Bounce clicks vs deep engagement: Google’s defense decrypted
Summarize this article with AI
A client calls in February
A client calls me one Tuesday in February 2026. E-commerce site. 820 SKUs. Auto parts catalog.
« Stéphane, we lost 37% of organic traffic between May and January. Google Search Console shows fewer impressions, fewer clicks. Our category pages were in positions 2-4. They’re still there. But sessions dropped. »
I look at the data. Average positions haven’t shifted. +0.2 position on average. Impressions down 28%. Clicks down 34%.
The ranking isn’t the problem. It’s the click-through rate.
I ask him: « Are you seeing AI Overviews appear on your main queries? » He says yes. On 60% of head queries that used to drive traffic, there’s now a purple block at the top of the page. Sometimes with his own products cited. But no direct link to his product sheet. Just the part name, price, sometimes an image.
The user clicks less. They get the answer in the Overview.
That’s exactly what Liz Reid calls a « bounce click. » A quick click, the user gets the info, returns to Google. Google says it removes this type of click. Not qualified visits.
Except my client was selling to these visitors. His conversion rate on those pages was 2.8%. Those clicks generated $14,000 in direct monthly revenue. Not useless bounces. Transactions.
What Google has been saying since August 2025
Liz Reid published a blog post in August 2025. She claims that the volume of organic clicks from Google Search to websites remains « relatively stable » year-over-year. She also says « quality clicks » (visits where the user doesn’t immediately return to Google) have increased.
In October 2025, she gives an interview to the Wall Street Journal. She explicitly uses the term « bounced clicks. » She says ad revenues with AI Overviews have remained « relatively stable. »
On April 23, 2026, she reappears on Bloomberg’s Odd Lots podcast. Same message. She describes « bounce clicks » as the category of clicks that AI Overviews reduce. Users who want to read longer continue clicking. Those seeking just a fact get the answer in the Overview and stop clicking.
She acknowledges fewer ad clicks for certain queries, but says increased query volume makes up for it.
Three public appearances. August, October, April. Same argument.
Zero data point.
No chart. No percentage. No year-over-year comparison. No segmentation by query type or site. According to the Search Engine Journal article published April 25, 2026, « Reid provided no data to support these claims in any of these three appearances. »
Google tracks whether users search more often, says Reid. But she doesn’t give the figures.
What independent studies show
Several independent studies have measured the impact of AI Overviews since the massive rollout in May 2025.
A BrightEdge study published in September 2025 showed that queries displaying an AI Overview had organic CTR 18 to 24% lower on average compared to the same queries without an Overview.
A Semrush study published in November 2025 observed a 15% decline in referral traffic from Google Search to a sample of 50,000 e-commerce sites between May and October 2025, with a steeper decline (27%) for sites in the « health » and « finance » sectors.
An Authoritas study published in January 2026 measured that 34% of commercial queries now display an AI Overview in US search results, versus 12% in April 2025.
These studies aren’t perfect. They’re based on samples. They don’t capture every vertical. But they converge: AI Overviews reduce CTR and referral traffic. Not just « bounce clicks. »
Google could refute these studies by publishing its own data. It doesn’t.
Bounce click vs qualified click: the line is blurry
The concept of « bounce click » seems clear. A user clicks, gets the info, returns. 5-second visit. No engagement. Google says it eliminates this type of click.
But the boundary between bounce click and qualified click isn’t sharp in e-commerce.
Real example. A user searches « Renault Clio 4 diesel oil filter price. » They see an AI Overview with three filter references, three prices, three sites. They click the first link. They land on a product sheet. They verify compatibility. They add to cart. They go back to Google to find another product (air filter). They return 20 minutes later to complete checkout.
First click: 12 seconds on site. Return to Google.
That’s a bounce click by Google’s definition. The user returned quickly. But that click generated a sale.
Another example. A user searches « magnesium deficiency symptoms. » They see an AI Overview with a symptom list. They click a supplement site. They read the full article. 3 minutes on page. They don’t order that day. They come back a week later via brand search and order.
That click isn’t a bounce. But it doesn’t convert immediately either.
Google measures « engagement » by time-to-return-to-search. But this metric doesn’t capture purchase intent, brand recall, or delayed impact.
In e-commerce, 70 to 85% of conversions happen after multiple touchpoints. A « bounce click » might be the first touchpoint. Removing it breaks the funnel.
What I observe in my e-commerce clients
I manage 47 active e-commerce projects. All tracked monthly. Since May 2025, I’ve segmented queries by whether they display an AI Overview or not.
Result: on queries with AI Overview, average CTR dropped 19% between May 2025 and March 2026. On queries without Overview, CTR dropped 3% (normal seasonal variation).
Across 12 clients in the « health and wellness » sector (supplements, natural cosmetics), organic traffic fell 30 to 45% between May 2025 and February 2026. Average positions stable or slightly up. The problem is CTR.
Across 8 clients in the « home equipment » sector (DIY, gardening), the decline is less pronounced: 12 to 18%. AI Overviews are less frequent there.
Across 5 clients in the « fashion and accessories » sector, the decline is 8 to 14%, with a partial recovery in January-February 2026 when Google reduced Overview display on certain brand queries.
These figures are orders of magnitude. Each site has its specifics. But the trend is clear: AI Overviews reduce traffic. Not just bounces.
I’m not saying all lost clicks were qualified. Some probably weren’t. But most of my clients had stable conversion rates before May 2025. If Google removed only unqualified bounces, conversion rate should have increased. It stayed flat or slightly declined (0 to -8% across sectors).
This suggests Google is also removing clicks that would have converted.
The volume compensation argument
Liz Reid says increased query volume compensates for lower CTR. Users search more often because they get quick answers.
It’s possible. But I don’t observe it.
I look at impressions in Google Search Console. Between May 2025 and March 2026, across my 47 e-commerce projects, impressions dropped 14% on average. Not increased.
That means Google shows my clients’ pages in results less often. Either because users search less (unlikely), or AI Overviews replace organic results in the visible viewport, or users scroll less.
The « users search more » argument doesn’t check out in my data.
I’m not saying this argument is false everywhere. Google might have global data showing query volume increase. But it doesn’t publish it. And my client data shows the opposite.
What you can do if you’re losing traffic
If you’re seeing organic traffic drop since May 2025, segment your queries.
Identify which ones display an AI Overview. Compare their CTR before/after. If the decline concentrates on these queries, you have an AI Overview impact.
Three levers:
1. Optimize to appear in AI Overviews. Google cites sources. If you’re cited, you get visibility (but not always a clickable link). I observe that sites with dense semantic structure and structured answers (FAQ schema, lists, tables) get cited more often. But CTR from the Overview is low (2 to 5% by my observation).
2. Strengthen long-tail and ultra-specific queries. AI Overviews are more frequent on head queries (1-3 words). On longer queries (5+ words), Overview display is less frequent (around 18% by my observation). Build content on these. Per-query volume is lower, but CTR stays stable.
3. Work on brand search. If users don’t click your generic category anymore, make them search your brand name. That means differentiated content, presence on other channels (social, email, display), and clear value prop. Brand search is less affected by AI Overviews (around 8% display by my observation).
These three levers don’t fully compensate for the loss. But they limit the damage.
Methodology note: To segment queries by AI Overview display, use a SERP tracking tool (SEMrush, Ahrefs, Sistrix) or a Python script with Selenium to capture SERPs. Cross with your Search Console data. If you lack the technical means, do a manual sample of 50-100 representative queries. Less precise, but gives you the trend.
Why Google doesn’t publish the data
Google could close this debate by publishing aggregated data. CTR evolution before/after AI Overviews, segmented by query type and site. Referral traffic evolution. Query volume evolution.
It doesn’t.
Why? A few hypotheses.
1. The data doesn’t support the narrative. If traffic has actually declined significantly, publishing figures would fuel criticism. Google prefers controlling the narrative in the absence of official contradictory data.
2. The data is too complex to summarize. Impact varies by sector, query type, geography. Publishing a global average would be misleading. Publishing segmented data would be too technical and could be misinterpreted.
3. Google doesn’t want to set a precedent. Publishing detailed data on one feature’s impact would create an expectation that Google do the same for every update. That would limit its flexibility.
4. Regulatory pressure. Google faces antitrust inquiries in the US and Europe. Publishing data showing traffic declines to third-party publishers could be used against it in those cases.
I don’t know which hypothesis is true. Maybe all of them. But the result is the same: you can’t count on Google to give you the figures. You have to measure yourself.
Losing traffic and want to understand why?
I spend 90 minutes with you on your Search Console, your SERPs, your queries. We segment what comes from AI Overviews, what comes from elsewhere. You leave with a costed action plan.
Book a strategic call — 45 minFrequently Asked Questions
What is a « bounce click » according to Google?
A quick click where the user gets the info then immediately returns to Google. Google says AI Overviews eliminate this type of click, but has published no data to prove it.
Do AI Overviews reduce organic traffic?
According to independent studies (BrightEdge, Semrush, Authoritas), yes. CTR drops 15-24% on queries displaying an AI Overview. Google says only « bounce clicks » are reduced, without public data.
How do I know if my site is impacted by AI Overviews?
Segment your queries in Search Console. Compare CTR and sessions before/after May 2025 on queries displaying an AI Overview. If the decline is concentrated there, you have an Overview impact.
Can I appear in AI Overviews and recover traffic?
Yes, but CTR from the Overview is low (2-5% observed). Optimize with structured answers, FAQ schema, lists, tables. That gives visibility, rarely direct clicks.
Which queries are most affected by AI Overviews?
Short (1-3 word) and informational queries. Long (5+ word) and transactional queries show fewer Overviews. Build long-tail content to limit impact.
