Track products in Google's AI Mode: the full picture across every Google surface

AI Mode tracking is now live in Productrise. It's the last of Google's surfaces we hadn't covered, and it means you can follow your products across regular search and now AI Mode, from the same set of queries you already track.

AI Mode doesn't behave like a product carousel. It tends to show far fewer products, often from different sellers and at higher prices, and it frequently picks products that don't rank in the carousel at all. So how you perform in one place can tell you very little about how you perform in the other. Tracking AI Mode closes that gap.

It also gives you a complete window into Google's Shopping Graph, the same system that powers the Universal Commerce Protocol. The Shopping Graph is what Google draws on to decide which products to show and how to present them, and AI Mode is one more surface it feeds.

I recorded a short walkthrough that shows how it works, from the Queries page through to the AI Mode rankings report.

Completing the picture across Google's surfaces

For a while now, Productrise has tracked organic product carousels in regular search in the detail you actually need: positions, prices, images, titles, reviews, and how all of that moves over time. More recently we added AI Overview tracking, which covers the products Google cites inside the AI Overview on a standard results page.

The AI Overview is the bridge between regular search and AI Mode. It sits on the normal results page, and the moment a shopper engages with it, Google can carry them into AI Mode with the same query, often without them noticing the search type changed. AI Mode is where that journey lands, and it was the one surface we couldn't show you. Now the loop is closed, and you can see how a product behaves from the carousel all the way through to AI Mode.

How AI Mode tracking works

The best part is that there's nothing new to set up. Productrise tracks the same queries you already have. You don't manage one set of keywords for search and another for AI Mode. We take your existing query, run it on both regular search and AI Mode every day, and report back.

There's a good reason we keep the query identical rather than asking you to write conversational prompts. Google is pushing more of its regular searchers into AI Mode using that same query: there's a prominent AI Mode button on the search box, a tab on the results page, and, for anyone on Chrome, an entry point right in the address bar. In most cases the shopper's original keyword carries straight over, so tracking that same keyword is what reflects how people actually arrive. If you do want to test longer, conversational phrasing, you can still add those as their own queries whenever you like.

For position, we use the same approach as regular search, which you can read about in how product positions and the position score work. We look at where a product appears inside AI Mode's answer relative to the full length of the response, turn that into a depth percentage, and normalize it to a 0 to 100 scale. That gives you a position for AI Mode that's comparable to the one you already read in the carousel reports.

You get the same signals you rely on in the carousel reports too: position over time, prices, sellers, titles, reviews, and how each of those moves day by day, now for the products Google recommends in AI Mode. Here's what that looks like in the report.

One more thing worth knowing: AI Mode doesn't always return products. Sometimes it answers a shopping query with a purely informational response and no shoppable listings at all. That's part of why AI Mode coverage can sit lower than carousel coverage on the same queries, and it's useful signal in its own right.

One identifier ties a product across every surface

The reason we can connect a product across regular search, AI Overviews, and AI Mode comes down to a single identifier that Google embeds in its own results. We call it the data_docid, and it's the backbone of our per-product tracking. It's worth being clear that this is not the product ID or GTIN you submit in Merchant Center. Those are identifiers you assign in your feed. The data_docid is Google's own internal reference, generated on its side and surfaced in the results, describing how Google catalogs the product across its surfaces. The two live in completely separate systems, so it's easy to conflate them, but they're not the same thing.

What makes it valuable here is that the same data_docid can appear on a product whether it's in an organic carousel, cited in an AI Overview, or ranking in AI Mode, as long as AI Mode is pulling that product from Google's Shopping Graph. When AI Mode answers with plain text and doesn't pull from the graph, there's no data_docid to follow, so we don't force a match. But when it is there, we can stitch a product's carousel history to its AI Mode appearance and read it as one identity rather than several disconnected ones.

That cross-surface consistency is really a window into how Google's Shopping Graph works. Google appears to keep a single underlying notion of each product, and the data_docid is the thread you can follow to watch that identity surface in different places. It's the same foundation that made our recent product tracking update possible, and it's why adding AI Mode was a natural next step rather than a separate build.

What we've already learned about AI Mode

We didn't wait for the feature to launch before studying this surface. Two data studies, each drawing on more than two million product listings, shaped how we think about AI Mode.

The first looked at how many products AI Mode shows. Running the same shopping queries on both surfaces on the same day, we found AI Mode returned roughly 95% fewer product listings than regular search, and that the two surfaces overlapped on only a small fraction of products. In other words, ranking well in carousels is no guarantee you turn up in AI Mode at all.

The second looked at price. When the exact same product appeared on both sides, the AI Mode price was about 21.6% higher on average, and across every listing on each side the typical AI Mode product sat well above the typical carousel product. On top of that, the lead seller for a matched product was different roughly half the time. That study was picked up widely across the press and even prompted a response from Google.

Here's a glimpse of how far that study traveled.

The takeaway for anyone selling online is simple to state and easy to miss: what ranks in your carousel, and at what price, can be very different from what Google recommends in AI Mode. Now you can watch both, for your own products, instead of guessing.

See coverage across every surface on the Queries page

The Queries page is where all of this comes together. For each keyword you track, it shows last-7-day coverage across surfaces side by side, so you can see how often a search returns a product carousel, an AI Overview, and AI Mode products, all in one view. Those numbers can tell you very different things about the same query, since one surface might return products almost every day while another rarely does. From there you can sort to find the keywords where AI Mode is already returning products, then click straight through into the report with the query, date range, and surface already applied.

Why tracking AI Mode matters for your product data

Knowing what Google recommends in AI Mode isn't just interesting, it's the input for fixing your product data with intent. AI and agentic surfaces draw on the same product data Google already uses, much of it flowing through the Shopping Graph, so the quality of that data can help you or work against you everywhere at once.

That's also why we'd steer you away from the pitch that you can hand a tool the keys, let it rewrite every title and attribute, and win AI shopping by default. Enrichment without oversight can flatten a distinct product range into generic filler, as we argued in our piece on why agentic commerce won't save a weak product foundation. Daily visibility data, including what AI Mode is doing, is what turns enrichment into something governed: you fix the products that are underperforming, then check whether the change actually moved anything.

It matters for what's coming, too. As the Universal Commerce Protocol rollout expands, more product data keeps flowing into the Shopping Graph, and being represented well there can affect whether your products get discovered in agent-driven shopping, not just today's carousels. Tracking AI Mode is one more way to see whether you're keeping up.

Start tracking your AI Mode visibility

You can create a free Productrise account and start tracking your rankings, competitors, prices, titles, reviews, and discounts without entering a credit card. That free SERP tracking is the foundation, and it's the fastest way to see where your products currently stand.

AI Mode tracking itself is available on the Growth plan and higher, alongside the deeper reports that follow a single product across surfaces. You can compare what's included on our pricing page, and this update is also on our changelog.

Regular search and AI Mode, for the first time in one place: you can follow your products from the carousel all the way through to what Google recommends in AI Mode. That's the complete window into the Shopping Graph we set out to build.