Over the last 90 days our site earned 262,099 impressions in Google and 1,570 clicks. That is a 0.60 percent click-through rate. Our best-ranked content does even worse: the informational articles that sit on page 1 for questions people ask tens of thousands of times a month convert at 0.27 percent, roughly ten times below the normal click-through rate for those positions. This is a first-party look, from our own Google Search Console, at what AI Overviews have done to organic clicks. Ranking is no longer the finish line.
📄 Download the full study as a PDF
Prefer a formatted, citable paper? This study is also available as a journal-style PDF, "Ranking Without Clicks," with the full page-level data and the raw dataset as CSV. → Download the PDF (journal format)
TL;DR
- In 90 days (April to July 2026) cleanor.app got 262,099 Google impressions and 1,570 clicks: a 0.60 percent site-wide CTR.
- Our informational articles rank well, averaging positions 7 to 9 (page 1) on big queries, yet earn only 0.27 percent CTR across 160,182 impressions, about ten times below the pre-AI norm for those positions.
- Tool pages, whose queries rarely trigger an AI Overview, earn 1.12 percent, roughly four times the CTR of the articles at similar positions.
- The worst-hit pages are classic AI-Overview queries: "what is the appdata folder" got 22,966 impressions and 50 clicks (0.22 percent); "clear cache vs clear data" got 5,030 impressions and 5 clicks (0.10 percent).
- The lesson is not to chase higher rankings. It is to be the source the AI answer cites. That is why we now publish original data instead of more explainer articles.
Ranking on page 1 stopped meaning traffic
For twenty years the deal was simple: rank on the first page and you get clicks. A published organic CTR benchmark puts a top-10 result somewhere between 2 and 30 percent depending on position, with even a modest page-1 spot around 3 percent. Our data breaks that deal. Across our informational articles that rank in Google's top 10, the click-through rate is not 3 percent. It is a quarter of a percent. We are on the page, and almost no one arrives.
The pages that prove it
These are real pages on cleanor.app, with their 90-day Search Console numbers. Each ranks on page 1 for a high-volume question, and each earns almost nothing:
| Query the page ranks for | Impressions | Avg position | Clicks | CTR |
|---|---|---|---|---|
| what is the appdata folder | 22,966 | 8.7 | 50 | 0.22% |
| why does youtube take so much storage | 10,755 | 7.5 | 22 | 0.20% |
| why is system data so high on iphone | 8,639 | 8.2 | 19 | 0.22% |
| why is chrome taking so much space | 6,415 | 7.4 | 39 | 0.61% |
| clear cache vs clear data | 5,030 | 8.1 | 5 | 0.10% |
| why is my phone still full after deleting photos | 3,548 | 9.6 | 5 | 0.14% |
| how to clean up telegram storage on iphone | 3,443 | 8.7 | 2 | 0.06% |
A page at position 8 would historically expect about 3 percent CTR. "How to clean up Telegram storage" sat at position 8.7 for 3,443 impressions and got two clicks. That is not a ranking problem. The ranking is fine. The clicks are gone before the user ever reaches the blue links.
It is the question, not the position
The clearest evidence that this is an AI-answer effect, and not just our pages being unappealing, is the split between our own page types. Our tool pages (image converters, unit calculators, and so on) earn 1.12 percent CTR: low, but roughly four times our articles. The difference is the query. Nobody's "webp to png converter" search is satisfied by a paragraph of AI text; they need the tool, so they click. But "why is system data so high" is answered completely by the AI Overview and the People Also Ask box sitting above every link. The question is resolved on the results page, and the click never happens.
Broken down by position, our page-1 CTR runs 2 to 9 times below the pre-AI benchmark at every rung: about 1.9 percent at position 7 (against a 3.5 percent norm), 1.3 percent at position 8 (against 3 percent), 0.5 percent at position 9 (against 2.5 percent). The decline is not concentrated at one position. The whole curve has been flattened for question-style queries.
Why AI Overviews collapse the click
An AI Overview is a generated answer Google places at the very top of the results, above the organic links, for queries it judges informational. Below it sits People Also Ask, then finally the blue links. For a query like "is it safe to delete the appdata folder," the AI Overview gives the yes-or-no and the reasoning in three sentences. The user has their answer. There is no reason to scroll past it to a page ranked eighth, so a page that would have collected a few percent of 23,000 impressions now collects a fraction of one percent. The page did nothing wrong. The results page changed shape.
What this means for anyone doing SEO
If you publish informational content, the ranking dashboards are now lying to you by omission. You can rank on page 1, watch impressions climb, and receive almost no traffic, because your position is below an answer the user never scrolls past. Three things follow from our data:
- Track clicks and CTR, not just position. A page-1 ranking on an AI-Overview query is not the win it used to be.
- Optimize to be cited, not just to rank. The traffic that remains flows to the sources the AI answer names and links. Clear, quotable, well-structured facts with a citable origin are what get pulled into the answer.
- Own something the AI cannot generate. Original data, first-hand measurements, and proprietary numbers are the content an AI Overview has to attribute, because it cannot make them up. That is the reasoning behind Cleanor Labs research: we would rather be the cited source of a number than the eighth link under an answer.
Method and caveats
Data is from Google Search Console for the domain property sc-domain:cleanor.app, 90 days ending 1 July 2026 (Search Console finalizes data about three days back). Site-wide totals (262,099 impressions, 1,570 clicks) are from the date dimension, which is complete. Page-level and query-level figures are impression-weighted from the page and query dimensions; Google anonymizes rare queries, so query-level sums undercount the true total, which is why we lead with page-level and site-wide numbers. "Normal" CTR is a published pre-AI organic benchmark (blended desktop and mobile, e.g. Advanced Web Ranking), used only as a reference line, not as our own historical baseline. The per-page dataset is available as CSV alongside this report.
Frequently asked questions
Do AI Overviews reduce click-through rate? In our first-party data, dramatically for informational queries. Our articles rank on page 1 (average positions 7 to 9) yet earn about 0.27 percent CTR, roughly ten times below the normal rate for those positions, because the AI Overview answers the question above the organic links.
What is a normal click-through rate for a page-1 Google result? Published pre-AI benchmarks put it around 3 percent for a middle-of-page-1 position and higher near the top. On question-style queries that now trigger AI Overviews, our measured CTR at those same positions is a fraction of that.
If I rank on page 1, why am I getting no traffic? Most likely your result sits below an AI Overview and a People Also Ask block that answer the query, so users never scroll to the organic links. Ranking is intact; the click is intercepted higher on the page. Check CTR by query in Search Console, not just position.
How do you get traffic when AI Overviews answer everything? Focus on queries that need a page (tools, purchases, first-hand experiences) and on being the source the AI answer cites. Original data and proprietary numbers are hard for an AI to answer without linking you, which is the most reliable way to earn both citations and clicks.