CTR Study 2026: What 5,460 B2B Pages Earned per Google Position

A non-branded #1 ranking earned a 0.44% click-through rate on our site.

That’s from 1.1 million Search Console impressions across 5,460 pages in Dutch, French, English and German, measured from June to mid-October 2026.

The verdict: on a B2B site in the AI Overview era, a ranking is a chance to be seen, not a promise of a click. What you rank with, and for whom, decides the click.

How we measured: first-party data from our own Search Console property (customerimpact.be), pulled through the API on 17 October 2026. No third-party estimates and no client data. The full method and filters are further down.

Key findings

1. A non-branded #1 position earned a 0.44% CTR. 262 of the 277 queries we held in the top 3 (with 20+ impressions) got zero clicks.
2. Brand searches clicked 37 times more often than generic searches in positions 1 to 3 (9.6% vs 0.26%).
3. Definition and glossary pages earned the least per impression. How-to pages on page one earned 0.76%, definition articles 0.30%, glossary entries 0.11%.
4. Question queries didn’t click less. In the top 10 they earned 0.19% against 0.09% for other queries.
5. Short queries clicked about twice as often as queries of three words or more, among the queries Google reports.
6. 86% of our clicks came from queries Search Console hides. The query report showed 457 of 3,212 clicks.
7. 40% of reported impressions weren’t people. One bot query alone produced 644,243 impressions.
8. Top-3 CTR rose as our pages matured. It went from 0.13% in July to 0.52% in early October.

How did we run this study?

We pulled every page and query row from the Google Search Console API for customerimpact.be. The window runs from 1 June to 15 October 2026: the period in which our library of about 5,500 articles went live.

One row is one page, one query and one month. CTR is total clicks divided by total impressions for a group. The 95% intervals in the charts come from a bootstrap over queries or pages, so a few lucky queries can’t carry a result.

Before calculating any number, we removed four kinds of noise:

• the bot query “best seo tool” (Dutch, desktop, zero clicks, up to 160,000 impressions a day since 10 October 2026)
• prompt-style queries of eight words or more
• search operators such as site:
• template placeholders such as {searchterms}

That removed 735,273 impressions and 3 clicks. What’s left: 48,824 queries, 5,460 pages, 1.09 million impressions and 456 clicks at query level, plus 3,279 clicks at page level.

We classified pages by their own title and queries by their wording. For example, a title starting with “What is” counts as a definition, one with “vs” as a comparison. Branded means the query contains our company or founder name.

What CTR does a #1 ranking earn on a B2B site?

A non-branded #1 position earned 0.44% (95% interval 0.26% to 0.67%). Position 3 earned 0.29%, positions 6 to 10 earned 0.07% and positions 11 to 20 just 0.03%.

The chart shows the drop per position bucket, with the number of impressions and clicks behind every bar:

Non-branded queries, customerimpact.be Search Console, 1 June to 15 October 2026. Whiskers are 95% bootstrap intervals.

That’s far below the click curves most marketers carry in their head. For keywords with an AI Overview, Ahrefs measured a 1.6% desktop CTR at position 1 in December 2025, against 7.3% two years earlier (Ahrefs, February 2026). Ours is lower again, and that’s with the noise removed.

Part of the gap is the type of demand. Most of our top-3 rankings are long, specific B2B terms with a handful of searches a month.

Say a Dutch query for a niche marketing term shows our page in position 2 five times a month. Nobody clicks, because the answer is already on the results page.

So a top-3 ranking earned nothing in 95% of cases: 262 of 277 queries got zero clicks.

Using our CTR by Google position benchmarks? Plan with the low end of the range.

Do brand searches still get clicked?

Yes, and it’s not close. Branded queries in positions 1 to 3 earned a 9.6% CTR, against 0.26% for non-branded queries in the same positions.

Branded vs non-branded CTR by position band, customerimpact.be, 1 June to 15 October 2026.

The branded sample’s small: 9 queries, 635 impressions, 61 clicks. So read 37 times as an order of magnitude, not a precise ratio.

But the direction makes sense. Someone who types your name wants your site, and no AI summary replaces that.

Generic searchers, for instance someone typing “lead scoring model”, get their answer on the results page and move on.

Does your plan only count generic rankings? Then track brand demand as its own line.

Which page types earn the most clicks?

How-to pages earned the most per impression, definition and glossary pages the least. We measured this at page level, which includes the clicks Search Console hides at query level.

For pages with an average position of 10 or better, how-tos earned 0.76% and definition articles 0.30%. The intervals don’t overlap. All groups sit at an average position between 7.9 and 9.0.

Page-level CTR of article pages averaging a top-10 position, customerimpact.be, 1 June to 15 October 2026.

Glossary entries did worst at 0.11% (148 pages, 15 clicks).

A short definition is exactly what an AI Overview or featured snippet can answer in full. That fits what we see in our glossary strategy for AI Overviews: glossary pages earn visibility and citations, not visits.

We can’t see which results pages showed an AI Overview, so this is a correlation.

Definition articles vs general guides (0.30% vs 0.41%) aren’t distinguishable either.

Write definitions to be cited. Write how-tos and comparisons to be clicked.

Do question queries lose clicks to AI answers?

Not on our site, and that’s worth knowing. In the top 10, question queries earned a 0.19% CTR, against 0.09% for other queries. “What is” queries showed the same split.

Non-branded queries ranking in the top 10, customerimpact.be, 1 June to 15 October 2026.

That one surprised us. We’d expected the opposite.

The common view is that questions are the queries AI Overviews swallow first, as in our piece on AI Overviews and fewer clicks.

Two caveats. The question group has only 41 clicks, and the intervals touch. So “question queries didn’t click less” is safe; “they click more” isn’t proven.

It doesn’t contradict the page-type result either. Say a definition page wins the odd “what is” click: it still misses the long tail that guides and how-tos collect.

Do long-tail queries click better?

Among the queries Google reports, no. One- and two-word queries in the top 10 earned a 0.15% CTR, queries of three words or more 0.08%.

Non-branded top-10 queries by word count, 8+ words excluded, customerimpact.be, 1 June to 15 October 2026.

But the long tail that does click mostly sits in queries Search Console doesn’t show.

So read this chart as “what you can see”, not “what pays”.

How much does Search Console hide?

Most of the clicks. Our property totals show 3,212 clicks, the query report only 457. 86% of clicks came from anonymised queries.

Property totals vs the sum of the query report, bot query removed from both, customerimpact.be, 1 June to 15 October 2026.

Google leaves these queries out of the report to protect user privacy but keeps them in the chart totals (Search Console Help). For impressions the query report still covered 69%.

So the hidden part is small in impressions but large in clicks.

That changes how you read CTR. Guides and how-tos showed only 7 to 8% of their clicks at query level.

A query-level CTR report makes them look far weaker than they are.

How much of your impression data is real?

Less than you’d think. Our four fixed filters removed 40% of all reported-query impressions.

One Dutch desktop query, “best seo tool”, produced 644,243 impressions without a single click.

Share of reported-query impressions per filter, customerimpact.be, 1 June to 15 October 2026.

We also flagged 1,482 queries with 50+ impressions, zero clicks and more than 99% desktop traffic. Many are agency terms that rank trackers check every day, such as “seo service” or “agence branding”. Others are long US questions comparing GEO tools that read like AI agents. Together they hold 13% of the remaining non-branded impressions.

We’ve kept those in the headline numbers, because the rule can catch real people too. With them removed, every result points the same way: position 1 rises to 0.56%, and the question and page-type gaps stay.

Check for one-query spikes and desktop-only zero-click queries before you report CTR.

Is CTR falling over time?

Not in our window. Non-branded CTR in positions 1 to 3 went from 0.13% in July to 0.29% in September and 0.52% in the first half of October. Positions 4 to 10 stayed flat around 0.1%.

Monthly non-branded CTR at fixed position bands, customerimpact.be, June to 15 October 2026.

The likeliest reason is maturity. New pages first rank top 3 for tiny long-tail queries, then for terms people actually click.

The monthly counts are small (6, 3, 30 and 35 clicks). It’s a direction, not a rate.

What are the limits of this study?

• One site. This is a young, multilingual B2B library of about 5,500 pages, not a cross-industry panel.
• 4.5 months. We couldn’t compare against a year without AI Overviews, because the site barely existed in Google before June 2026.
• Small click counts. 456 query-level clicks and 3,279 page-level clicks. We show intervals and only call a gap real when they don’t overlap.
• No AI Overview flag. Search Console doesn’t say which results showed an AI Overview, so every AI link here is a correlation.
• Average positions. Search Console reports an average per row, so “position 1” means “averaged to about 1”.
• Our own classification. Page and query types come from rules on titles and wording, which will misfile some cases.

Frequently asked questions

What is a good organic CTR for a B2B website in 2026?

It depends on brand share and page type. On our site non-branded page-one queries earned well under 1%, brand searches around 10%. Compare yourself with your own past data, not a generic curve.

Why is my Search Console CTR so low?

Often it’s bot and rank-tracker impressions plus hidden queries. Remove one-query spikes and desktop-only zero-click queries, then compare page-level clicks with impressions.

Do AI Overviews reduce clicks on definition pages?

Our data fits that pattern: definition and glossary pages earned the lowest CTR at the same average position. We can’t prove the AI Overview caused it, because Search Console doesn’t flag which results showed one.

Can I reproduce this study on my own site?

Yes. Pull page and query rows per month through the Search Console API, apply the four filters above and compare page-level with query-level clicks. Our Search Console API guide covers the pull.

What should you do with this?

Plan content for clicks where clicks still happen: how-tos, comparisons, data and brand. Plan definitions for citations, and improve your organic CTR where the click is still there to win.

Want to know what your own Search Console data says once the noise is out? Our SEO team runs this same analysis on your property and shows you which pages earn clicks and which only collect impressions.