Ranking and being clicked are two different problems

If your website isn’t bringing anyone in, the assumption is almost always that you don’t rank — that you’re buried on page four and the answer is more SEO.

Sometimes that’s right. Often it isn’t, and you can find out which in about ten minutes without paying anybody. The two things fail separately: appearing in the results and being chosen from them are different problems with different fixes, and only one of them is expensive.

Impressions and clicks

An impression means your page appeared in somebody’s search results. A click means they picked it.

From your analytics, a page that appears two hundred times and is never clicked looks exactly like a page nobody ever finds. Same flat zero. But they are opposite situations: in one, the hard, slow, expensive work is already done and you are losing people in the last inch; in the other you haven’t started.

Hardly anybody checks, because impressions aren’t the number people look at. Visitors are.

The check

You need Google Search Console. It is free, and if you have never set it up, that is the first job — it wants you to prove you own the domain, usually with a DNS record.

Then: Performance → check that Impressions and Average position are both switched on. They are toggles above the chart, and average position is the one people leave off. Set the range to the last 28 days, open the Pages tab, and sort by impressions.

Now read down the list for rows where the average position is under ten and the clicks are at or near zero. Every one of those is a page that Google is already putting in front of people who then choose somebody else.

What it looked like on our own site

We got API access to our own Search Console data yesterday and ran exactly this. The honest numbers, 28 days to 24 August 2026, on a blog that is about two months old:

1,153 impressions. 15 clicks. Average position 12.8.

Small, and we’re not going to dress it up. But the interesting part is the distribution. 52 pages had an average position of ten or better. Forty-four of those got no clicks at all. Between them the top-ten pages picked up 960 impressions and 12 clicks.

The individual cases are sharper than the total. A post about chasing unpaid invoices: 136 impressions, average position 9.8, zero clicks. One on customer service: 44 impressions at position 6.1, zero. One on the EU AI Act: 50 impressions at position 8.7, zero.

Meanwhile the best-performing post on the site sat at position 8.3 and converted 83 impressions into 3 clicks. Same neighbourhood of the results page, different outcome. Nothing about ranking explains that gap.

So what causes it

Three candidates. We have not tested which applies to us, and I’m not going to pretend otherwise.

1. Your title and description don’t match what they typed. Someone searches a specific worry; your link offers a general topic. They scroll. This is the cheapest thing in all of search marketing to change and the one most often left as whatever the page was called when it was written.

2. Something above you already answered it. An AI overview, a featured snippet, a map panel. The person got what they needed without leaving. We wrote about that shift in when Google answers instead of linking — and it appears to be happening to us now, which is a fair result.

3. The query never wanted a click. Someone checking your opening hours or your phone number is done as soon as they see it. That is a satisfied customer, not a lost one.

Only the first is fully in your hands. It is also free.

What to actually do

Take the three or four pages with the most impressions and no clicks. Not all of them — the ones with volume.

Look at what people are actually typing. The Queries tab, filtered to that page, tells you the words they used. Then say those words in your title.

Write a description that answers rather than teases. “What we do about late payers” is a brochure. “How to chase an overdue invoice without souring the relationship” is an answer. Google may ignore what you write and pull its own text from the page — but it uses the description when it fits the query better, so it is still worth writing.

Search the term yourself and look at what you’re competing with. Ten seconds. If an AI summary and four adverts sit above your link, the title is not your problem and no amount of rewriting fixes it.

Then leave it a month. Changes take time to show, and Google re-crawls on its own schedule — which we went through in what “update my website on Google” actually means.

What argues the other way

At small numbers, zero clicks may just be arithmetic. Most of our 44 zero-click pages had single-digit impressions, where getting nothing is unremarkable rather than diagnostic. The pattern across many pages is worth something; any one page with eleven impressions is worth almost nothing. Don’t rebuild a page on a sample of eleven.

Average position is an average. A page listed at 6.1 was probably never at 6.1 for anyone — it was third for one query and fifteenth for another. It is a useful sorting tool, not a description of what any real person saw.

Search Console does not tell you everything, and the numbers don’t fully reconcile. Google withholds queries below a privacy threshold: on our site only about a quarter of impressions had a query attached at all. And the per-page figures added up to more than the site total, because the two are aggregated differently. It is good enough to decide what to work on. It is not good enough to quote to three decimal places, and anyone who does is overselling it.

And clicks are not the point either. Traffic to a page that never turns into an enquiry is a vanity number wearing a hat. Fix the pages where a click would actually be worth something.

The same question is now being asked about AI assistants rather than search results, and the answer there is less settled than the headlines suggest — we went through one widely-reported example in Reddit nearly vanishing from ChatGPT’s citations.

The short version

Before you spend money on ranking better, spend ten minutes finding out whether you already rank. Turn on impressions and average position in Search Console, sort your pages, and look for the ones sitting in the top ten with nothing to show for it. On our own site that was 44 pages out of 52. If it’s happening to you, the expensive work is already done and what’s left is a sentence.

The figures above are our own Google Search Console data for lolasquared.com, 28 July to 24 August 2026, pulled via the Search Console API on 25 August 2026. They describe one small, two-month-old site and are offered as a worked example, not as a benchmark for anyone else. Two caveats we would want stated if these were someone else’s numbers: only about a quarter of the impressions had a query attached, the rest being withheld by Google’s privacy threshold; and the per-page totals sum to more than the site total, because Search Console aggregates the two differently. No third-party click-through-rate benchmarks are quoted here, because the published ones vary too widely to be worth citing.