How Much Traffic Do AI Overviews Really Take? Every Major Study Compared

If a client asks you how much traffic AI Overviews are costing them, there is no single honest number to quote. There is, however, a well-supported range, and a clear explanation for why the published figures disagree. This article lines up the major studies, explains what each one actually measured, and turns the findings into something you can use in a client conversation.

The headline is simple. When Google shows an AI Overview above the organic results, fewer people click the organic results. How many fewer depends on the query, the position you hold and when the data was collected.

What do the main studies say?

The most cited studies agree on direction and disagree on size. Every credible study finds that AI Overviews reduce organic clicks. The estimates of the reduction for the top result run from roughly one third to more than half.

Two things stand out. First, the effect appears to have grown over time: the same research group reported a much larger gap in its later dataset than in its first. Second, Pew's study is the only one in this list based on observed behaviour of a panel of real users rather than aggregated click data. Its result, roughly half as many clicks when a summary appears, sits comfortably inside the range the click-data studies report.

Several of the figures circulating online come from blog posts that summarise other blog posts. Before you put a number in a client deck, trace it back to the original publisher and check the date, sample and method.

Why do the numbers disagree so much?

The studies disagree because they measure different things. Four differences explain most of the spread.

Query mix

AI Overviews appear far more often on informational questions than on commercial or navigational searches. A study weighted towards "what is" and "how to" queries will show a bigger drop than one weighted towards product or local searches, because a summary can fully satisfy an informational question. Nobody needs to click to learn the boiling point of water.

Which position is measured

Most click-data studies look at the first organic result. Lower positions were already receiving few clicks before AI Overviews, so the relative drop there is harder to measure and often looks smaller or noisier.

Time period

AI Overviews have changed shape since launch. They have appeared on more queries, grown longer, and changed how and where they show source links. A dataset from early 2025 and one from 2026 are not measuring the same product.

How "an AI Overview appeared" is detected

Rank trackers detect AI Overviews by scraping results from data centres, which is not identical to what a signed-in user on a phone sees. Pew avoided this by recording what panel members actually saw, which is one reason its result is useful as a sanity check.

How often do AI Overviews appear?

Estimates of prevalence vary as much as estimates of impact. Trackers report AI Overviews on anywhere from about a quarter to more than half of the queries they monitor, and Pew found an AI summary on roughly one in five searches in its panel. The difference again comes down to which queries are being sampled. For your clients, the relevant number is how often AI Overviews appear on their keyword set, which you can measure directly.

Zero-click behaviour adds context. Even before AI Overviews, a majority of Google searches ended without a click to any website, and studies of searches that trigger an AI Overview report zero-click rates above 80%. AI Overviews did not create zero-click search. They accelerated it on the queries where they appear.

Which queries lose the most clicks?

Informational queries with short, factual answers lose the most. Queries where the person still needs to act, compare or buy lose the least. A practical way to sort a client's keyword set is into three groups.

1. Answerable in a summary. Definitions, quick facts, simple how-tos. Expect the largest click losses. The goal shifts from winning the click to being the cited source.
2. Research and comparison. "Best X for Y", "X vs Y", buying guides. Summaries appear but people still click to see detail, prices and proof. Losses are moderate and the citation is valuable.
3. Transactional, local and branded. "Book", "near me", product and brand names. AI Overviews appear less often and click losses are smallest.

When you group a client's top 200 queries this way and weight them by traffic, you get a far better estimate of their exposure than any industry average.

Is a citation inside an AI Overview worth anything?

Yes, but its value is different from a ranking. A cited page earns a link inside the summary, brand visibility at the top of the page, and a chance at the click from the people who want more depth. Several industry analyses report that visitors arriving from AI answers convert at higher rates than average organic visitors, possibly because the summary has already filtered out people who only wanted a quick fact. That evidence is still early and mostly vendor-reported, so present it to clients as a promising signal rather than a proven rule.

There is also a defensive argument. If your client is not cited, a competitor or a publisher is. The summary still frames the answer, and the brand named in it gains the trust.

For how answer engines decide which pages to cite, see how ChatGPT and Perplexity choose which sources to cite. For the broader picture, our guide to AEO and GEO covers the full strategy.

How to measure the impact on a client's own site

Industry studies give you the range. Your client's own data gives you the answer. Here is a measurement approach that works with tools you already have.

Step 1: Find the queries that trigger AI Overviews

Use your rank tracker to flag which of the client's tracked keywords show an AI Overview. If your tracker does not record this, check the top 50 queries by impressions manually from a clean browser session and note the result.

Step 2: Compare click-through rate before and after

In Google Search Console, export query-level clicks, impressions and average position for a period before AI Overviews appeared on those queries and a comparable period after. Filter to queries where average position stayed roughly the same. If position held steady but CTR fell, the AI Overview is the likely cause.

Step 3: Use the generative AI report for impressions

Search Console now includes a report showing impressions in Google's generative AI features. It tells you how often the client's pages were shown as links inside AI Overviews and AI Mode. It does not show clicks separately; those are blended into the normal web search totals. Use it to show clients where they are being cited, not to calculate CTR.

Step 4: Watch conversions, not only sessions

If sessions from organic search fall but leads or sales hold steady, the lost clicks were mostly low-intent visits. That is a very different story from falling revenue, and it should change how you report. Our guide to proving SEO value in a zero-click world covers how to present this.

What this means for agency strategy

The research points to four practical shifts.

Re-forecast informational content. Traffic forecasts built on pre-2025 click-through curves will overstate what informational pages deliver. Rebuild forecasts using the client's own post-AI-Overview CTR.

Add citation as a goal. For queries where a summary appears, being cited is now part of winning. Track it alongside rankings so the client can see progress even when clicks fall.

Shift effort towards commercial intent. Comparison pages, product pages, local pages and proprietary data hold their clicks better. Rebalance content plans towards them.

Reset expectations early. Clients who see falling traffic with steady rankings assume something is broken. A short explanation with their own before-and-after data prevents that conversation from going badly.

How Rankwyn helps

Rankwyn's AEO Citation Radar tracks whether each client site is cited in AI answers for the queries you care about, alongside traditional rankings, so you can report citations and clicks side by side across a whole portfolio.

The bottom line

AI Overviews take a substantial share of organic clicks, with credible estimates running from about a third to more than half for the top result. The size of the effect depends on the query. Instead of quoting an industry average, measure your client's exposure on their own keyword set, add citations to what you track, and steer content towards the queries that still earn visits.