The AI DownsideDocumenting AI's downsides

AI Search

Google's AI Mode Shows You Pricier Products Than Search Does

A new study says the identical item costs 21.6% more in AI Mode. The exact figure is arguable; the price comparison it quietly removes is not.

Editorial illustration for “Google's AI Mode Shows You Pricier Products Than Search Does”.

Ask Google what to buy and, more and more, it doesn’t hand you a page of results to weigh up. It hands you an answer. AI Mode names a product, shows a price, and moves on, the way a confident shop assistant might if the shop assistant were also the shop. It feels efficient. A new study says it may also cost you more.

Productrise, a firm that tracks Google Shopping listings, compared the prices Google’s AI Mode shows against the prices in ordinary search. Its headline claim: when the very same product turns up in both, AI Mode’s price is 21.6% higher on average. That number has caveats stacked on it — and we’ll get to every one of them, because they matter — but the finding underneath is worth sitting with.

The interesting part isn’t whether Google is quietly marking things up. On the evidence, it almost certainly isn’t. The interesting part is that the AI shopping interface removes the one thing that normally keeps a price honest: your ability to see what everyone else is charging.

The finding, and the asterisks on it

The study ran for 23 days, from 9 to 31 August 2026, across the United States and the United Kingdom, tracking more than two million product listings and over a hundred thousand results pages. On matched products — the identical item, same query, same day — AI Mode came out 21.6% dearer on average. Prices actually differed on 38.1% of those matched products, and when they did, AI Mode was the pricier one 68.4% of the time.

There is a bigger, scarier number floating around: across all products, AI Mode’s median price was $149 against $100 in traditional search, roughly 49% higher. Treat that one with suspicion. Only 1.28% of the products in ordinary search also appeared in AI Mode, so the wider comparison is mostly measuring two different shopping baskets, not the same goods at two prices. If AI Mode tends to surface the premium tier of a category, its median rises without anyone charging you a penny more for a given item. The defensible figure is the matched one, and even that is an average of a minority of cases where the prices diverge at all.

Why the same product drifts pricier

Google’s explanation, given to Futurism, is the obvious one and probably the right one: AI Mode and the search results page “are powered by the same data source, Google Shopping Graph.” If both surfaces read from the same catalogue, a systematic markup would be very strange. So the price gap, where it exists, is almost certainly not a different price — it’s a different choice.

Here is the mechanism the demo skips. Traditional search shows you a carousel: in the study, an average of 27.8 products per query, a grid of competing sellers for the same thing. AI Mode shows about 3.9. To get from twenty-eight offers down to four, something has to pick, and that picker has its own logic — it might favour the manufacturer’s own listing over a cheaper third-party reseller, or the seller with the tidiest structured data over the one with the keenest price. None of that requires malice. It just requires an algorithm optimising for a “good answer” rather than the lowest total cost, and a single lead offer standing in for a market.

The comparison you can no longer see

This is the actual consumer harm, and it survives every caveat the study attracts. Comparison shopping is what disciplines prices. When you can see fifteen sellers of the same drill in a grid, the dear ones are obvious and you scroll past them. When an AI hands you one drill and a price, you have nothing to scroll past. The high price isn’t hidden exactly — it’s just presented without the context that would tell you it’s high.

Google’s rebuttal is that “shoppers can easily click into a product listing to compare prices for that product across retailers.” That’s true, and it’s the right instinct. But it quietly concedes the point: the comparison now lives one click deeper, behind an answer that is designed to feel complete. Defaults are powerful precisely because most people accept them. An interface that shows you one option and buries the price grid a tap away will, on average, sell more of the option it shows.

The problem isn’t that the AI charges you more. It’s that it shows you one price and takes away the twenty-seven others you’d have used to know it was too high.

To be fair: what the study doesn’t show

The case for scepticism is genuine, and we’d be doing exactly the reverse-hype we complain about if we skipped it. Start with the source. Productrise sells rank-tracking, feed audits and feed optimisation — tools whose entire pitch is helping merchants rank better in Google Shopping. A study concluding that Google’s AI shopping surface behaves badly is not a neutral bystander’s finding, and that belongs on the label.

Then the method. MediaPost noted that the wider price gaps are inflated by outliers, in one case a used item in ordinary search matched against a new one in AI Mode — not the same product at two prices, but two different products. PPC Land flagged an unexplained discrepancy in the disagreement rates that Productrise didn’t reconcile. And commenters poking at the examples found the usual real-world noise: a listing that looked $5 cheaper until a delivery charge closed the gap, or a manufacturer’s page that costs more but is less likely to be a knock-off than the bargain reseller beneath it. Cheaper is not always better, and “lead offer” is not always “worse deal.”

Finally, Google says it hasn’t verified the claims, which is fair — it’s a third party’s dataset, not Google’s own. Take all of that together and the honest reading is narrow: not “Google overcharges through AI,” but “an interested party’s large study found that AI Mode tends to surface pricier options and far fewer of them, and Google hasn’t shown otherwise.”

Which leaves one fair question pointing the other way. A study from a company that sells ranking tools is easy to wave off, and Google has effectively waved it off. But Google is the one party that could settle the matter outright: it holds the Shopping Graph, the AI Mode logs and the search logs, and could publish its own matched-price comparison whenever it liked. Until it does, “we haven’t verified the claims” is not the same thing as “the claims are wrong” — and the party best placed to check is the one saying least.

Why this is more than a shopping quirk

Even at its most cautious, the finding lands on a nerve, because it isn’t really about drills and delivery fees. It’s about what happens when an answer engine replaces a results page. We’ve argued before that AI search is making Google worse and that Google keeps nudging the plain search button out of reach. Shopping is the same story with a price tag attached: the more Google answers instead of listing, the more it decides on your behalf, and the less you can check its working.

That shift has a cost the launch videos never mention. An AI Overview that summarises ten articles at least lets you click through to the ten; a shopping answer that names one product from a field of twenty-eight has quietly made the decision comparison shopping was supposed to be yours. It is the retail version of what AI Overviews did to the open web — convenience up front, and a narrower set of choices you never quite see being narrowed.

What to do until it’s clearer

None of this means abandoning Google, and it certainly doesn’t mean trusting a single study to the decimal point. It means shopping like someone who knows the answer in front of them is one option, not the option:

  • Do price-sensitive shopping in the grid, not the answer. Drop into traditional search results or the Shopping tab, where you still get the full list of sellers. For anything above pocket-money money, that comparison is worth the extra tap.
  • Compare the total, including delivery. A headline price that looks lower can lose once postage and fees are added — and it can also win. The point is to check, not to assume the AI’s pick did the checking for you.
  • Treat the single result as a starting point. Take the product AI Mode names, then search that exact model across a couple of retailers or a comparison site. You’re not distrusting the recommendation, just refusing to let it be the whole search.
  • Notice when you’re being pushed into AI Mode. For some queries a plain results page is getting hard to reach. When you can still choose the ordinary listing, and price matters, choose it.

The tidy version of the future Google is selling is one where you ask and it answers. The study is a useful reminder that an answer is a choice someone else made, and that when the choosing happens out of sight, the thing most likely to quietly slip is the price. The fix isn’t to fear the tool. It’s to keep asking to see the other twenty-seven.

Frequently asked questions

Does Google's AI Mode really show higher prices than normal search?

According to a September 2026 study by the firm Productrise, yes, on average: when the identical product appeared in both AI Mode and traditional search on the same day, AI Mode's price was 21.6% higher. But prices only differed on about 38% of matched products, the firm sells shopping-ranking tools, and Google has not confirmed the finding, so treat the exact figure as a claim worth checking rather than a settled fact.

Why would the same product cost more in AI Mode?

The study doesn't prove Google marks anything up, and Google says both surfaces use the same underlying Shopping Graph data. The likelier mechanism is selection. AI Mode picks a single 'lead offer' per product from far fewer listings, and if it favours the manufacturer's own page or one particular seller over the cheapest third-party reseller, the price you see drifts upward even though the underlying data is identical.

Is the 21.6% figure trustworthy?

Cautiously. It is a large sample and several outlets reported it, but it comes from a company that sells Google Shopping visibility tools, the wider '49% higher' figure compares different products rather than matched ones, and trade coverage flagged both outlier listings and an unexplained gap in the disagreement rates. It is a strong prompt to check for yourself, not a verdict.

How do I avoid overpaying because of AI Mode?

For anything price-sensitive, run the search in traditional Google, the Shopping tab, or a dedicated comparison site, where you still get the full grid of sellers. Compare the total price including delivery, since a listing that looks cheaper can lose once postage is added. And don't assume the single result AI Mode offers is the best deal — it is one option, chosen by a system that shows you far fewer of them.

Can I turn AI Mode off?

Not reliably, and that is part of the complaint. Google increasingly makes AI Mode or AI Overviews the default for many queries, and for some searches it is close to the only surface offered, so 'just use normal search' is getting harder. Where a plain results page is still available, it remains the better place to compare prices.

Sources

  1. Google AI Mode shows the same products 21.6% more expensive than traditional search [Data Study] — Productrise (Hugo Huijer); 2M+ listings, 100k+ responses, 9-31 Aug 2026, US + UKProductrise
  2. ‘Google's AI Mode Is Ripping Shoppers Off With Vastly Higher Prices, Study Claims’ — carries Google's response: same Shopping Graph data, ‘shoppers can easily click into a product listing to compare prices’, claims not yet verifiedFuturism
  3. AI Mode shows the same products 21.6% more expensive, Productrise finds — methodology and an unexplained disagreement-rate discrepancyPPC Land
  4. Google AI Mode Prices Differ From Search For Same Items — notes 3.9 vs 27.8 products per response and outlier used-vs-new matching inflating the wider gapMediaPost
  5. Google AI Mode Prices Differ From Product Carousel For Same ItemsSearch Engine Journal
  6. Ask HN discussion — ‘Google AI Mode shows same products 21.6% more expensive than traditional search’ (4 Sep 2026): delivery-fee confounds, manufacturer-page vs cheaper reseller, Google phasing out plain searchHacker News

Related grievances

All articles →