AI Performance Insights: Measuring Your Brand on Google's AI Surfaces (2026)

Adil Ziani

Ecommerce Google Ads Consultant, ex-Google, Google Partner

You cannot optimize what you do not measure

For the past year, every ecommerce client has asked me some version of the same question: "Are we visible when people shop through AI?" Until now, my honest answer was that we could infer it, but not measure it properly. Traffic from AI Mode or Gemini blended into broader buckets, and nobody could tell you their share of voice on AI surfaces.

That changed with AI Performance Insights, a new reporting feature in Merchant Center. It shows how your brand performs across Google's AI surfaces: AI Mode, Gemini and other AI-powered experiences. It is rolling out first in Australia, Canada, India, New Zealand and the U.S. If you sell in those markets, this report should be in your weekly rotation already. If you sell elsewhere, use the head start to prepare.

What the report shows and how to read it

The point of the report is separation. Instead of AI-driven activity hiding inside your overall Search and Shopping numbers, you can now see how often your products surface in AI experiences and what that presence produces. When I read it for a client, I look for three things in order.

First, presence: are we showing up on AI surfaces at all, and for which products? Second, the gap: how does our AI surface activity compare with classic Search and Shopping for the same catalog? A brand can dominate Shopping and be nearly invisible in AI Mode, and without this report you would never know. Third, the trend: AI surfaces are growing, so flat AI numbers while the channel expands means you are losing share even if nothing looks broken.

Build a simple AI visibility KPI stack

Do not drown in the data. I track four numbers per account, monthly:

• Share of AI surface impressions: impressions on AI surfaces as a percentage of total impressions. This is your headline visibility metric.
• AI-sourced clicks: raw volume and trend. Early on, trend matters more than volume.
• AI-sourced conversions and revenue: the number your CFO cares about, even while it is small.
• AI vs classic ratio: AI surface performance compared to Search and Shopping on CTR and conversion rate, to see if AI traffic is qualitatively different for your catalog.

Four numbers, one page, once a month. That is enough to spot both problems and opportunities before your competitors do.

What to do when AI visibility is weak

Weak numbers in this report are usually a data problem, not an ads problem. Here is the checklist I run:

• Feed completeness first. Missing attributes, thin descriptions and generic titles hurt more on AI surfaces than anywhere else, because the AI has to explain your product in its answers.
• Add Conversational Attributes. This related Merchant Center feature feeds the matching on AI surfaces. It captures how real people describe your products in conversation: use cases, occasions, problems solved. Fill them for your top sellers before anything else.
• Ship schema.org markup on your site. Product, Offer, AggregateRating and Review markup gives AI systems structured facts to trust and quote.
• Check availability and price accuracy. AI experiences deprioritize products whose data looks stale or inconsistent.

My monthly review routine

This is the loop I run on client accounts. First Monday of the month: pull AI Performance Insights, update the four KPIs, and note the biggest mover in both directions. Then pick one improvement action: a batch of Conversational Attributes, a feed fix on a weak category, or a schema gap on key templates. One action per month, executed properly, compounds fast. Then log the numbers so that in six months you have a real trendline instead of anecdotes.

Bottom line

AI Performance Insights turns AI visibility from a guess into a metric. The playbook is not complicated: measure your share on AI surfaces, compare it to your classic channels, fix the data gaps that hold you back, and review monthly. The brands that build this habit in 2026 will own the shelf space everyone else starts fighting for in 2027.