---
title: "Position 2.6. 4,204 Impressions. One Click."
description: "Our launch-schedule page ranked near the top of Google and earned one click from 4,204 impressions. Here’s what that row changed about how I check search visibility."
url: "https://groas.com/post/position-2-6-4-204-impressions-one-click"
image: "https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/36651900-ed38-4a8c-943b-ff92ccad7112.png"
published: "2026-10-17T05:25:44.140Z"
modified: "2026-10-17T05:25:44.214Z"
---

October 17, 2026 · 10 min read

# Position 2.6. 4,204 Impressions. One Click.

[Alexander PerelmanHead Of Product @ groas](https://groas.com/author/alexander-perelman)[LinkedIn](https://www.linkedin.com/in/alexander-433793253/)

![Tabletop model town square: a crowd takes leaflets from a big answer booth, while one figure in yellow walks to the open, empty kiosk just behind it.](https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/36651900-ed38-4a8c-943b-ff92ccad7112.png)

In this article

1. [Scene 2: A launch question meets a crowded results page](#scene-2-a-launch-question-meets-a-crowded-results-page)
2. [Scene 3: I checked the obvious suspects first](#scene-3-i-checked-the-obvious-suspects-first)
3. [Scene 4: An answer can end the search before the click](#scene-4-an-answer-can-end-the-search-before-the-click)
4. [Scene 5: Another page on our site gave me a contrast](#scene-5-another-page-on-our-site-gave-me-a-contrast)
5. [Coda: The row I would pull this afternoon](#coda-the-row-i-would-pull-this-afternoon)

Tuesday morning, cold coffee, Search Console export open on the second monitor. I was sorting our own blog by impressions, looking for something to brag about, when one row stopped me: the campaign launch schedule piece. **Position 2.6. 4,204 impressions. One click.** I read it twice because I assumed I had filtered the dates wrong.

I have killed ad groups for less. If a keyword spends and buys nothing, you do not call it a win because the impression share looks good. You ask where the clicks went. This time the account was ours, and the page had cleared the hurdle every SEO checklist tells you to clear. Google was showing it near the top. Almost nobody opened it.

## Scene 2: A launch question meets a crowded results page

The search behind that row was the sort of question I like: someone about to launch Google Ads wants to know what the first 2 to 4 weeks should look like. Say you are spending $20k a month and you type that in. You want a checklist, a timeline, someone to tell you when to touch bids and when to leave the account alone. **The intent was not hard to recognise.**

The results page can answer plenty of that question before a searcher reaches a blue link. There may be ads at the top and an AI answer that lays out a schedule in a few neat steps. Our guide can rank below that answer while doing the slower, more detailed work of explaining the same schedule. I cannot prove that all 4,204 impressions came with an AI box. I was not sitting beside 4,204 searchers, and Search Console does not replay their screens. But the possibility was hard to ignore: the reader might have got enough of an answer without needing our page.

This is not just a complaint about where Google puts the furniture. In one analysis of 300,000 keywords, position-one click-through fell [from 7.3% to 1.6% when an AI Overview appeared](https://www.marketscale.com/industries/marketing-tech/googles-ai-overviews-roughly-tripled-on-brand-name-searches-in-september). A separate Pew panel of 68,879 searches found [only 8% of visits with an AI summary produced a website click versus 15% without](https://bosio.digital/articles/ai-overviews-zero-click-search). Neither study diagnoses our row. Both explain why a high position and a visit can no longer be treated as the same event.

I went back to the export. One click still looked like a mistake.

## Scene 3: I checked the obvious suspects first

When an ad group looks dead, I start with the boring stuff. Boring breaks more accounts than clever causes. So I ran the same inspection on our page: position, then the result a person would actually see, then the question the result was supposed to answer.

**Position was not the missing piece.** An average of 2.6 puts the page near the top across those impressions, though an average cannot tell me its exact position on every search. In an ordinary set of results, that would be encouraging. [Average organic click-through runs around 16% for position 2 and 11% for position 3](https://www.thecssagency.com/blog/impressions-vs-clicks). Apply rates like those mechanically to 4,204 impressions and you start imagining hundreds of visits. Search pages are not that tidy. Query mix, ads and answer features can all get between an average position and a click. Still: one visit was enough to make me look twice.

Then the title and snippet. I used to tell clients a rewrite fixes everything. I was wrong often enough to stop saying it. Our title was plain and matched the query. The URL was clean; the date was fresh. Could a better meta description have earned more than one click? Sure. But I could not persuade myself that a line of snippet copy was the whole explanation for a gap this wide. One agency described [a commercial page at position 6.0 with 887 impressions and zero clicks, and another at 5.77 with hundreds of impressions and the same zero](https://bosio.digital/articles/ai-overviews-zero-click-search). Visitors who did arrive engaged normally. Different pages, same awkward question: what happened before the visit?

That question took me away from the page itself. No amount of staring at its paragraphs would show me what a searcher saw above the link.

## Scene 4: An answer can end the search before the click

I keep a rule from client work. When tracking is clean and the offer is fine, stop blaming the ad and look at the page the click lands on. Here the visit barely happened, so I flipped the rule: stop looking at our page and look at what might sit above it. **If the results page gives someone a usable week-by-week launch plan, our more detailed version has to earn a click they may no longer need.**

Pew found [only 1% clicked a link inside an AI summary](https://bosio.digital/articles/ai-overviews-zero-click-search), and more than a quarter of those sessions ended right there. That does not tell me whether our guide appeared in a summary or whether anyone read it there. It does tell me why “we rank” is a thin answer to “did anyone choose us?”

Then there is a second distinction. An answer can use a page as a source, name a brand in its text, do both, or do neither. Tracking across 3,119 terms found [cited brands earned 35% higher organic clicks and 91% higher paid clicks than uncited brands on the same queries](https://bosio.digital/articles/ai-overviews-zero-click-search). That is an association, not a promise that one citation will fix our traffic. But it made the question sharper. A result can be counted as visible without being the source the answer chooses to show.

**Ranked but not chosen** was the phrase I wrote beside the row. I could not add “not cited” to it yet. Search Console had given me impressions, position and a click, not a record of every answer that appeared. I needed to check the answer separately.

## Scene 5: Another page on our site gave me a contrast

A week later I pulled our bot logs for a different reason and found something I had not been looking for. Our Google Ads updates changelog, a boring dated page I barely think about, had 377 recorded fetches during the period in which the launch schedule page earned its single click. Same domain. Same Search Console account. One page had an impressive search row and almost no human visits; the other kept showing up in machine activity. **Those logs showed fetching, not proof that an AI answer cited the changelog 377 times.** The contrast was still worth examining.

![Monitor showing an AI answer above a blue search result](https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/d62f0d85-fdbe-4e37-b43f-76563e2fff49.png)

The pages were built differently. The launch guide tells its story in long paragraphs. The changelog gives short dated entries, each answering one thing in one place. If a system needs a compact passage to quote, I can see which page makes that job easier. I had spent years making pages readable for people. I was looking at evidence that I also needed to make the answer easy to find _within_ the page, without turning the whole thing into chopped-up machine food.

That distinction matters because there is no magic switch to flip. Google needs [indexing and snippet eligibility for AI Overviews, not a special file or tag](https://www.itomic.com.au/how-to-get-your-business-into-google-ai-overviews/). Pages practitioners see cited tend to rank already and give a [direct one- or two-sentence answer near the top, supported by clear headings, lists, tables and FAQs](https://www.itomic.com.au/how-to-get-your-business-into-google-ai-overviews/). I am not claiming the changelog’s structure caused every fetch, or that the same structure would have rescued the guide’s 4,204 impressions. I am saying the comparison gave me a specific edit to try instead of another vague instruction to “improve visibility.”

## Coda: The row I would pull this afternoon

If I were looking for the same problem on another site, I would start where I started on ours: Search Console. Set Performance to the last 28 days, sort Pages by impressions and scan for a high-impression page with very few clicks. An average position under 5 and click-through under half a percent would get my attention. Then open the queries behind it. **A page row tells you where to investigate; it does not tell you which search or result layout explains the gap.**

Export the row. Circle it if that helps. I circled ours. The point is to stop treating it as a ranking problem by default. Our page had ranked. The thing I did not know was whether searchers were being answered on the results page, whether an AI answer named us, and whether it linked to us when it needed a source.

For that, I would ask the page’s question in plain buyer words where buyers now ask it: Google with the AI box expanded, ChatGPT, Gemini and Perplexity. Log out first. Run the question two or three times and record what comes back, rather than keeping the answer you liked best. Practitioners doing this systematically use a [fixed set of 20 to 50 buyer-worded prompts, recording mentions, positions and links by engine before calculating mention rate and share of voice weekly](https://ujeebu.com/blog/what-is-ai-visibility). One check is a glimpse, not a trend. Brand lists can change order between runs; one analysis found the same order returned [about one time in a thousand](https://brand24.com/blog/ai-brand-visibility-tracking-tools/). Three runs would not settle everything, but they would stop me from treating one answer as permanent.

![Spreadsheet with a high-impression, low-click row circled in red](https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/9f20e468-5514-43c0-b765-bbf9e2aa6b0f.png)

When I run those checks, I write down two separate things: **Were we named? Were we linked?** An answer can recommend a brand in the text and cite someone else underneath. It can also cite a page while recommending a competitor. ChatGPT pulls brand sites directly more often, while AI Overviews and Gemini lean heavily on third parties, so reviews, forums and comparisons can matter to [mentions and citations in different ways](https://ujeebu.com/blog/what-is-ai-visibility). I used to tell clients to fix the page and wait. Sometimes the work sits off the site. First I want to know which problem I have.

For our launch schedule page, I made two edits I could defend to a client. I put a direct, two-sentence answer at the top, covering the weeks in order, so someone can understand the timeline before reading the explanation. Then I tightened the structure underneath: H2s, a simple table for weeks one through four, and three FAQs in buyer wording. No redesign, no new post, no begging for links. I cannot promise those edits will earn a citation. Anyone promising that from one test is selling you something. I can say the guide now makes its answer easier to locate, the way the changelog already did.

I keep the export pinned above my desk, the 4,204 row highlighted in yellow. Once a week I search the page’s main query with the AI box open, then ask it in ChatGPT and Gemini while logged out. I write down whether we were named and whether we were linked. When neither happens for three weeks straight, the page goes on the rewrite list. When an answer names us but links elsewhere, I look beyond the page at the reviews and comparisons it chose to cite. It is a smaller, more useful routine than admiring an average position.

It is also the part I stopped doing entirely by hand. groas runs that tracking for our site as it does for clients, [across Google and AI answers from one program](https://groas.com/for-businesses), with changes logged and a strategist accountable for them. I still like checking one query myself, the way I still like pulling one search term report myself. Old habits. But a Tuesday-morning sort of Search Console no longer gets to tell me the whole story.

If you have twenty minutes, spend them on one page, not the whole site. Pick the high-impression row with the fewest visits, run its question through Google and one AI engine, and write down whether you were named and linked. Then give the page one paragraph that answers the question directly. I left our export open when I finished, yellow highlight still on the 4,204 row, coffee ring drying beside the keyboard.

I still check rankings. I just stopped treating them as the receipt. That Search Console tab is open as I write this, cursor blinking on the next row down.

## Frequently Asked Questions

### Why does a page rank near the top of Google but get almost no clicks?

Because the results page may answer the searcher's question before they reach a blue link, through ads and AI summaries. Studies cited show position-one click-through fell from 7.3% to 1.6% when an AI Overview appeared, and only 8% of searches with an AI summary produced a website click versus 15% without.

### Can Search Console tell you exactly why a page isn't getting clicks?

No. A page row tells you where to investigate, but it does not tell you which search or result layout explains the gap. It gives impressions, average position and clicks, not a record of every AI answer that appeared. An average position like 2.6 also cannot tell you the exact position on every individual search.

### How can I check whether AI answers mention or link to my site?

Search the page's main query in plain buyer wording in Google with the AI box expanded, and in ChatGPT, Gemini and Perplexity. Log out first, run it two or three times, and record what comes back rather than keeping the answer you liked best. Write down separately whether you were named and whether you were linked.

### What's the difference between being mentioned in an AI answer and being cited in it?

An answer can recommend a brand in its text and cite someone else underneath, or cite a page while recommending a competitor. Tracking across 3,119 terms found cited brands earned 35% higher organic clicks and 91% higher paid clicks than uncited brands on the same queries, though that is an association, not a guarantee.

### What makes a page more likely to be used as a source in AI Overviews?

Pages practitioners see cited tend to rank already and give a direct one- or two-sentence answer near the top, supported by clear headings, lists, tables and FAQs. Google needs indexing and snippet eligibility for AI Overviews, not a special file or tag.

### How did they fix the high-impression, low-click page?

With two edits: a direct two-sentence answer at the top covering the weeks in order, so readers understand the timeline before the explanation, and a tightened structure underneath using H2s, a simple table for weeks one through four, and three FAQs in buyer wording. No redesign or new post, and no promise the edits will earn a citation.

### What routine should I follow to track AI visibility over time?

Once a week, search the page's main query with the AI box open, then in ChatGPT and Gemini while logged out, recording whether you were named and linked. When neither happens for three weeks, put the page on the rewrite list. When an answer names you but links elsewhere, look beyond your page at the reviews and comparisons it cites.

## Pay For Results, Not For Hours

Businesses buy the outcome, agencies resell it, and groas answers for it either way.

[See If You Qualify](https://groas.typeform.com/to/xC1bQNUT)

## Structured data

```json
{"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://graph.groas.com/entity/groas","name":"groas","url":"https://groas.com/","logo":{"@type":"ImageObject","url":"https://groas.com/icon-512.png","width":512,"height":512},"sameAs":["https://www.linkedin.com/company/groas/"]},{"@type":"WebSite","@id":"https://groas.com/#website","name":"groas","url":"https://groas.com/","inLanguage":"en-US","publisher":{"@id":"https://graph.groas.com/entity/groas"}},{"@type":"WebPage","@id":"https://groas.com/post/position-2-6-4-204-impressions-one-click#webpage","url":"https://groas.com/post/position-2-6-4-204-impressions-one-click","name":"Position 2.6. 4,204 Impressions. One Click.","description":"Our launch-schedule page ranked near the top of Google and earned one click from 4,204 impressions. Here’s what that row changed about how I check search visibility.","inLanguage":"en-US","isPartOf":{"@id":"https://groas.com/#website"},"about":{"@id":"https://graph.groas.com/entity/groas"},"primaryImageOfPage":{"@type":"ImageObject","url":"https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/36651900-ed38-4a8c-943b-ff92ccad7112.png"},"breadcrumb":{"@id":"https://groas.com/post/position-2-6-4-204-impressions-one-click#breadcrumb"},"dateModified":"2026-10-17T05:25:44.214Z","mainEntity":{"@id":"https://groas.com/post/position-2-6-4-204-impressions-one-click#article"}},{"@type":"BreadcrumbList","@id":"https://groas.com/post/position-2-6-4-204-impressions-one-click#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://groas.com/"},{"@type":"ListItem","position":2,"name":"Blog","item":"https://groas.com/blog"},{"@type":"ListItem","position":3,"name":"Position 2.6. 4,204 Impressions. One Click.","item":"https://groas.com/post/position-2-6-4-204-impressions-one-click"}]},{"@type":"BlogPosting","@id":"https://groas.com/post/position-2-6-4-204-impressions-one-click#article","url":"https://groas.com/post/position-2-6-4-204-impressions-one-click","mainEntityOfPage":{"@id":"https://groas.com/post/position-2-6-4-204-impressions-one-click#webpage"},"isPartOf":{"@id":"https://groas.com/blog#blog"},"headline":"Position 2.6. 4,204 Impressions. One Click.","description":"Our launch-schedule page ranked near the top of Google and earned one click from 4,204 impressions. Here’s what that row changed about how I check search visibility.","datePublished":"2026-10-17T05:25:44.140Z","dateModified":"2026-10-17T05:25:44.214Z","image":{"@type":"ImageObject","url":"https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/36651900-ed38-4a8c-943b-ff92ccad7112.png"},"author":{"@id":"https://groas.com/author/alexander-perelman#person"},"publisher":{"@id":"https://graph.groas.com/entity/groas"},"keywords":"page, answer, impressions, click, position, 204 impressions, search, 204","wordCount":2021,"inLanguage":"en-US"},{"@type":"Person","@id":"https://groas.com/author/alexander-perelman#person","name":"Alexander Perelman","jobTitle":"Head Of Product @ groas","description":"Ex Goldman Sachs and Ex Stanford Computer Science","image":"https://groas.com/media/blog/4f17cac3a81acc224cc1b2cfab8f94bbc295a3089d13e8421125771ef8d4064f.jpg","worksFor":{"@id":"https://graph.groas.com/entity/groas"},"sameAs":["https://www.linkedin.com/in/alexander-433793253/"]},{"@type":"FAQPage","@id":"https://groas.com/post/position-2-6-4-204-impressions-one-click#faq","mainEntity":[{"@type":"Question","name":"Why does a page rank near the top of Google but get almost no clicks?","acceptedAnswer":{"@type":"Answer","text":"Because the results page may answer the searcher's question before they reach a blue link, through ads and AI summaries. Studies cited show position-one click-through fell from 7.3% to 1.6% when an AI Overview appeared, and only 8% of searches with an AI summary produced a website click versus 15% without."}},{"@type":"Question","name":"Can Search Console tell you exactly why a page isn't getting clicks?","acceptedAnswer":{"@type":"Answer","text":"No. A page row tells you where to investigate, but it does not tell you which search or result layout explains the gap. It gives impressions, average position and clicks, not a record of every AI answer that appeared. An average position like 2.6 also cannot tell you the exact position on every individual search."}},{"@type":"Question","name":"How can I check whether AI answers mention or link to my site?","acceptedAnswer":{"@type":"Answer","text":"Search the page's main query in plain buyer wording in Google with the AI box expanded, and in ChatGPT, Gemini and Perplexity. Log out first, run it two or three times, and record what comes back rather than keeping the answer you liked best. Write down separately whether you were named and whether you were linked."}},{"@type":"Question","name":"What's the difference between being mentioned in an AI answer and being cited in it?","acceptedAnswer":{"@type":"Answer","text":"An answer can recommend a brand in its text and cite someone else underneath, or cite a page while recommending a competitor. Tracking across 3,119 terms found cited brands earned 35% higher organic clicks and 91% higher paid clicks than uncited brands on the same queries, though that is an association, not a guarantee."}},{"@type":"Question","name":"What makes a page more likely to be used as a source in AI Overviews?","acceptedAnswer":{"@type":"Answer","text":"Pages practitioners see cited tend to rank already and give a direct one- or two-sentence answer near the top, supported by clear headings, lists, tables and FAQs. Google needs indexing and snippet eligibility for AI Overviews, not a special file or tag."}},{"@type":"Question","name":"How did they fix the high-impression, low-click page?","acceptedAnswer":{"@type":"Answer","text":"With two edits: a direct two-sentence answer at the top covering the weeks in order, so readers understand the timeline before the explanation, and a tightened structure underneath using H2s, a simple table for weeks one through four, and three FAQs in buyer wording. No redesign or new post, and no promise the edits will earn a citation."}},{"@type":"Question","name":"What routine should I follow to track AI visibility over time?","acceptedAnswer":{"@type":"Answer","text":"Once a week, search the page's main query with the AI box open, then in ChatGPT and Gemini while logged out, recording whether you were named and linked. When neither happens for three weeks, put the page on the rewrite list. When an answer names you but links elsewhere, look beyond your page at the reviews and comparisons it cites."}}]}]}
```