Google Merchant Center Just Started Talking About AI Search. Are You Listening?
Google Merchant Center quietly picked up a new block of metrics — AI Performance Insights. Three of them: AI Search Intent, AI Search Terms, and AI Attributes. Sounds like just another line in the interface you can skip past. It's not. This is Google telling you, in plain text, exactly how AI search matches your products to user queries. Ignoring this is like holding a treasure map and never unfolding it.
What actually happened
Before, Merchant Center showed you the classics — clicks, impressions, positions. Now there's a data layer specifically about how AI search perceives your products. AI Search Intent shows the intent people are searching with. AI Search Terms shows the exact phrasing the model uses to match a query to your product. AI Attributes shows which product characteristics got recognized — and which are missing and dragging your visibility down.
This is an official signal from Google, not someone's guess based on indirect clues. And the signal is direct: your data feed is no longer read only by a ranking algorithm — it's read by a language model trying to figure out whether your product fits the meaning of the query, not just the set of words in the title.
Why this changes how you build a product card
For years, product card optimization was built around keywords: stuff the right phrase into the title and description, get the traffic. With AI search, the logic works a bit differently. The model evaluates intent as a whole — why does this person want this product, in what context, with what characteristics. If your description is formal and thin — "sneakers, size 9, black" — the model can't confidently say the product answers a real query like "sneakers for running on asphalt in winter." The keywords match. The meaning doesn't.
Here's the direct consequence: missing attributes aren't a minor checkbox issue — they're a real reason your product isn't surfacing in AI results. An empty "material" or "season" field isn't a technical feed gap. It's a hole in your visibility.
What to do right now
First — go into Merchant Center and look at AI Performance Insights yourself. Don't rely on my summary. Everyone's data is different: some of you have gaps in attributes, some have a mismatch between what people search and how the product is described.
Second — take the actual AI Search Terms and carefully work them into your descriptions and metadata. Not stuffing a keyword list — rewriting the text so it answers the real intent Google is showing you.
Third — close the gaps in attributes. This is the most boring work and the most underrated. Completeness of the card is literally what the AI uses to build its understanding of the product. The more honest, precise attributes you have, the more confidently the model matches your product to the query.
Fourth — make this a regular practice, not a one-time update. AI search moves fast. Intents and terms shift, and whatever closed the gap today will be stale again next quarter.
My take
It's all like sport: a single sprint doesn't work here, consistency does. The SEO world for product sites has been moving from "optimize for the keyword" to "optimize for the meaning" for a while now. Google Merchant Center just handed you a tool to see that shift with your own eyes instead of guessing from logs. Whoever starts systematically closing these gaps first takes the visibility while everyone else is still fighting over positions under the old rules. If you want to figure this out for your own project instead of guessing blind, we can talk it through in a consultation or audit.