The Comparison Layer Is Becoming an AI Decision Layer
Comparison sites were built to sit between shoppers and merchants. Assistants are becoming the layer between shoppers and the market.
Whether shopping comparison sites are dead is the wrong question. The function they performed, normalizing merchant data into a consumer decision, is moving inside the assistant. Some comparison businesses will survive as suppliers to that layer. Those that depend on people browsing their pages have a harder road than the headline traffic numbers suggest.
What the feed does
ChatGPT's shopping experience draws on a merchant-controlled product feed. Merchants push a structured file of titles, prices, stock and media rather than waiting to be crawled, and refreshes are accepted as often as every 15 minutes. That is the raw material comparison engines have always run on. Idealo, Europe's best-known example, builds its database from CSV files and API integrations supplied by the retailers themselves. The assistant now has direct access to the same structured merchant data that historically powered comparison engines, alongside web pages, reviews and editorial coverage.
The shift shows up in how answers are sourced. Profound analyzed more than 281,000 prompt runs between July and August and found that ChatGPT's shopping answers relied almost entirely on web search through early 2026, with feed-integrated retrieval accelerating from May and overtaking web search overall in August. The same report puts Shopify at about 35% of feed-integrated retrieval. This is measurement of outputs, which suggests a rapid move toward structured merchant data without proving how the system works internally. The direction is clear enough to plan around.
Where the decision moment goes
Adobe measured AI-sourced traffic to US retail sites up 393% year over year in the first quarter of 2026. By July, AI-referred visits were worth 53% more per visit than non-AI visits, against a year earlier when non-AI visits were worth 128% more. Adobe serves retailers and the base was small, so read the figures as directional. What they show is a shopper who arrives at the merchant with the decision largely made, where before that shopper arrived at a comparison site mid-decision. What comparison sites lose with the click is the decision moment.
Publishers with affiliate models felt it first. NerdWallet reported credit card revenue down 24% year over year in the first quarter of 2025, citing organic search headwinds. On Google, organic click-through on queries showing an AI Overview fell from 1.76% to 0.61%. A commission business needs the click, and the assistant answers inside the conversation.
The unresolved question is transaction ownership
OpenAI's Instant Checkout, launched in September 2025, was retired after roughly five months, with around thirty Shopify merchants integrated and sales close to zero. Purchases moved into retailer apps inside ChatGPT. Consumers remain cautious as well: 73% use AI in shopping journeys, 14% trust it for autonomous purchasing, and 89% double-check what it tells them. Commerce does not need full autonomy to change, though. The highest-value moment in a purchase sits upstream of payment, at the recommendation that decides what reaches the shortlist.
From winning the click to winning inclusion
A comparison site competed for traffic. A brand now competes at four AI decision gates: inclusion in the shortlist, a positive recommendation, survival when the shopper raises objections, and final selection. Each gate is decided inside the assistant, out of view of the brand's analytics. Comparison does not disappear. It becomes an invisible layer inside AI decision-making.
Why the pressure grows
Three forces widen the gap.
Advertising comes first. ChatGPT ads launched in the US on February 9, 2026, and product-feed ads followed in June, using the same Google Shopping-format catalog merchants already maintain. Organic ranking is currently not influenced by paid placement, but a sponsored card beneath the answer recreates the comparison site's revenue model inside the assistant.
Triangulation is second. ChatGPT weighs feed data against reviews, editorial roundups and comparison articles. A product supported by independent "best X for Y" coverage gains another signal that can reinforce its representation inside the assistant. Comparison sites are feeding the system that is reducing their traffic.
The personal agent is third. Today the assistant belongs to the platform. The next step is the consumer's own agent, holding standing preferences and a budget and reading merchant feeds through open protocols. The timeline is uncertain, but the direction is consistent. Google's Universal Commerce Protocol covers discovery through post-sale, and its council now includes Amazon, Meta, Microsoft, Salesforce and Stripe. Morgan Stanley expects nearly half of online shoppers to use AI shopping agents by 2030, accounting for roughly 25% of their spending. A personal agent has no reason to view an ad or accept an affiliate ranking. It reads the feed, checks reviews and returns a shortlist that serves one person. In Europe, 63% of shoppers already use AI to compare brands, models, prices and reviews, so the ground is prepared before the feed arrives.
Is there another use case?
Yes, for businesses that own something an agent cannot get from a feed: independent testing, verified reviews, long price histories, trust signals. Comparison businesses that license those inputs to agents have a future. The wider lesson for traditional web businesses is to be the source the agent cites or the page it skips.
What follows for brands
The next generation of commerce measurement will ask whether AI systems selected your brand when consumers asked what they should buy, against which competitors, and how often. Whether consumers found your page will become a weaker measure of commercial success. That is a recommendation-share question, and it will decide more revenue than search ranking within this planning cycle.
Sources: Lengow, ChatGPT feed spec · Darkroom, How ChatGPT Shopping Works · Profound findings via Elsop · Idealo · Adobe, AI traffic · Spangle, State of AI Shopping · Scrunch via AOL · ALM Corp · Red Van Workshop · Elogic · Passionfruit, ChatGPT ads · Tru Commerce · Fin, agentic commerce · commercetools trends · commercetools stats
Comments ()