AI Overviews CTR in 2026: Why Google, Ahrefs, Pew, and Seer Seem to Disagree

01

Dataset and period

AI Overviews clearly change click behavior, but the headline numbers are not directly interchangeable. Pew measured browsing sessions, Ahrefs compared ranking-page CTR across keyword sets, and Seer analyzed client search data with its own methodology.

AI Overviews are associated with lower click-through behavior in several independent datasets.

But there is no single universal number for:

AI Overviews reduce clicks by X%

because different studies measure different populations, queries, positions, devices, and outcomes.

Three commonly cited findings illustrate the problem.

Pew Research Center.

In a study based on browsing activity from 900 U.S. adults, Pew found users clicked a traditional search result on about 8% of visits when an AI summary appeared, versus 15% when no AI summary appeared.

Clicks on links inside the AI summary itself occurred in about 1% of visits with an AI summary.

Ahrefs.

Ahrefs reported that the presence of an AI Overview correlated with roughly a 58% lower average CTR for the top-ranking page in its updated study.

This is not the same metric as Pew’s share of search visits producing a click.

Seer Interactive.

Data reported by Search Engine Land showed AI Overview CTR rising from a low point in late 2025 into early 2026.

The reported comparison included roughly:

• no AI Overview: 3.3% CTR;
• cited in AI Overview: 2.1%;
• not cited: 0.9%.

Different dataset.

Different denominator.

Different conclusion.

Why these studies are not actually contradictory.

Suppose one study asks:

What percentage of search visits led to any result click?

Another asks:

What is the CTR of the page ranking first?

Another asks:

What is the CTR for pages cited in AI Overviews versus pages not cited?

All three can be correct simultaneously.

The web loves one number because one number fits into a screenshot.

Measurement is crueler.

Pew’s methodology.

Pew analyzed web browsing data from 900 U.S. adults for March 2025.

The broader search dataset contained tens of thousands of Google searches.

Pew compared what people did after search pages with and without AI-generated summaries.

Key finding:

Users were less likely to click result links when an AI summary appeared.

Pew also found longer and question-like queries were more likely to trigger AI summaries.

That matters because query type is a confounder.

Informational questions already behave differently from branded navigation.

Sources reviewed.

1. Google users are less likely to click on links when an AI summary appears — Pew Research Center; accessed 2026-08-06. ^[1]
2. Update: AI Overviews Reduce Clicks by 58% — Ahrefs; accessed 2026-08-06. ^[2]
3. Google AI Overviews CTR shows early signs of recovery: Study — Search Engine Land; accessed 2026-08-06. ^[3]
4. AI features and your website — Google Search Central; accessed 2026-08-06. ^[4]

02

Methodology

Ahrefs methodology.

Ahrefs selected large keyword sets with and without AI Overviews and compared estimated or observed click-through behavior for ranking pages according to its published methodology.

Its updated study reported a large negative correlation for the top-ranking page.

This is especially relevant for publishers whose old model was:

rank #1
→ expect large traffic share

AI Overviews can break that historical relationship.

Seer’s different angle.

Seer’s reported data suggested cited pages perform better than uncited pages on AI Overview SERPs.

That does not mean cited pages outperform equivalent rankings on clean SERPs.

The reported numbers still showed stronger CTR where no AI Overview appeared.

So two statements can both be true:

1. AI Overviews reduce overall click opportunity.
2. If an AI Overview exists, getting cited can improve your relative click opportunity.

That distinction is strategically important.

Google’s position.

Google has said that clicks from search pages with AI Overviews can be higher quality, meaning users may spend more time on destination sites.

Google also says generative AI features increase search usage.

Those statements concern:

• engagement quality;
• query volume;
• user behavior.

They do not automatically rebut every study showing lower click-through rates.

More searches can coexist with lower CTR per search.

Stop using position as the only KPI.

A page can remain:

average position: 1.8

while clicks decline.

Track:

• impressions;
• clicks;
• CTR;
• AI feature exposure;
• citations;
• landing sessions;
• conversions;
• branded demand.

If impressions stay stable and CTR falls, SERP presentation may be changing even when ranking holds.

03

Result

Build a clean before-and-after cohort.

For your own site:

1. Select queries that consistently show AI Overviews.
2. Select comparable queries without AI Overviews.
3. Match intent where possible.
4. Record rank.
5. Record device.
6. Record date.
7. compare CTR.
8. compare conversion.

Do not compare:

branded navigational query

with:

ten-word medical question

and call the difference an AI effect.

Citation matters.

If your page is cited in the AI Overview, track it separately.

The Seer-reported data suggests citation can improve relative CTR within AI Overview SERPs.

Use:

AIO + cited
AIO + not cited
No AIO

as three buckets.

That is more useful than:

AI / not AI

Industry differences matter.

Ahrefs’ broader 2026 CTR benchmarks show significant variation by industry.

Answer-heavy niches can suffer more zero-click behavior.

Transactional searches can retain more click pressure because the user still needs:

• product;
• booking;
• purchase;
• login;
• quote.

Do not import a publisher CTR benchmark into ecommerce planning.

What to report to clients.

Good:

In our tracked query cohort, AI Overview queries produced 1.2% CTR versus 2.4% for comparable non-AI queries during July.

Weak:

AI reduces traffic by 58%.

The second statement imports someone else’s methodology into your client’s business.

Use external research as context.

Use first-party data for decisions.

04

Limitations

FAQ.

Do AI Overviews reduce clicks?

Several independent studies have found lower click behavior when AI summaries or Overviews appear.

Is the reduction exactly 58%?

No. That is a finding from a specific Ahrefs methodology, not a universal constant.

Does being cited help?

Some research reported higher CTR for cited pages than uncited pages on AI Overview SERPs.

Does Google say clicks are better?

Google has said clicks from AI Overview search pages can be higher quality by engagement measures.

Can search usage increase while website CTR falls?

Yes. More total searches and lower click rate per search can coexist.

Measurement checklist.

• Query intent segmented.
• AI Overview presence recorded.
• Citation status recorded.
• Rank recorded.
• Device recorded.
• CTR measured.
• Sessions measured.
• Conversion measured.
• External benchmarks labeled.
• No universal reduction claimed.
• Time period explicit.

Verdict.

The serious conclusion is not “AI Overviews cut clicks by exactly 58%.”

The serious conclusion is that ranking position no longer predicts click opportunity as reliably as it once did.

Measure the search surface, the citation state, and the business outcome together.

Verification record.

• Pew methodology and click figures were checked.
• Ahrefs’ 58% result is explicitly attributed to its study methodology.
• Seer figures are attributed through Search Engine Land.
• Google’s higher-quality-click claim is kept separate from CTR estimates.

Duplication and search-intent record.

No prior package reconciles the competing AI Overview CTR studies as its primary search intent.

References

Sources behind this record

1. Google users are less likely to click on links when an AI summary appears — Pew Research Center (accessed August 6, 2026)
2. Update: AI Overviews Reduce Clicks by 58% — Ahrefs (accessed August 6, 2026)
3. Google AI Overviews CTR shows early signs of recovery: Study — Search Engine Land (accessed August 6, 2026)
4. AI features and your website — Google Search Central (accessed August 6, 2026)

Corrections

Correction history

No corrections recorded.

To report an error, use the public corrections path.