Productrise, a platform that monitors organic product visibility in Google Shopping, said on 29 July 2026 that AI Mode surfaced roughly 95% fewer product listings than standard Google search. The figure comes from 21 days of testing in July 2026 across the United States and the United Kingdom, covering more than 100,000 identical shopping searches and over two million product listings. Productrise sells tracking for the exact surface it measured, so the number arrives with a commercial interest attached: the reporting by PPC Land is independent, but the underlying study is the vendor's own.
How Productrise ran the comparison
Productrise put each query through standard Google search and through AI Mode on the same day, so the two result sets are directly comparable. The sessions were clean and non-personalised, and the queries were product-centric rather than informational. Over 21 days in July 2026 that produced more than 100,000 query pairs across the two markets and more than two million tracked product listings.
The design matters, because AI Mode output moves from day to day and personalised sessions can inflate or suppress product results. Same-query, same-day pairing removes most of that noise. It does not remove the vendor's interest in the answer.
Two separate effects add up to the 95% figure
The first effect is whether a shopping query returns any products at all. Productrise found that standard search returned at least one product on about 88% of shopping queries, while AI Mode did so on about 23%. Roughly three-quarters of product-intent queries in AI Mode came back with no shoppable listing.
The second effect is how many products appear once any appear. Among queries that did return products, standard search showed an average of 22.5 listings per results page against 4.3 in AI Mode, which Productrise describes as roughly 5.3 times fewer competitive slots. Multiply the two effects and you get the headline: about 95% fewer product listings surfaced overall.
Two worked examples appear in the study. For the query "neck cream", standard search returned listings while AI Mode answered with quality reviews and no shoppable listing at all. The query "1/2 inch plywood" was used to show the volume gap.
Standard search against AI Mode, measured
The table below holds the four numbers Productrise published for US and UK shopping queries over 21 days in July 2026.
| Measure | Standard Google search | AI Mode |
|---|---|---|
| Shopping queries returning at least one product | about 88% | about 23% |
| Average product listings shown, when products appeared | 22.5 | 4.3 |
| Competitive slots once products appear | baseline | about 5.3x fewer |
| Products appearing on both surfaces | about 0.8% | about 0.8% |
The 0.8% overlap is the number retailers should react to
Productrise found that only about 0.8% of products overlapped between standard search and AI Mode. Ranking in a conventional Shopping carousel is therefore close to no guarantee of appearing when the same query runs through AI Mode. In practice these are two separate shelves that happen to sit in the same shop.
That is heavier news than the 95% drop. A smaller shelf can still be won. A shelf stocked from what looks like a different rulebook cannot be won with the same work.
One product feed, two different answers
PPC Land noted that Google's Shopping Graph feeds both surfaces. If a single data source produces almost entirely different result sets, the deciding factor is Google's ranking or selection logic rather than feed availability. Google has not published the selection criteria for product listings inside AI Mode, so nobody outside the company knows what clears the bar for a conversational answer.
The founder's explanation is a guess, and he says so
Productrise founder Hugo Huijer offered what he called a best guess for the shorter lists, tying it to the conversational interface. If the system expects a follow-up anyway, whether that is a request for something cheaper, a different brand or faster shipping, it does not need to put twenty products up front. Huijer also suggested that appearing among four or five recommended products carries different value than appearing among twenty, and said click-through rates from AI Mode sessions should be considerably higher.
All of that is his speculation and his expectation rather than measured data. The study counted listings and overlap, not clicks, so the higher click-through claim is a hypothesis to test in your own analytics, not a finding to plan against.
What this means for marketers in Thailand
The study covered the United States and the United Kingdom only, and Productrise made no claim about Thai search results. For a Thai retailer or exporter selling into those two markets, the useful read is narrow. Feed and product-data work built for Shopping carousels does not appear to transfer to AI Mode, and there is no disclosed way to buy around the selection logic.
Two responses hold up without guessing at Google's criteria. Measure AI Mode separately instead of assuming Shopping performance describes it. Then keep investing in the product data you control, since it is the input to both surfaces even when the outputs differ. For a Thai team that means treating conversational product discovery as its own AI-visibility problem rather than a line in a Shopping report. The commercial work underneath, whether that is ecommerce marketing or Shopify SEO, still runs on product data quality, which is the one lever here that is not speculative.
Frequently asked questions
Does AI Mode show product listings at all?
Yes, but on far fewer queries. Productrise found AI Mode returned at least one product on about 23% of shopping queries against about 88% in standard search, over 21 days of US and UK testing in July 2026.
Will my Google Shopping rankings carry over to AI Mode?
Mostly not, on this data. Only about 0.8% of products overlapped between the two surfaces, so a strong Shopping position is close to no guarantee of appearing in AI Mode for the same query.
Was Thailand included in the study?
No. Productrise tested the United States and the United Kingdom only, and the study makes no claim about Thai search results or Thai shoppers.
Are click-through rates from AI Mode really higher?
The study does not measure that. Productrise founder Hugo Huijer said he expects click-through rates from AI Mode sessions to be considerably higher, and he labelled his reasoning a best guess. The analysis counted product listings and overlap, not clicks.
Can a retailer influence what AI Mode selects?
Google has not published the selection criteria, so any tactic sold as a fix for AI Mode product visibility is untested. Productrise's overlap figure points at ranking or selection logic rather than feed gaps, since PPC Land noted that the Shopping Graph feeds both surfaces from the same data.
The short version: one surface got much smaller, and the rules that win the other one do not appear to win it. If you sell in the US or UK and your product visibility reporting still counts Google Search as a single surface, splitting AI Mode out is the first change worth making, and a good moment to review what your product data is doing in each.
