Track products in AI Overviews: the bridge between search and AI Mode

AI Overview tracking is now live in Productrise. We track the products Google cites or lists inside the AI Overview on a search results page, right next to the popular product carousels, deals, and knowledge panels we've tracked for months. This post explains why the AI Overview needed a different tracking approach than carousels, how we handle it, and what we can and can't capture.

I recorded a short walkthrough that shows how it works, from the Queries page through to the Product Rankings report.

AI Overviews are the bridge between regular search and AI Mode

The AI Overview shows up on a standard search results page, often at or near the top. For shopping queries it frequently names products, links out to sellers, and pulls in references.

What matters most is where it leads. When a shopper engages with the overview, Google expands it and carries them into AI Mode with the same query. AI Mode is a different ecosystem. It runs on the same Google Shopping Graph, but it can select, rank, and sort products very differently from traditional search. In a recent study we found that when the same product appears in both places, AI Mode tends to show it at a higher price and surfaces far fewer products overall. The overview sits in the middle, moving shoppers from regular search toward AI Mode, which is why knowing what's inside it matters.

Why AI Overviews require a different kind of tracking

In a product carousel, every listing follows the same shape: a title, a price, a seller, an image, and a rating, all in predictable slots. AI Mode shopping cards are structured elements we can read directly too. The AI Overview works differently.

Products in an AI Overview live inside prose. A product might appear as a link in the middle of a sentence, a name in a bulleted list, or a reference card beside a paragraph. There's no fixed template, and the seller names and prices Google shows in that text are often incomplete or missing. Reading a title, seller, and price out of free-flowing text isn't reliable, so we don't try to.

This is what one of these AI Overviews looks like on a live shopping search, with products cited inside the answer.

How we track them, and what we can't

Even though the text is unstructured, Google embeds a stable product identifier in the links and references inside the overview. That identifier is the data_docid, the same one we use to follow products across carousels and AI Mode, explained in the data_docid identifier. We pull the data_docid out of the overview, then enrich it by matching it against product data we already have for that query, using that clean title, seller, price, image, and ratings to fill in the product.

This method depends on being able to identify the product, and there are a few cases where we can't. Sometimes the data_docid isn't in our database for that query, so we have no prior data to enrich it with. Sometimes a mentioned product doesn't carry a data_docid at all, which leaves us unable to tell which exact product in Google's Shopping Graph is being referenced. And sometimes a product is named in the prose but not linked, so there's no identifier to work from. In each of those cases we skip the product rather than guess. In our data so far, roughly 40% of the AI Overview blocks we detect don't produce an enriched product row for reasons like these. We still record that the overview appeared and where it sat on the page, and this gap tends to shrink the longer you track, since more of the products Google cites will already be in your history.

Once a product is matched, it shows up in Product Rankings as its own line, tagged as an AI Overview element.

You can filter any report by SERP element type, which lets you isolate the products shown only in AI Overviews and keep them separate from your carousels, paid shopping, and knowledge panels. This works in both the individual Product Rankings report and the Query Groups report, so you can look at AI Overview visibility on its own for a single query or across a whole group of keywords.

A revamped Queries page

You can now use the revamped Queries page to filter your queries based on how often AI Overviews appear, which makes it easy to find the keywords that return these results. It shows last-7-day SERP metrics for every keyword in one view, including a new AI Overview coverage column next to your carousel stats. Sort by how often an AI Overview appears at the top of the page and the keywords where Google is already citing products rise to the top. Click any metric and you land in Product Rankings with the query, date range, and SERP element type already applied.

The new column sits right alongside the carousel metrics you already use.

AI Mode tracking is coming soon

As covered above, the AI Overview is the bridge between regular search and AI Mode. It sits on the standard results page, and the moment a shopper engages with it, Google expands it and carries them into AI Mode with the same query, often without them noticing the search type changed. That makes it the point where a shopping journey can quietly shift onto a surface that ranks products on its own terms. With AI Overview tracking live, we now cover that middle step: you can see which of your queries trigger an overview and which products Google cites inside it.

The last piece is AI Mode itself. We already store the data_docid identifiers for products appearing in AI Mode, which is what makes the enrichment above possible in the first place. Bringing AI Mode in as its own trackable surface is what we're building now, and we hope to have it ready soon. Once it's live, you'll be able to follow a single product across all three surfaces: where it ranks in regular search, whether it gets cited in the AI Overview, and how Google presents it once a shopper crosses into AI Mode.

Start tracking your AI Overview visibility

AI Overview tracking is part of your daily SERP tracking. If you're not on Productrise yet, you can create a free account and start following your rankings, prices, competitors, and AI Overviews without entering a credit card. This update is also on our changelog.