Marketplaces Layer AI, Preserve Ads
Andrew Yates, CEO of Promoted.ai, on when marketplaces should layer on ads
Amazon cannot replace keyword search with a fully generative interface without risking the two things that make commerce search work at scale, speed for shoppers and monetization for sellers. Most shopping queries are short navigation tasks where a fast ranked list is enough, and Amazon already layers a chat assistant, Rufus, on top of that core flow rather than swapping the whole surface. That preserves low latency, familiar buying behavior, and the sponsored listing inventory that sits inside results pages.
-
Amazon launched Rufus in February 2024 as a shopping assistant inside the existing app experience, first in beta for a small U.S. cohort. The design choice matters. Generative AI was added as a helper for research and comparison, not as a wholesale replacement for standard search results.
-
Amazon search is also an ad marketplace. Sponsored Products are CPC ads that appear at the top of, alongside, and within shopping results, and take shoppers straight to product detail pages. A pure chat answer would shrink these placements and make seller bidding, measurement, and click based pricing harder to preserve.
-
The broader pattern across marketplaces is to keep improving ranking and relevance first, then layer monetization on top. In this ecosystem, ads work because marketplaces control placement inside search and discovery surfaces. That is why the mature winners add AI into ranking and assistance incrementally, instead of blowing up the page layout that already converts.
The next step is a hybrid model. Generative AI will handle research, comparison, and follow up questions, while the main results page remains a fast ranked grid with embedded ads and clear product links. As inference gets cheaper, the winners will be the marketplaces that fuse conversational help into search without giving up the economics and control of the results page.