The short answerAbout 1%. Across 13 sites we manage, from August to October 2026, a non-brand position-1 Google result earned a 1.05% CTR (95% interval 0.88% to 1.24%), position 2 earned 1.06%, position 3 0.52% and position 10 0.24%, a fraction of the 7% to 28% that published blended curves give.
| Position | Non-brand CTR | 95% interval | Impressions |
|---|---|---|---|
| 1 | 1.05% | 0.88% to 1.24% | 56,200 |
| 2 | 1.06% | 0.86% to 1.28% | 36,676 |
| 3 | 0.52% | 0.30% to 0.83% | 51,462 |
| 4 | 0.55% | 0.36% to 0.80% | 50,808 |
| 5 | 0.71% | 0.42% to 1.17% | 45,224 |
| 6 | 0.51% | 0.32% to 0.82% | 53,391 |
| 7 | 0.36% | 0.22% to 0.56% | 58,711 |
| 8 | 0.27% | 0.18% to 0.40% | 65,919 |
| 9 | 0.26% | 0.19% to 0.35% | 55,494 |
| 10 | 0.24% | 0.19% to 0.30% | 41,409 |
| 11 to 20 | 0.10% | 0.08% to 0.12% | 138,324 |
CTR here is Google's own definition: clicks divided by impressions, read from Search Console. The sample is 13 sites in B2B software, legal services, education, finance and health, 1 August to 31 October 2026, 44,620 non-brand site-and-query pairs and 653,618 non-brand impressions at positions 1 to 20.
If someone hands you a traffic forecast built on a 28% position-1 CTR, this table is the reason to send it back. Our guide to SEO forecasting covers how to build the forecast itself. This page gives you a dated curve to check one against.
How we ran the study
We wrote the method down before the study's data pull began, and added one amendment during analysis: a third automated-query check, and 500 resampling draws for the US and device intervals.
The data. The Search Analytics API in Google Search Console, web search, final data only, for 1 August to 31 October 2026 (92 days). We pulled every row at the level of date, query and device for the 13 sites we actively managed on 5 November 2026: our own site and 12 client sites. That came to 391,806 query rows. We do not name the clients or attach any figure to a named site.
Position. Search Console reports a daily average position for each query. We rounded it to the nearest whole number and kept positions 1 to 20, with 11 to 20 pooled into one bucket.
Brand. We removed every query that contains the site's own brand name, using a fixed pattern per site written before the pull. Brand searches behave so differently that mixing them in would hide the curve.
CTR per position. Total clicks divided by total impressions for every query row in that position. Bigger queries count for more, the way they do in your own Search Console.
Intervals. We resampled whole site-and-query pairs 1,000 times and read the middle 95% of the results. A query that shows up on many days moves as one unit, so a few busy queries cannot fake precision.
What the curve leaves out. Google withholds rare queries from query-level tables. Google's documentation says "Anonymized queries are those that aren't issued by more than a few dozen users over a two-to-three month period." On our 13 sites the named query rows held 28.7% of all clicks and 43.2% of all impressions in the window. So this curve covers named queries only, which is also all you can see per query in your own account. Ahrefs, reading 887,534 Search Console properties, put the hidden share at 46.77% of clicks in April 2025. Our sites hide more: 71.3% of clicks.
Positions 1 and 2 both earn about 1% on non-brand searches
The curve does not halve with each step down the page. Pooled across devices, positions 1 and 2 show no measurable difference, at 1.05% and 1.06% with overlapping intervals. Within each device, position 1 still leads: 0.65% against 0.43% on desktop and 3.47% against 2.56% on mobile. Desktop, where CTR runs lowest, made up 85.9% of position-1 impressions and 70.6% of position-2 impressions, so the pooled tie partly reflects position 1's heavier desktop mix. Below that, the curve steps down in blocks. Positions 3 to 6 sit together between 0.51% and 0.71%, 7 to 10 between 0.24% and 0.36%, and page two returns 0.10%.
That has a practical edge. Position 1 earned roughly double position 3 on these sites. Moving from position 2 to position 1 is less clear-cut: the pooled figure shows no measurable gain, and the device split suggests a gain exists. Before a plan spends a quarter fighting for the top spot on a query already at position 2, check which devices that query's impressions come from.
Brand searches click almost 25 times more often at position 1
At position 1, brand queries earned a 26.02% CTR. Non-brand earned 1.05%.
| Position | Brand CTR | Non-brand CTR |
|---|---|---|
| 1 | 26.02% | 1.05% |
| 2 | 5.48% | 1.06% |
| 3 | 8.14% | 0.52% |
Brand queries were 7.4% of impressions in our sample and 50.2% of clicks. That is one reason published curves look healthier than a non-brand forecast should: on our sites, half the clicks in a blended curve came from people who already wanted you. It is not the whole gap. Advanced Web Ranking sees the same direction at a smaller gap, and its unbranded position-1 figure is still far above ours: "At position 1, branded terms average a 27.3% CTR against 19.4% for unbranded (US, all devices, July 2026)."
If an agency report shows your total CTR rising, ask what share of the clicks came from your brand name. A program that grows brand search is doing something real, but it is a different result from winning new non-brand demand, and the report should say which one it is.
Mobile clicks five times more often than desktop at the top
At position 1, non-brand CTR on mobile was 3.47% (interval 2.91% to 4.14%). On desktop it was 0.65% (0.51% to 0.80%). The intervals do not touch at positions 1 to 4.
| Position | Desktop CTR | Mobile CTR |
|---|---|---|
| 1 | 0.65% | 3.47% |
| 2 | 0.43% | 2.56% |
| 3 | 0.28% | 1.42% |
| 4 | 0.25% | 1.51% |
| 5 | 0.38% | 1.33% |
The device mix is the stranger number. Desktop held 85.9% of non-brand position-1 impressions and 68.9% of all non-brand impressions at positions 1 to 20, but only 42.4% of the clicks. StatCounter put US web page views at roughly half desktop over the same months (52.4% desktop and 45.43% mobile in October 2026). The two counts differ in scope: StatCounter measures US page views on sites running its tracker, while our device mix covers searches from all countries, and about two-thirds of the non-brand clicks behind it came from the US.
We cannot prove why. One reading fits the facts: some desktop impressions are not people. In September 2025, Google appeared to disable, or to test removing, the parameter that let tools load 100 results per page. Search Engine Journal reported that "Multiple SEO professionals saw sharp declines in desktop impressions in Google Search Console starting September 10, with average position increasing accordingly." Search Engine Roundtable reported Tyler Gargula's count across about 320 properties: "87.7% of those properties / sites experienced drops in impressions and 77.6% of sites experienced drops in the number of unique ranking keywords". If tools were loading results pages on desktop before that change, some may still be doing it another way. Our window starts eleven months after that change and desktop still dominates top-position impressions. Search Engine Journal called the earlier explanation, that bots had inflated desktop impressions before the change, "a community theory at this stage, not a confirmed Google explanation." Our use of it for August to October 2026 is less tested still, so treat the tool-traffic reading as a hypothesis.
Advanced Web Ranking also finds mobile ahead at the top, and its tool page says the two devices "are reported separately because CTR runs higher on mobile at the top positions". In July 2026 it put position-1 organic CTR at 35.74% on mobile and 20.02% on desktop, about 1.8 times. Ours runs from 0.65% on desktop to 3.47% on mobile, so on our data mobile clicks about five times as often, a far wider split than Advanced Web Ranking's.
Obvious bot queries explain only a small part of the gap
We tested the automated-traffic idea directly. For the US, we took the 1,000 non-brand queries with the most impressions at positions 1 to 10 and checked each one's monthly US search volume in DataForSEO. A query counted as likely automated when its 92-day impressions were more than three times its three-month search volume, or when it had no measurable volume and 100 or more impressions. After the data came in, we added a third test: a query with 100 or more US impressions that Google Ads could not look up at all, because it carried search operators, quotes, more than 10 words or 80 characters, or a character the lookup rejects. That third test also swept in a few ordinary searches that contain a slash.
131 of the 1,000 queries met one of those tests. They held 11.6% of US non-brand impressions at positions 1 to 3, 22.4% at positions 1 to 10, and earned 26 clicks in 92 days. Some are easy to spot, such as a boolean search for a named company, built from quoted terms chained with OR. Others read like prompts written for or by AI tools, such as 'what solutions help optimize content coverage for query fan-out?', which has no measurable search volume.
Then we removed them and redrew the US curve. The US intervals come from 500 resampling draws.
| Position | US, all queries | US, without flagged queries | 95% interval |
|---|---|---|---|
| 1 | 1.15% | 1.25% | 1.06% to 1.47% |
| 2 | 1.40% | 1.55% | 1.25% to 1.93% |
| 3 | 1.01% | 1.17% | 0.93% to 1.46% |
Position 1 moved by a tenth of a point. In the US, desktop position 1 went from 0.65% to 0.71% and mobile from 3.57% to 3.84%. The low curve survives the cleanup. Whatever is dragging desktop CTR down is not limited to queries a volume check can catch.
The curve holds when you change the sample
We ran three checks, all fixed in the method before the data came in.
- Every site weighted equally, using sites with at least 200 impressions in a position: position 1 averaged 1.27% across 7 sites, position 3 0.69% and position 10 0.28%.
- Without the largest site: position 1 fell to 0.88%.
- Without the one SEO agency site in the sample: position 1 rose to 1.42%, position 2 to 1.48% and position 10 to 0.29%.
Across these three checks and the cleaned US curve, non-brand position 1 stayed below 1.5%.
Sites do differ. Seven sites had at least 1,000 non-brand impressions at positions 1 to 3. At those positions, the site with the highest CTR earned more than 40 times the CTR of the site with the lowest, and the median site earned 0.48%. The spread is wide enough that your own curve beats any benchmark, ours included.
Why published CTR curves disagree so widely
Across the published curves in the table below, position-1 CTR runs from 3.6% (First Page Sage, with an AI answer on the page) to 27.6% (Backlinko, 2019). Blended or no-AI figures run from 7.1% to 27.6%. Our non-brand 1.05% sits below every published figure in it. Most of them are right about what they measured. They measured different things.
| Source | Data window | Position 1 CTR | What it measures |
|---|---|---|---|
| This study | August to October 2026 | 1.05% | 13 sites, non-brand, Search Console, all devices |
| First Page Sage | March to August 2026 | 7.1% blended, 22.6% without an AI answer, 3.6% with one | 1,184 domains, weighted to US query volume |
| Advanced Web Ranking | July 2026 | 20.02% across all results pages, 9.69% with an AI Overview | US desktop |
| Backlinko | published 2019 | 27.6% | 12,166,560 queries, before AI Overviews |
Five things move the number:
- Unit. A keyword-level study counts one CTR per keyword and averages. Search Console pools every impression, and long-tail queries with a handful of impressions each carry real weight.
- Brand. Blending brand queries in lifts the top of the curve sharply, as the brand table above shows.
- Device. Desktop-only and mobile-only curves sit far apart on our data.
- AI Overviews. Ahrefs found that "the presence of an AI Overview now correlates with a 58% lower average clickthrough rate for the top-ranking page," on desktop data from December 2025. Advanced Web Ranking measured a 52% cut.
- Year. Backlinko's 27.6% comes from a study first published in 2019. First Page Sage says its own blended figure "fell from 39.8% in our 2025 report to 7.1%".
First Page Sage's page says "The blended column is the one to use for traffic forecasting." For a B2B site forecasting non-brand demand, our data says even its blended 7.1% is about seven times our non-brand position-1 figure.
AI Overviews sit on top of almost all of this
Search Console does not tell you which queries showed an AI Overview, so our curve cannot split by it. Google counts an AI Overview as one result: "An AI Overview occupies a single position in search results, and all links in the AI Overview are assigned that same position." In our own October 2026 panel of 180 B2B software searches, Google showed an AI Overview on every one (LoudFace research). We think most non-brand queries in this sample carried one too. That is an inference, and it explains only part of the gap: curves measured with an AI answer present still put position 1 at 3.6% to 9.69%.
Research outside our sample points the same way. Pew Research Center, reading 68,879 Google searches by 900 US adults in March 2025, found "Users who encountered an AI summary clicked on a traditional search result link in 8% of all visits." Without one, they clicked on 15% of visits.
The newest data softens that a little. Seer Interactive's 2026 update found that "Organic CTR on AIO-present queries climbed from a floor of 1.3% in December 2025 to 2.4% in February 2026," and that being cited inside the Overview "delivers +120% more organic clicks per impression versus when you are not cited." That is why a citation inside the Overview is now part of the ranking job, and why our AI Overviews work reports per-prompt visibility and average cited position in AI Overviews every week. The optimization side is in our guide to Google AI Overviews optimization for B2B SaaS.
What to do with this curve
Check every forecast against a dated curve. Ask which CTR curve the forecast uses, its year, and whether it includes brand searches. A non-brand forecast built on a 2019 curve overstates clicks many times over on our data.
Build your own curve first. Pull a quarter of your own data from Search Console by query, remove your brand terms, and compute clicks divided by impressions per rounded position. Your own number beats ours: on our sites, the top-three CTR of the highest-CTR site was more than 40 times that of the lowest.
Split desktop from mobile before you judge a page. A desktop-heavy impression count with almost no clicks may not be people at all. Look at the queries behind it before you rewrite a title tag to fix it.
Treat CTR as a diagnostic. We judge an organic program on qualified pipeline, not on impressions or clicks. A page can earn citations in AI answers, shape a shortlist and send a buyer to search your brand name later, and none of that shows up as a click on the page that did the work. Track the SEO KPIs that tie to revenue, and read CTR to explain why one of them moved.
If you want Google rankings and AI citations run as one program, that is how our SEO and AEO service works.
Limits
This is 13 sites. The mix leans B2B and the sites are mostly small, and about 61% of their clicks came from the US. Search Console hides the rarest queries, which held most of the clicks on these sites. Google reports average position as the average position of the site's topmost result, so a rounded daily position is an approximation. Two of the 13 sites contribute almost nothing to the named query rows, and two sites' brand names are ordinary words, so some generic queries on those sites were removed as brand. We could not see which queries showed an AI Overview. The automated-query test catches only queries that look automated by volume or form. We kept the raw rows, the search volume reads and the scripts behind every number on this page.
