AI Overviews in GSC: How to turn every impression into clicks for your e-commerce

A client calls me. He has an €8,000 budget and one certainty: SEO is dead.

Tuesday morning. The phone rings.

Julien, founder of an online fashion brand. 370 products. 12,000 organic sessions a month, down 28% over the last six months. He’d invested €8,000 in an « SEO-optimized » redesign a year earlier. Result? Nothing.

He tells me: « Stéphane, SEO is finished. Google answers directly, people don’t click anymore. »

I ask if he’s looked at the new Generative AI reports in his Search Console. Silence. He didn’t even have access yet—the rollout wasn’t complete on his account. But I knew it was coming. Marie Haynes had just published her analysis, and I was already cross-referencing data from another client.

Three weeks later, he gets access. We pull the export. And that’s when it hits. 47 pages from his catalog were appearing regularly in the AI Overview, AI Mode, or Discover AI features. 47 pages generating thousands of impressions… and zero tracking, zero action.

We stopped the old work. We restructured. We invested the next €8,000 in the right places. +820% clicks on these pages from standard search. No paid ads. Here’s how.

Le rapport Generative AI m’a permis d’identifier 47 pages. Voici leur répartition par type.

What Google finally shows you (and what it doesn’t)

The « Generative AI » report, rolling out progressively in Search Console, is a dedicated view of your page impressions in generative AI features: AI Overviews, AI Mode, and AI versions of Discover. You filter by page, country, device, date. You see impressions, not clicks, not CTR, not ranking position.

Many complained. « What good is a report without clicks? »

For me, it’s the best report Google has shipped in years. Why? Because knowing which pages the AI deems relevant enough to include in its answers is pure gold. Here, an impression is a signal of algorithmic trust.

I observe across my clients that 60 to 70% of pages getting AI Overview impressions are not the pages performing best in standard search. They’re often guides, tech specs, buried category pages. This report reveals them.

And the absence of clicks is not a bug. In an AI Overview, the user gets their answer directly. The click, if it happens, comes after—to the source, or via follow-up search. The value lives elsewhere: in your brand association with the answer, and in the opportunity to convert those impressions into qualified traffic via standard search, once you understand why the AI picked you.

Voici le processus que j’ai appliqué pour Julien. Il se reproduit facilement sur n’importe quel e-commerce.

3-step method to turn AI impressions into clicks (and sales)

Here’s the process I built for Julien. You can replicate it for any e-commerce getting first access to the report.

Step 1: Extract, filter, map.

Export the list of pages with at least 10 AI impressions over 28 days. Keep URLs clean: no parameters, no duplicates. Sort by page type: product card, category, guide, brand page. You’ll get a matrix. Julien had 47 pages: 22 product cards, 18 catégories, 7 blog posts.

Step 2: Cross with standard search performance.

Take the same period—or better, the prior period. For each AI page, check its organic clicks, average CTR, average position. This cross-check is what makes the diagnosis sharp. I built a Google Sheet with 4 columns:

• URL
• AI Impressions (GSC Generative AI)
• Standard search clicks (GSC Search)
• Standard search CTR

Calculate a simple ratio: AI Impressions divided by Standard Search Clicks. Pages with high ratios are nuggets: Google deems them very relevant for AI, but users don’t click via the SERP. Why? Either the answer is in the overview, or the page isn’t compelling enough.

Step 3: Prioritize pages to rework by commercial potential.

Don’t scatter your efforts. Of Julien’s 47 pages, we picked the 15 tied to high-margin catégories and bestselling products. We concentrated effort there.

Golden rule: Work first on pages that solve queries with latent purchase intent: compare, price, reviews, best X for Y.

What we changed on those 15 pages: concrete, numbered examples

I spotted a pattern in Julien’s 15 priority pages. All product cards ranked in AI Overview had a comparison table of 3–5 rows that the AI lifted verbatim. Below it, nothing to justify a click: no unique details, no original visuals.

I applied 3 types of adjustments.

1. Add content the AI cannot summarize—for example, for the card « Summer linen dress—Petrol blue, » I embedded a video of the fabric’s drape in motion, filmed on a phone, with comments on the feel. The AI can’t generate that. Clicks from standard search jumped 4×.

2. Revisit structured markup. I corrected FAQ and BreadcrumbList data on catégories. I removed HowTo markup on pages where Google preferred crafting its own synthesis. Result: the AI Overview shifted from a summary to a « Popular products » block with images and prices. Direct clicks from SERP to product cards went up.

3. Strengthen internal links from AI pages to commercial pages. A guide page « How to care for washed linen » appeared in 39% of brand-related AI Overviews. I added a direct link at the top to the washed linen collection with a buy button. Sales attributed to that guide jumped 62% the next month.

Results after 4 months: standard search clicks on the 15 reworked pages climbed +820%. AI impressions stayed flat, but visitor-to-customer conversion on those pages rose 34%. Zero ad spend.

The trap agencies love (and my clients avoid)

When a new report drops, the temptation is strong: optimize for it. In other words, build content designed only to appear in the AI Overview. It’s a mistake. I say this with conviction backed by 600 audits: chasing the AI Overview for the AI Overview is like chasing featured snippets in 2018—you grab impressions with no value, and you can cannibalize your commercial pages.

A recent example: an appliance site bulked up category descriptions with detailed definitions, hoping to hit the AI Overview. They got in. AI impressions exploded +2,300%. But overall clicks dropped 18% in 3 months. Why? Because the AI delivered everything. The user had no reason to visit. And the category pages, now encyclopedic, lost their commercial punch.

The right approach—the one I always use—flips this: use the Generative AI report as a sensor, not a target. It tells you what Google values. Your job is to enrich those pages without gutting them, adding only what humans can bring. The goal isn’t to stay there; it’s to make users want to click.

5 immediate actions for your e-commerce (even without report access)

Don’t have access to the Generative AI report yet? You can act today with the data you have.

1. Find « always top 5 but rarely clicked » pages. Cross-reference your standard search positions and CTR. A CTR below 1.5% on rank #1 or #2 signals the AI Overview likely captures the click. These pages are your natural candidates.
2. Grab long-tail queries where you rank 1–3 with low CTR. You’ll often find the same intent patterns an AI Overview exploits. Adapt your pages by adding non-summarizable micro-content: comparators, dynamic pricing, video testimonials.
3. Audit your structured data. FAQ, HowTo, and Product markup deserve close attention. Overly generous markup gives the AI chunks to synthesize without linking back. Surgical markup can trigger a richer, more click-friendly display.
4. Build « hub » pages around key catégories. A page aggregating reviews, comparison tests, and real-time pricing is nearly impossible to summarize fully. Google will use it partially, but the hunger for detail drives the click.
5. Track conversion KPIs, not AI impressions. Success isn’t measured by how often the AI cites you. It’s measured by qualified visits and sales from those pages.

What I see across about 50 accounts with the report now

Stéphane Jambu

SEO & AI Engineer

I build growth systems / AI / Neuroscience | 650+ clients · 80 LinkedIn testimonials · 30 years of expertise · 15 years of systems running without me.

Follow on LinkedIn