Last week, a productrise.app study landed on the Hacker News front page with an uncomfortable finding: for the same queries, products recommended by Google AI Mode averaged 21.6% more expensive than the top results from traditional Google Search.
The question isn't just whether AI Mode is recommending the "right" product. It's whether the algorithm has learned to favor what's most profitable over what's most useful.
How the 21.6% Was Measured
productrise.app's methodology isn't complex, but a few details are worth unpacking:
- Sample: Primarily electronics queries (headphones, keyboards, monitors, etc.), with the same query set run through both traditional Google Search and AI Mode.
- Comparison method: Price comparison between the top result (or first item in the recommendation list) from each channel.
- Finding: The average price of AI Mode's top recommendation was 21.6% higher than traditional search's top result.
Worth applying some skepticism here: the original post didn't disclose exact sample size and was concentrated in electronics. Nobody has reproduced this across all product categories. That uncertainty matters for what follows.
It's Not About the 21.6% — It's About Ranking Position
"21.6% more expensive" can be read two ways:
- AI Mode's recommendations include some expensive items, and they drag the average up by 21.6%.
- AI Mode systematically elevates pricier products to the top of its recommendation list.
The productrise.app data points to interpretation #2 — a ranking position bias.
In other words, expensive products aren't just "present" in AI Mode's answers. They're being actively pushed to the first or second position — exactly where users are most likely to click.
The distinction matters:
- If expensive items are merely mixed in, users can scroll past and compare alternatives.
- If expensive items are pinned to the top, most users' attention locks onto that answer and stays there.
That's a much uglier commercialization signal. AI Mode presents itself as giving you "an answer" — when it's actually giving you "an answer sorted by bid price."
Google's Dual Identity: Search Engine vs. Shopping Advertiser
Google has a structural conflict of interest — it's simultaneously a search engine and one of the world's largest shopping advertising platforms.
- Traditional search: Ad slots carry "Ad" or "Sponsored" labels. They're paid placements, but users can see them.
- Shopping ads: Google Shopping and Performance Max are paid-placement products at their core.
- AI Mode: Recommendations appear as a single AI-generated paragraph with no indication that any sentence might be sponsored.
The problem lives in that third bullet.
In traditional search, users know which results are paid and which are organic. AI Mode blends paid placements and organic recommendations into one seamless AI-written block. Users can't tell which sentence is the answer and which is an ad.
This effectively bakes Google Shopping's revenue model into the search "answer" itself — without telling users.
Compare that to competitors like Perplexity: they embed links in answers too, but they at least label sources and mark "sponsored" placements. Google AI Mode currently has no such mechanism.
Practical Impact for Merchants and Users
Merchants: SEO Playbook Needs a Rewrite
The last decade of e-commerce strategy can be summarized as "win SEO rankings":
- Optimize keywords
- Build backlinks
- Write review content
- Bid on Google Shopping ads to boost placement
AI Mode changes the equation. What you're fighting for is no longer "a position on the SERP" — it's the specific recommendation sentence the AI generates.
The industry is already discussing a new concept: AEO (Answer Engine Optimization), sometimes called AI Engine Optimization.
The new core objectives:
- Your product needs to be "seen" by AI training data
- Your product descriptions need to be deemed "worth recommending" by the AI
- Your pricing, inventory, and review data need to be machine-readable
There's no standard playbook yet. Whoever figures it out first captures the AI Mode traffic windfall.
Users: Cross-Check Before Buying Through AI Mode
Honestly, users don't have great options in the short term. AI Mode's output looks authoritative — a lengthy AI-written justification that most people wouldn't suspect is a sales pitch.
What you can do right now:
- For any product AI Mode recommends, cross-check on Amazon, price comparison sites, and traditional search results.
- Be suspicious when you see AI recommendations with no "Ad" label — not guaranteed advertising, but the probability has gone up.
- For anything priced higher than you expected, check historical pricing (CamelCamelCamel, Keepa, etc.).
In the medium term, either Google adds disclosure (unlikely — it would directly hit Shopping ad revenue) or regulators step in.
Editorial Take
This isn't the same fight as the last decade of SEO wars.
Traditional Google Search has long been criticized for "too many ads," but at least ad slots were visible — "Ad" and "Sponsored" labels were unambiguous. SEO was fundamentally a competition within the "free results" lane.
AI Mode breaks that structure. AI-generated recommendations are a unified "answer" — there's no boundary between "this is organic" and "this is paid." Users get deceived without ever knowing they were deceived.
What's sneakier: AI Mode's recommendation language usually reads as "objective." Phrases like "Based on reviews, features, and value, we recommend X" make users far less likely to question the output.
A few things I haven't fully resolved:
- How representative is the 21.6% number? I didn't see specific sample size from productrise.app, and the focus was on electronics. Across all categories, the number could be higher or lower. More independent replication is needed.
- Could AI Mode's training data itself be contaminated? Is it possible that Google Shopping advertiser data influenced AI Mode's preferences during training? That's a deeper question than result-ranking bias.
- Do competitors like Perplexity and Claude with Search have the same problem? In principle, any company doing "search + AI recommendation" faces this conflict. Google is just the most visible case.
My read: AI search commercialization will trigger regulatory action faster than traditional search ads ever did. Traditional search at least let users distinguish paid from organic. AI search removes that protective layer entirely.
Sources: