CTR by position is the share of clicks a result gets depending on where it shows up in Google, and if you ask the search engine today how much that is, the AI summary answers with a lot of confidence: 25% to 35% for the first spot, 15% to 20% for the second and under 5% from there down. That table comes from an agency blog that never explains where it got it, and it is the one that ends up in plenty of SEO reports as "what to expect".
So we went to see what happens in real accounts. We reviewed Search Console for 153 company websites in the US and Latin America, from October 4 to 31, 2026, using non-branded searches only: at the median, the first position had a CTR of 5.2% and the fifth, 0.8%.
My read is that the CTR curve a typical site sees without brand searches is several times lower than the tables going around, and that what moved in 2026 was the middle of the page (positions 4 to 10), while the top 3 held steady.
What is CTR by position?
CTR (click-through rate, or the percentage of clicks) is clicks divided by impressions. CTR by position splits that calculation by the spot your result holds: how much a result that shows up first, second or tenth gets, on average.
If you plot those numbers you get the CTR curve, which always slopes down to the right (the lower you appear, the less people pick you).
One important detail: in Search Console, position is the average position of each query over the period. A search "in position 1" can mix days in first place with days in second, and a page can show up third for one query and fortieth for another. That is why the curve is built by query and not by page.
If the formula feels distant, the basics are in what CTR is, and the CTR calculator does the math for you.
What is CTR by position in Google today?
This is the October curve for non-branded searches, with the median per site. May sits next to it, to compare the same position at two points in the year:
| Position | Sites | CTR October 2026 | CTR May 2026 | Oct/May ratio | Dropped in |
|---|---|---|---|---|---|
| 1 | 150 | 5.2% (p25 2.1%, p75 14.2%) | 4.7% | 1.10 | 43% |
| 2 | 149 | 3.0% | 3.1% | 1.02 | 48% |
| 3 | 153 | 1.8% | 1.7% | 0.93 | 53% |
| 4 | 152 | 1.3% | 1.3% | 0.89 | 61% |
| 5 | 153 | 0.8% | 1.1% | 0.87 | 62% |
| 6 | 152 | 0.6% | 0.8% | 0.87 | 58% |
| 7 | 150 | 0.4% | 0.5% | 0.92 | 55% |
| 8 | 151 | 0.3% | 0.4% | 0.77 | 62% |
| 9 | 149 | 0.3% | 0.3% | 0.81 | 60% |
| 10 | 146 | 0.25% | 0.28% | 0.91 | 56% |
| 11 to 20 | 152 | 0.25% | 0.25% | 1.09 | 43% |
| Top 3 grouped | 153 | 3.1% | 3.3% | 0.98 | 51% |
| 4 to 10 grouped | 153 | 0.57% | 0.63% | 0.95 | 53% |
The last column is the share of sites where the CTR for that position dropped between May and October.
The first thing that jumps out is the distance: within the same domain, position 1 had 5.6 times the CTR of position 5 (median; p25 2.7, p75 10.4, across 149 sites). In other words, moving from fifth to first on a non-branded query multiplies its clicks by somewhere between three and ten.
The second is the range. In first place, a quarter of the accounts were below 2.1% and another quarter above 14.2%. With that much spread, an industry average says little about your case, which is why the end of this article covers how to get your own curve.
Why is your CTR lower than the tables going around?
Because those tables usually measure something else. Four differences in method explain most of the gap.
The first is brand. When someone searches for your name they have already decided where they want to go, so the CTR of those queries is much higher. In the sample, 28% of total clicks came from branded searches (median; 26% in the US and 37% in Latin America).
A table that does not separate that traffic mixes two very different behaviors. In my CTR by position benchmarks for Chile, which include it and weight by impressions, the first spot reaches 18.86%, and in Peru, 20.74%. Those numbers and this curve's 5.2% are each right for what they measure, and how much each part weighs is something we measured across 131 sites.
The second is how you average. If you add up the clicks and impressions of every account together, the biggest ones push the result: with the same data, that pooled average gives 9.8% for first place, almost double the 5.2% median.
The median describes the site in the middle. The pooled average describes the ones with the most traffic.
The third is average position and the long tail. Search Console includes thousands of long queries with few impressions, and many show up first for a result nobody expected to see there. The most quoted studies (like Advanced Web Ranking, which publishes averages by position every quarter, or the 27.6% figure for the first result that Backlinko cites) do not publish the per-site median without brand, so their tables are not directly comparable with this curve.
How much those long queries weigh in your account is something we measured in long-tail keywords.
The fourth is AI summaries. Links inside AI Overviews share the position of the block, and links in AI Mode are counted inside web search, so that traffic (and the lack of it) is mixed into the same curve.
Ahrefs estimated that on queries with an AI Overview the CTR of the first result was 34.5% lower than expected without the summary, with 2.6% for position 1 (desktop, March 2025). It is a different sample, but it points the same way.
A homegrown example: the calculator on my site had 1,818 impressions over the last 90 days, at an average position of 9.7, and got 3 clicks. That is 0.17% (yes, the tool that calculates CTR has a pretty sad one), a bit below the 0.25% for position 10 on this curve.
Did CTR by position change in 2026?
We compared October with May 2026, at the same position and within each domain, and looked at the share of accounts where CTR dropped. The top 3 barely changed, and the middle of the page changed quite a bit.
For the top 3 grouped, the ratio was 0.98 and CTR dropped in 51% of sites, a statistical tie. First place even went up at the median (ratio 1.10, dropping in only 43%).
On the other hand, at every position from 4 to 10, CTR dropped in more than half of the accounts, between 55% and 62%. Position 8 fell the most, with a ratio of 0.77.
In Latin America the pattern was sharper. In positions 4 to 10, CTR dropped in 67% of the region's 27 sites (median change of -10%), against 48% of the 90 in the US (median change of +5%).
In positions 5 to 9, the drop in Latin America reached between 64% and 81% of domains. At the same time, first place in Latin America had a CTR of 8.6% in October, compared with 5.2% in the US.
Two caveats. This is a comparison, with no proven cause: May and October have different seasonality, and if May still carried some of the Search Console impressions bug, its CTR would be underestimated (which would work against the drop we see). And the gap between Latin America and the US may come from the industry mix or from how AI Overviews and AI Mode rolled out in each country, something this data cannot separate.
Where do the clicks come from?
In October, at the median per site and without brand, the top 3 had 14% of impressions and 46% of clicks.
Positions 4 to 10 had 58% of the visibility and 43% of the traffic, and from position 11 down, 22% of what was shown and just 5% of what was visited (each band is a separate median, which is why they do not add up to 100).
Read another way: most of your non-branded visibility lives in the middle of the results page, which is exactly the band where CTR dropped in more accounts this year. If your non-branded organic traffic is falling and your positions are not moving, part of the answer may be there.
What is a good organic CTR?
It depends on position, brand and the type of search, so the useful question is whether your CTR is fine for the spot where you appear. On this curve, 3% in second place without brand is the median, and 1% in sixth is above it.
At the whole-site level, mixing all positions, Ahrefs in Spanish published in July 2026 that a good organic CTR usually sits between 1% and 2%. My site had 0.42% over the last 90 days, at an average position of 15, which fits a site that shows up a lot on Google's second page.
Watch out for an effect that confuses a lot of teams: when you start appearing in more searches (more content, new keywords in low positions), impressions grow faster than clicks and total CTR drops. That can be good news, and it is why this metric should always be compared at the same position.
What moves CTR at each position?
The factors every guide repeats are real, though they move less than the mix of queries:
-
The title and meta description: they are your shop window on Google. A title that answers the search intent beats one written only for the keyword (the SEO title tag article covers how to measure each change).
-
Rich results: stars, prices or questions take up more space and lift CTR when they apply.
-
Ads: when there are four Google Ads at the top, the first organic result sits lower on the screen and gets fewer clicks.
-
AI summaries and other modules: maps, videos, the questions block and AI Overviews take part of the attention before the first blue link.
-
Intent: on an informational query the person compares; on a navigational one, they go straight in.
My recommendation is to start with queries that have lots of impressions in positions 4 to 10 and a CTR below their median. That is where a better title has more room to move traffic than in first place, where the curve is already high.
How do you calculate your site's CTR curve in Search Console?
It is the same measurement as the study, applied to a single domain. You can do it with the API, with the Search Analytics for Sheets extension or with Looker Studio (formerly Data Studio), and it ends in a pivot table in Google Sheets:
-
Download 28 days of web search data by query, with clicks, impressions and position. The interface stops at 1,000 rows, so on a large site the API or the extension is the better option.
-
Separate branded searches with a regular expression that includes your name, its variants and the most common typos. If you miss variants, first place comes out inflated.
-
Round each query's average position to the nearest whole number (1 to 10, and group 11 to 20).
-
Build the pivot table with position in the rows and the sum of clicks and of impressions in the values, and calculate CTR as one sum divided by the other. Do not average the percentages of each query, because one with 10 appearances would weigh the same as one with 10,000.
-
Drop positions with fewer than 100 impressions in the period: with that little data, the number changes with a couple of visits.
-
Repeat for another month and compare at the same position. Pick months after September 2025 and April 2026 so you do not cross the changes in how appearances are counted.
-
If you have very different sections (blog, categories, product pages), build one curve per section, because sitelinks and queries with maps or carousels can move a section's CTR a lot.
With that you have your own curve to compare against the 153-site one, and a number far more useful for your strategy than any industry average.
Methodology and limits of the study
We used Search Console (web search) from company websites in the US and Latin America that I have access to. Out of 357 properties we dropped unverified and subfolder ones and kept one per domain: 229 candidates.
With criteria set before looking at results (1,000 or more impressions and 100 or more non-branded clicks in each period), 153 remained. The main period was October 4 to 31, 2026 and the comparison, May 3 to 30, both 28 days long and after the changes in how impressions are counted.
For each site and period we pulled up to 25,000 queries. We excluded as branded the ones containing the domain name (without spaces or hyphens). Position is the rounded average position, and CTR by position was calculated as clicks divided by impressions, only with 100 or more impressions at that position in both periods.
We publish per-site medians with p25 and p75. The change is the median of the October/May ratio and the share of accounts where it dropped. The country is the one that brings the most clicks in October (US 90, Latin America 27, other 36).
The limits: it is a convenience sample, mostly US, that does not represent the whole web. Rounded position mixes days in neighboring spots, and anonymized queries do not appear. In 59 large domains the 25,000-query cap leaves out part of the long tail, although without them (90 sites) the pattern repeats: a stable top 3 and fifth place dropping in 61%.
The domain-based filter lets parent brands and products with other names slip through, and AI Overviews and AI Mode links cannot be separated. The Latin America cut (27 sites) clears the minimum of 20 but is small. And all of this is a comparison, with no proven causality.
What I would do with this data
If I had one afternoon, I would build my non-branded CTR curve with the steps above and put it next to the 153-site one.
Then I would look for the queries with the most impressions in positions 4 to 10 and a CTR below the median for their spot, and I would start with the titles of those pages.
And to prioritize new content, I would use the real CTR of the position it is reasonable to reach instead of the 30% from the tables, because that changes quite a bit how much traffic each keyword is worth. All of it can be calculated for free, before asking for budget for next year's SEO plan.