Google AI Mode shows same products 21.6% more expensive than traditional search

TEXT START: Over 23 days (August 9 to 31, 2026), we tracked more than 2 million product listings across more than 100,000 SERPs and AI Mode responses.

The Dissection

This is a useful observational study wrapped in a merchant-facing sales funnel.

The real finding is not simply that AI Mode displays prices 21.6% higher. It is that AI Mode is replacing broad retrieval with narrow recommendation: roughly four products instead of twenty-eight, only 1.28% overlap with traditional listings, and a different lead seller nearly half the time. The interface is compressing the market before the shopper ever sees it.

The article then performs the standard transition ritual. It converts a possible loss of consumer choice into an optimization opportunity for brands: improve feeds, enrich product data, monitor “GEO,” and buy better visibility. The platform’s growing control over demand is treated as a technical ranking problem. That conveniently turns Productrise’s tracking product into the proposed remedy.

The Core Fallacy

The article conflates an observed outcome with a proven preference mechanism. Higher prices in AI Mode show that the system selected different offers. They do not establish that Google deliberately favors expensive inventory, that price is being consciously deprioritized, or that the platform is steering users toward higher margins.

The deeper error is the optimization fallacy: assuming merchants can preserve agency by supplying better data to the gatekeeper. Under the Discontinuity Thesis, this is not merely SEO with a new interface. It is control over economic coordination migrating to whoever owns the model, data, interface, and transaction route. Merchants are being trained into servitor behavior—format yourself correctly for the Sovereign and hope the machine grants access to demand.

Hidden Assumptions

• A stable product identifier means equivalent commercial offers. It does not. Seller, condition, bundle, warranty, shipping, promotions, and new-versus-used status can materially change the price.
• The “lead offer” is a clean price comparison. The study itself shows that the lead seller changes 49.6% of the time, so the comparison often concerns different offers for the same underlying product.
• The median difference across all listings is meaningful. With only 1.28% overlap, it primarily demonstrates a different product mix, not systematic repricing of identical demand.
• A very large sample eliminates methodological uncertainty. It does not. Large numbers cannot repair selection bias, unmatched query intent, missing category controls, or absent confidence intervals.
• Same query and same moment guarantee equivalent conditions. Location, device, personalization, account state, inventory volatility, and retrieval randomness may still differ.
• Displayed price predicts consumer behavior. The article supplies no click, conversion, or shopping-path data; the claim that most users will not inspect other sellers is plausible but unmeasured.
• Better feeds and richer product data will improve selection. That is a merchant hypothesis, not a demonstrated result.
• Google’s incentive to accelerate purchases proves the direction of the system. It is a reasonable inference, not evidence supplied by this study.

The 21.6% headline is therefore sharper than the proof beneath it. The data supports “AI Mode selects different, often costlier offers,” not “AI Mode has been proven to make the same purchase more expensive.”

Social Function

Primary classification: transition management.

Secondary classifications: partial truth, commercial propaganda, prestige signaling, and ideological anesthetic.

It tells shoppers that the comparison layer is becoming less transparent, then tells merchants to adapt to the new choke point instead of challenging it. The uncomfortable structural fact—algorithmic control over what counts as a visible market—is softened into a subscription opportunity. The article does not deny the transition; it teaches its subjects how to remain useful inside it.

The Verdict

The study catches an important symptom: AI Mode is not a neutral price-comparison tool. It is a selective allocation layer that can hide cheaper alternatives, anchor users on fewer offers, and shift bargaining power toward the platform.

But the article mistakes the symptom for a ranking glitch and the power transfer for a merchant-growth opportunity. Its evidence does not prove deliberate price exploitation, yet it does show the old open-comparison model being compressed and privatized. Under DT logic, the decisive asset is not the cheapest product or the best feed. It is control of the AI gate through which buyers and sellers must pass. Google is moving toward Sovereign status; merchants are being reduced to servitors competing for machine-selected visibility.