Google tests AI Overviews clicks: the finding that changes everything for e-commerce
Summarize this article with AI
38% fewer clicks: a client calls me on a Tuesday morning
An online retailer. 1,200 SKUs in electronics. Investing €3,200 per month in content for 18 months. Organic traffic: 12,000 sessions in January 2026. By March, 7,440. That’s –38%. He doesn’t understand.
I pull up his Search Console data. Out of 47 main queries, 31 display an AI Overview above the blue results. Coincidence? I don’t believe in coincidences.
That’s why I apply the DOSE framework—Diagnose, Optimize, Structure, Evaluate—taught by Guillaume Attias at BMO Academy. First, understand the real impact of AI Overviews on e-commerce clicks.
A randomized study conducted by the Indian School of Business and Carnegie Mellon University, published in April 2026, tested AI Overviews’ impact on 1,065 American participants. Result: a 38% drop in organic clicks when AI Overviews were present, and a 33% increase in zero-click searches. Interestingly: removing AI Overviews didn’t affect user satisfaction, perceived quality, or ease of finding information. Users aren’t happier—but sites lose clicks.
Liz Reid, Google’s Search lead, reiterated on Bloomberg’s Odd Lots podcast that these lost clicks are « bounce clicks »—quick visits where someone grabs a fact and leaves. According to her, AI Overviews eliminate these valueless visits. The problem: Google has never provided data to back this claim. It’s a narrative, not a demonstration.
Among e-commerce sites I observe each week, the effect is brutal. The first step is a clinical diagnosis.
Diagnose: why Google talks about « bounce clicks » and what the term hides
An e-commerce bounce click isn’t a simple round trip. It’s a visitor checking a price, comparing two models, reading a critical review in 10 seconds, then leaving. That click may not lead to an immediate purchase, but it feeds the decision. Liz Reid’s argument, reported by Search Engine Journal, falls flat for a retail site.
The ISB-CMU study confirms that 38% of organic clicks disappear when an AI Overview is present. In practice, that means on a query like « best quiet cordless vacuum 2026, » the top natural result loses an average of 20 to 35 CTR points. I see it on my clients’ Search Console dashboards: click curves plunge the moment AI Overviews activate.
38% of organic clicks lost according to the field study (Indian School of Business & Carnegie Mellon). 33% more zero-click searches. Removing AI Overviews doesn’t improve user satisfaction.
How do you diagnose whether your store is affected? Go to Search Console, filter for queries containing « best, » « comparison, » « guide, » « how to choose. » Track CTR trends since January 2026. If the average has dropped more than 10 points and clicks have fallen while impressions stayed flat or rose, you’re in the thick of it. List those queries and their associated pages. That’s step 1 of DOSE—raw diagnosis, no sugar coating. Once you’ve confirmed this, you move to optimization.
Optimize: revenue climbs, but traffic plummets—where’s the opportunity?
On April 23, 2026, Alphabet reported Q1 results. Search advertising revenue jumped 12% year-over-year. Microsoft announced double-digit growth in its own ad revenue. Advertisers keep investing heavily. Yet organic traffic is melting. Where’s the money going? Into platform pockets, not into merchant clicks.
The good news: AI Overviews aren’t walls. They cite sources. On average, an AI Overview references 3 to 5 sites. If your content is structured, clear, and authoritative, you can become one of those sources. And clicks arriving from those citations, even if sparse, are often highly qualified clicks.
Optimization starts with on-page fundamentals:
- Titles and meta descriptions that directly answer the query, with natural bullet lists.
- Concise content: answers in 60 to 90 words under each subheading, ideal for automatic extraction by Google.
- schema.org markup: Product and FAQ are essential for appearing in rich results, and Google uses them to build AI Overviews.
I observed a home décor client increase organic clicks by 12% since implementing these optimizations on 50 priority product sheets, without changing a single line of heavy code. Overall traffic continued falling 9%, but organic conversion rate climbed 14%. Fewer visits, more value. The goal isn’t volume—it’s relevance.
Structure: don’t fight AI Overviews, build the architecture that captures clicks
Facing cannibalization, the temptation is to produce even more content. Wrong move. What you need is an armature. A semantic silo is a set of interconnected pages organized by topic logic, covering all facets of a subject. It’s the method I’ve applied since 2016, taught at BMO Academy.
When Google needs to generate an AI Overview for a complex query like « how to maintain my electric bike, » it scans the web for authority signals and comprehensive coverage. A site offering a pillar page « electric bike maintenance » with 8 child articles on niche topics (battery, drivetrain, tires, brakes) and seamless internal linking has far higher odds of being cited than 15 isolated product sheets.
Let’s take a concrete case. An organic food e-commerce retailer: 300 product sheets, zero guide content. Traffic down 24% in 3 months. We built a silo around « vegetable oils » with 7 core queries and 22 article pages. Result after 4 months: 19% of silo queries display an AI Overview citing the site. Average CTR for silo pages rises from 1.8% to 4.2%. Qualified traffic from those pages jumps 63%. Overall site traffic stays 9% below pre-AI Overview levels, but organic revenue stabilized. The structure absorbed the shock.
Structure transforms your site into a citation magnet. Where Google wants to summarize, you become the natural source.
Evaluate: forget total traffic, focus on visit value
With AI Overviews, organic traffic is no longer a reliable gauge. The metric that matters is revenue per organic session. I measured this with a fashion client: overall traffic down 28%, but SEO revenue dropped only 6%, because organic conversion rate climbed from 2.1% to 3.7%. Visitors arriving after ignoring an AI Overview, or from a citation in the box, are buyers already far along in their decision.
To evaluate real impact, segment in Google Analytics 4 landing pages matching URLs cited by AI Overviews. Use referral parameters if you’ve enabled tracking, or more simply, isolate sessions whose landing page is a guide or pillar page, and compare their conversion rate to standard product pages. You’ll often see a gap of 1.5 to 2 times higher.
Next, track queries in Search Console by separating those with an AI Overview from those without. CTR is no longer the sole gauge. Look at click count and estimated conversion rate. If a page loses 30% of clicks but maintains or increases revenue per click, you’re on the right path. Évaluation, the fourth pillar of DOSE, pulls you out of volume obsession and into profitability logic. What’s the point mourning 1,000 lost sessions if 200 sessions generate more?
Act: 3 concrete actions for your store before next quarter
You won’t find magic formulas here. You’ll find an action plan.
- Audit your 20 most vulnerable queries. In Search Console, extract queries that lost the most clicks since January 2026. Check for each whether an AI Overview is present. List associated pages. That’s your immediate dashboard.
- Optimize content to be cited. For each vulnerable page, add a Q&A block with FAQ schema. Write direct answers in under 60 words. Include bullet lists and comparison tables. Google loves digestible structures for AI Overviews.
- Build a semantic silo on your most profitable theme. Identify 5 to 10 informational queries around your core range, create a pillar page, and link your existing content in a silo. Don’t create 50 pages: intelligently structure 12 to 15 pages in a silo. Let architecture do the talking.
These three actions cost zero in ad spend. They demand rigor and architectural vision. That’s exactly what I deliver to my clients in a 30-minute live audit.
How many of your pages survive the AI Overviews filter?
Your free 30-minute audit
I review live with you the 15 pages losing the most clicks since AI Overviews arrived. You leave with a concrete action plan.
Book a strategic call — 45 minFrequently Asked Questions
What exactly is an AI Overview?
AI Overviews (formerly SGE) are answers generated by Google’s AI, displayed at the top of search results. They synthesize multiple sources and answer the query directly, reducing the need to click a traditional blue link.
My traffic has dropped since January 2026. Is it AI Overviews?
Check Search Console for queries that lost the most CTR. If they’re informational or comparative and an AI Overview appears for those terms, that’s likely the cause. Auditing your top 50 queries confirms this.
How do I optimize content to be cited by AI Overviews?
Favor clear structure with lists, short answers (60 to 90 words), FAQ and Product schema markup, and demonstrated topic authority through semantic silos.
Are semantic silos still effective against AI Overviews?
Yes. A well-built silo raises the odds of being an AI Overview source because it shows complete expertise on a topic. I observe my clients get more citations after structuring.
Should I worry about my revenue?
Not necessarily. Overall traffic may fall, but remaining traffic can convert better. Measure revenue per organic session to guide strategy rather than click count.
