Sin categoría

Are AI Overviews Killing Your Website Traffic? What the Data Shows

AI Overviews are consistently associated with fewer clicks on regular search results, and one randomized experiment now points to a causal effect. That does not mean search has stopped being worth your time. It means rankings alone are a weaker measure of success, so you should judge search by leads and sales from the queries that still lead to action.

What does the research actually say?

The headline numbers differ because each study measures something different. Read together, they point in the same direction.

Study What it measured Result
Pew Research Center Behavior of 900 U.S. adults in March 2025 Clicks on a traditional result in 8% of visits with an AI summary, 15% without
Ahrefs Desktop click-through rate at position one across 300,000 keywords About 58% lower than forecast for keywords with an AI Overview
Field experiment by Agarwal, Toulouse School of Economics seminar listing Users randomly assigned to search with or without AI Overviews Outbound organic clicks down 39.8% where Overviews appeared
SparkToro All U.S. Google searches, January to April 2026 68.01% ended without a click

Each has limits worth knowing. Pew states that its data shows association rather than cause. Ahrefs compares against a forecast and covers desktop only. SparkToro compares different panels over time, calls that comparison “a bit of apples and oranges” and excludes the Google mobile app. The field experiment is the strongest design of the group, but we could only review its published abstract, so treat 39.8% as a preliminary estimate.

Is every search affected the same way?

No. Pew found that AI summaries appeared on about 18% of the searches in its dataset. They showed up for 8% of one- or two-word searches and 53% of searches with ten or more words. Searches using question words triggered them 60% of the time.

That pattern matters for your plan. Ahrefs also reports, in its first version of this study, that 99.2% of keywords triggering AI Overviews are informational. Pew’s sample and Ahrefs’ keyword data are not the same thing, but both suggest the exposure is highest on questions rather than on short, direct searches.

Our reading: a blog post answering a general question carries more risk than a service page built for someone ready to hire. We would not stop writing helpful content, but we would stop counting informational traffic as the main return on it.

Is AI Mode the real cause?

Not yet, at least by SparkToro’s measure. It found that only 0.34% of U.S. Google searches went to AI Mode from January to April 2026. The abstract of the field experiment does report exploratory evidence of even larger traffic reductions under an AI Mode condition, so this is worth watching. For now it is a smaller part of the picture for most small businesses.

Do the clicks that remain convert better?

Google says so. Its documentation states that clicks from AI Overview results tend to show higher engagement. Independent evidence is less supportive: the field experiment’s abstract found no difference in downstream engagement among users who clicked through. The two statements do not measure exactly the same thing, so we would not plan around either one. Measure your own conversion rate instead.

How can you tell if your business is affected?

You cannot isolate AI Overview clicks in the standard report. Google counts AI Overview and AI Mode traffic within overall search traffic, under the “Web” search type in the Search Console Performance report. Google’s newer guide also points site owners to a Generative AI performance report in Search Console, so check whether it is available for your property.

We recommend a simple monthly review:

  1. Split your pages into two groups. Put informational articles in one and service, location and booking pages in the other.
  2. Compare impressions against clicks for each group. Impressions holding steady while clicks fall is the pattern to look for.
  3. Search your top queries by hand. Note where an AI Overview appears and whether you are cited.
  4. Connect search to outcomes. Track forms, calls and bookings in Google Analytics, because traffic alone will no longer tell you the story.

What should a business owner do about it?

These are our recommendations, based on the evidence above.

  • Protect the pages closest to revenue. Make service pages specific about what you do, where you do it, who it is for and how to start. Vague pages give people little reason to click.
  • Offer what a summary cannot. Google advises creating content based on first-hand knowledge of the topic. Original photos, real process details, pricing logic and expert opinion are harder to reduce to a few lines.
  • Keep your pages eligible. Google says there are no special requirements beyond being indexed and eligible for a snippet. Be wary of anyone selling a secret AI markup.
  • Build demand beyond Google. If fewer searches lead to your site, being known by name, by email subscribers and on social platforms becomes more valuable.
  • Reset how you report. Replace “traffic is down” with leads, qualified calls and revenue per channel.

Frequently asked questions

Should I block Google from using my content in AI Overviews?

Usually not, unless you have a specific reason. Google offers controls such as nosnippet and max-snippet, but they also limit how your page appears in search. For most service businesses, being absent is a bigger risk than being summarized.

Are AI Overviews bad for local businesses?

We could not find a verified study that isolates small local businesses, so nobody should give you a precise number. Local searches with clear intent may behave differently from general questions. Your own Search Console and call data is the best evidence available.

Is SEO still worth paying for?

In our opinion, yes, when it is tied to outcomes. A plan that reports only rankings and sessions is now incomplete. Ask any provider how they will connect the work to leads.

How much traffic will I lose?

Nobody can tell you that in advance. The studies above measure different samples and periods. Compare your own data over at least a few months before making major decisions.

Need a second opinion on your search numbers?

JONTI Code Design is a bilingual studio in San José, Costa Rica, working in GMT-6 with clients mainly in the United States. We help businesses with technical and local SEO and with visibility in AI search. If your traffic has changed and you want to know whether it matters, write to us at info@jonti.co.cr.

Sources