We've heard this was coming, and now it's here. VersaFeed can confirm that AI Performance reporting is officially live for a small pool of retailers inside Google Merchant Center (Analytics > Products > AI performance). It likely helps if you are submitting AI conversational attributes to Google but this doesn't seem like a requirement. If you've got access, congrats! You've got an early look at how Google wants you to think about AI Overviews and AI Mode. If you don't have it yet, here's what to expect.
The headline metric for the AI performance insights is "Share of Voice," and it's worth pausing on because Google is using the term differently than you're used to. In the past, share of voice has referred to impression share as of all the impressions you were eligible for, how many did you actually get? Here, it means something closer to market share: of the total AI impressions captured by you and the competitors Google has selected, what percentage was yours? Same phrase, different math. Don't assume you know what you're looking at just because you've seen "share of voice" a hundred times before.
What we do like is that Google breaks this out by funnel stage, something we wish they'd do everywhere, not just AI:
- Discovery : broad, feature or style driven searches. Shows query frequency and your share of voice.
- Evaluation : comparison shopping, reviews, spec-checking. Same two stats.
- Purchases : not literal purchases, just purchase-intent queries, split into local availability ("near me" searches), ready-to-buy (specific SKUs, sizes, models), and pricing (deals, discounts, "cheap X").
Beyond the benchmarking, the report actually gives you something actionable: it flags product attributes that are missing from your descriptions. It's not naming the exact fields, it's telling you the values aren't showing up in the text Google can read. Nine times out of ten, this happens on variant products, where a parent listing has five sizes and three colors but one generic description covering all of them. It's a quick fix, and it's exactly the kind of gap VersaFeed closes by hand-tailoring descriptions per variant instead of relying on one-size-fits-all copy.
There's also a query term report showing the language shoppers are actually using in AI search. Think of this as a search terms report for the AI era. If your titles, descriptions, and product highlights aren't speaking that language, you're leaving visibility on the table, and it's something our AI enhancement tooling is built to catch and fix.
One caveat: everything here is filterable by time period, country, category (seems like like this is related to higher-level GPCs), and traffic type, but right now it's limited strictly to organic AI traffic within AI Experiences. No paid AI Mode data yet, and it's still a limited rollout. That said, if Google is building out this level of granularity for a beta feature, it's a strong signal AI-driven shopping surfaces aren't a side project anymore, they're becoming a real channel worth optimizing for and the retailers paying attention now will have a head start when this rolls out more broadly.
At VersaFeed, we are serious about AI so if you have concerns about where you are at here, let's connect.