Google AI Mode Sends 57% of Product Links to Google Shopping

Rixot Editorial Team•November 9, 2026•10 min read

A funnel of product links curving back into a Google Shopping grid while a few links escape to outside publisher sites

When shoppers ask Google's AI Mode what to buy, most of the links it shows lead right back to Google. A new study of 55,198 AI Mode answers found that 57% of product links point to Google Shopping pages, and 46.5% of U.S. product answers link only to Google's own properties. For e-commerce, that quietly rewrites where a backlink has to live to matter.

Summary

57% of product links in Google AI Mode go to Google Shopping product pages, measured link by link.
46.5% of U.S. product answers link only to Google-owned pages — a separate, per-answer measure, not the same number as the 57%.
The data comes from Findrix, which analyzed 55,198 AI Mode answers to 18,400 buyer questions across 35 U.S. markets (14–17 October 2026), as reported by Search Engine Land on 29 October 2026.
By category, the share ran from 53.8% (activewear) to 60.4% (swimwear).
Other engines behaved differently: none of Gemini's measured links went to google.com, and ChatGPT showed ~25 links per answer versus roughly three for AI Mode.
The takeaway for link building: when Google keeps product clicks inside its own ecosystem, off-Google authority — editorial links and brand mentions on real, relevant publisher sites — becomes the path back into AI answers.

What the study actually measured

Findrix ran 18,400 buyer-style questions through Google's AI Mode across 35 U.S. markets over four days in mid-October 2026, capturing 55,198 answers. Crucially, it measured the links shown in the answers — not clicks, not purchases, and not rankings. So the headline is about where AI Mode chooses to point, which is the part link builders can actually influence.

Two numbers get quoted together and shouldn't be. The 57% is counted link by link: of every product link AI Mode displayed, 57% went to a Google Shopping product page. The 46.5% is counted answer by answer: in 46.5% of U.S. product answers, every link stayed on a Google-owned page. One describes the mix of links; the other describes how often the answer is a closed loop. Reporting them as one figure overstates the effect.

Why the category spread matters

The Google-Shopping share was not uniform. Swimwear topped out at 60.4% of answers, women's everyday fashion at 57%, sun care at 54.1% and activewear at 53.8%. Visual, catalog-heavy categories — where Google already has a deep Shopping feed — lean hardest on Google's own pages. Categories where buyers want comparison, context or expertise leave more room for outside sources, and that room is where an editorial link can still be seen.

Frequently asked questions

What is Google AI Mode?

AI Mode is Google's conversational search surface: instead of a list of ten blue links, it returns a generated answer with a small number of cited links. For product queries it behaves like a shopping assistant, summarizing options and linking out to where it thinks you should go next.

What does "57% of product links go to Google Shopping" mean?

Counted link by link across 55,198 answers, 57% of the product links AI Mode displayed pointed to a Google Shopping product page rather than to a retailer's or publisher's own site. It is a measure of where the answer sends attention, not of what shoppers ultimately clicked or bought.

How is that different from the 46.5% figure?

The 46.5% is a per-answer measure: in nearly half of U.S. product answers, every single link stayed inside Google's own pages. The 57% is a per-link measure across all answers. They describe related but distinct things, so they should not be added together or used interchangeably.

Why does this matter for backlinks?

If classic product-page clicks from search keep shrinking because AI Mode routes them into Google Shopping, ranking a product page is no longer the whole game. Getting cited inside the AI answer — and building the off-Google authority that earns those citations — becomes the way to stay visible. That authority is built with relevant editorial links and brand mentions, not with another Shopping listing.

How is AI Mode different from ChatGPT or Gemini for links?

They cite very differently. AI Mode showed roughly three links per answer and leaned heavily on Google's own pages; ChatGPT averaged about 25 links per answer; and none of Gemini's measured links pointed to google.com at all. A link profile that only optimizes for one engine misses where the others actually send traffic.

Do backlinks still help if AI keeps clicks inside Google?

Yes — arguably more. AI engines ground their answers in sources they trust, and that trust is still built on who links to and mentions you across the open web. A relevant, in-content link on a real-traffic publisher is exactly the kind of signal that puts a brand into the pool of sources an AI answer can draw from.

What is the difference between a product listing and an editorial backlink here?

A Shopping listing lives inside Google and competes on feed data and price. An editorial backlink lives on an independent site, in context, and signals authority to both classic search and AI engines. When the listing surface is crowded with Google's own links, the editorial signal is what still comes from outside the walled garden.

How should an e-commerce brand respond to this study?

Treat it as a directional signal, not a settled benchmark — the authors kept their method brief. Then diversify away from Google-owned surfaces: earn relevant editorial coverage, brand mentions and in-content links on publishers your buyers actually read, and track citations in AI answers alongside rankings.

Why this is a backlink story, not just an e-commerce one

It is tempting to file this under "another zero-click scare." But the mechanism is specific and it points straight at link strategy. When AI Mode answers a product question, it is choosing a handful of sources to trust and surface. For 57% of product links, that choice is Google's own Shopping page. That is Google keeping the transaction inside its ecosystem — and it means a retailer's own domain is increasingly not the thing being linked.

So the question shifts from "how do I rank my product page?" to "how do I become a source the AI answer pulls from?" The answer looks a lot like old-fashioned authority: be cited, in context, on sites with real readers and real topical relevance. Those citations are what AI engines use to decide who is credible, and they are earned with editorial links and mentions, not with feed optimization.

Google is a walled garden; the open web is the moat

The 46.5% of answers that link only to Google pages describe a closed loop you cannot win from inside. You cannot out-Google Google on its own Shopping surface. What you can do is build presence on the open web — the independent publishers, niche blogs and trade sites that cover your category — so that when any engine looks for a credible outside source, your brand is already there.

Different engines, different link math

The cross-engine numbers make diversification concrete. Gemini pointed none of its measured links at google.com, and ChatGPT showed about 25 links per answer against AI Mode's three. An off-Google link that AI Mode ignores may be exactly what ChatGPT cites. Building authority across relevant, real publishers hedges against any single engine's link policy.

How product discovery is being re-routed

Step back and the pattern is bigger than one study. Classic search sent a shopper from a query to ten links, most of them to retailers and publishers. AI Mode compresses that to about three links per answer, and well over half of them loop back into Google Shopping. The funnel didn't disappear — it moved inside Google, where the exit doors are Shopping listings rather than independent sites.

That changes the economics of an e-commerce click. A retailer used to compete for a top organic position and win the visit directly. Now, for a large share of product questions, the "win" is a slot in Google's own Shopping grid — a surface governed by feed quality, price competitiveness and ads, not by the editorial authority that search once rewarded. The open-web click a strong backlink profile helped you earn is simply offered less often.

None of this is a reason to abandon a well-optimized product page — it still matters for the queries AI Mode does send outward, and for classic search. But it is a strong reason to stop treating the product page as the finish line. The brands that stay visible as answer engines mature will be the ones that also exist, credibly, off Google: quoted in a category roundup, linked from a review, mentioned by a publication a buyer already trusts. Those are the signals that survive whichever way a given engine decides to route its links next quarter.

Where the outside links still are

The same study shows the escape routes. Just over half of answers were not closed Google loops; the rest surfaced at least one outside source. Categories built on comparison and expertise — not just catalog images — left more of that room. And other engines leave far more of it: ChatGPT's roughly 25 links per answer and Gemini's complete absence of google.com links mean the open web is still where most AI citations point. A brand with relevant editorial links is eligible for those slots; a brand with only a Shopping feed is not.

Why authority, not feed data, wins the citation

AI answers are grounded in sources the model treats as credible, and credibility on the open web is still a function of who links to and writes about you. That is why a topically relevant, in-content link on a real publisher does something a Shopping listing cannot: it tells every engine — Google's AI Mode, ChatGPT, Gemini — that an independent, trusted site vouched for you. When the paid, Google-owned surface is saturated, that independent signal is the differentiator, and it is the one piece of the picture a competitor cannot buy with a better price feed alone.

What to do about it

For brands that sell things, the practical response is to stop treating search visibility as a product-page ranking problem and start treating it as an authority problem:

Earn relevant editorial links on real-traffic publishers in your category, so AI engines have an outside source to cite. You can browse and filter donor sites by niche, authority and traffic in the Rixot marketplace.
Pair the link with a brand mention in the same article — AI answers increasingly surface named brands, not just URLs.
Diversify across engines. Don't optimize only for Google AI Mode; ChatGPT and Gemini cite the open web far more, so off-Google links do double duty.
Measure citations, not just rankings. Track whether your brand appears as a source in AI answers for your key product questions, and treat that as a KPI next to position.
Prioritize relevance over raw DR. A topically matched placement on a site your buyers read is worth more to an AI engine than a generic high-authority link; our per-link pricing is built around that trade-off.

What this does not mean

It does not mean product pages are dead or that Google will keep every click forever — AI Mode is young and its link behavior is changing month to month. It also does not mean the 57% is a hard law: Findrix published limited methodology, so treat it as a strong directional reading rather than a fixed benchmark. The durable lesson is structural, not numerical: as answer engines decide what to link, the brands with real off-Google authority are the ones that keep showing up. Expect the exact percentages to move as Google tunes AI Mode and as rivals like ChatGPT and Gemini keep linking out more aggressively. What will not move is the underlying rule — engines cite sources they trust, and trust on the open web is still earned with relevant links and mentions. Build that, and you stay visible no matter which way the next product answer decides to point.