shift is that ChatGPT is moving toward the same logic. OpenAI has spent its early life as a general assistant with no obvious way to pay for itself beyond subscriptions. Commerce answers are the natural fit: high intent, clear value, willing advertisers. Once an assistant recommends products, the distance between "helpful answer" and "paid placement" is very short.

For brands, this means the discovery surface is fragmenting again. You optimized for Google, then for Amazon search, then for social. Now there is a third front: being the answer an assistant gives when someone asks what to buy. That is harder to game than a keyword, because the assistant decides what to surface, and increasingly it decides partly based on who paid.

The catch nobody has solved yet

Two problems sit underneath this optimism. First, disclosure. When an assistant blends a paid recommendation into a natural reply, users may not know they're seeing an ad. That is a regulatory and trust risk, and both platforms will have to navigate labeling rules that don't fully exist yet. If sponsored answers feel like manipulation, engagement numbers can reverse fast.

Second, dependence. If assistants become the front door to shopping, sellers lose the direct relationship they spent years building. You won't rank; you'll bid, or you'll be invisible. That is great for the platforms and precarious for everyone selling through them, especially smaller brands without ad budgets to compete for prompt-level visibility.

The honest read: this convergence is early and the metrics are self-reported by companies that benefit from hype. A single earnings-call statistic is a signal, not a settled trend. But the direction is clear enough to plan around. Treat conversational assistants as a paid channel in the making, and start asking now how your products get recommended when a machine is doing the recommending.