Google gives every website owner a free report on exactly how their site performs in search. Most owners open it once, see four big numbers across the top, decide the site is doing fine or doing badly, and never go back. The trouble is that two of those four numbers are averages — and both of them routinely say the opposite of what is really happening.
This article uses real figures from two sites we host. One of the customers agreed to be named. The other preferred not to be, so their numbers are here and their name is not.
The four numbers, in plain words
Open Google Search Console, pick your site, click Performance. Across the top:
| Clicks | How many people clicked through to your site from a Google result. The only one of the four that is a count of something real. |
|---|---|
| Impressions | How many times a link to your site appeared in someone's results. Appeared — not was read, not was noticed. |
| Average CTR | Clicks divided by impressions, across everything. |
| Average position | Where your site ranked, averaged over every search term and every time it appeared. |
Two counts and two averages. The averages are the ones that mislead, and they mislead in different ways.
Average position averages things that have nothing to do with each other
BBBOX builds quick-assembly modular homes in Taiwan. Their site runs on EZsite. Over the three months to mid-October 2026, Search Console reports:
| Clicks | 3,380 |
|---|---|
| Impressions | 102,000 |
| Average CTR | 3.3% |
| Average position | 6.5 |
Position 6.5 reads like "we are near the bottom of page one". It is not a place on a page. It is the mean of hundreds of different searches, some where the site ranks first and some where it ranks twentieth, weighted by nothing you care about. Click the Queries tab underneath and that single number falls apart into this:
| Search term | Impressions | Clicks | CTR | Position |
|---|---|---|---|---|
| 組合屋申請建照費用 (what a prefab building permit costs) | 458 | 128 | 27.9% | 1.3 |
| bb box組合屋 (their brand plus the product) | 112 | 55 | 49.1% | 1.0 |
| 組合屋價格 (prefab house price) | 5,092 | 165 | 3.2% | 5.8 |
| 組合屋 (prefab house) | 10,666 | 214 | 2.0% | 8.5 |
| 貨櫃屋 (container house) | 3,957 | 34 | 0.9% | 8.3 |
Read the first row against the last one, because that comparison is the entire point of this article.
A question about permit costs put the site in front of people 458 times and brought 128 of them in. The generic phrase "container house" appeared 3,957 times — nearly nine times the exposure — and brought 34.
One-ninth of the impressions, close to four times the visitors. Not because one word is magic, but because of what each search means. Somebody typing "container house" is browsing a category. Somebody typing "what does a prefab building permit cost" has a plot of land and a worry, and the page that answers the worry gets the click — 27.9% of the time, which is a rate no amount of ranking for a broad word will produce.
The site average, 3.3%, contains both of these and describes neither. You cannot act on 3.3%. You can act on "the permit-cost page earns 28% and there are probably ten more questions like it we have never written down".
One irrelevant search can poison your whole CTR
The second site belongs to a household cleaning brand we build for. Same three months:
| Clicks | 1,400 |
|---|---|
| Impressions | 112,000 |
| Average CTR | 1.3% |
| Average position | 6.8 |
Ranking near the top and converting at 1.3% looks like a page that is being seen and ignored. It would be easy to spend a month rewriting titles on the strength of that number.
Then you open the Queries tab and find a single search term — a Malay phrase for felt fabric, which the brand does not sell and has never sold — sitting at 21,902 impressions and 32 clicks. A 0.1% click rate on one-fifth of every impression the site received.
Google is showing this site to twenty-two thousand people looking for craft material. Those people are not going to click, and nothing the brand writes will make them. They are not customers; they are weather.
Set that one row aside and the same site reads completely differently. The searches that are actually about the brand behave like this:
| Search term | Impressions | Clicks | CTR | Position |
|---|---|---|---|---|
| brand name + "malaysia" | 107 | 33 | 30.8% | 3.4 |
| brand name + "ingredients" | 223 | 30 | 13.5% | 6.9 |
| brand name alone | 2,017 | 186 | 9.2% | 5.5 |
Nobody has a problem here. The site converts the people who want it at between 9% and 31%. The 1.3% headline was measuring an audience the business never asked for.
This is the general rule, and it is worth keeping: an average taken over searches you do not want is not a measurement of your website. Before you accept any CTR figure, find out which query is supplying the impressions.
How to actually read the report
Six steps. It takes about fifteen minutes the first time.
1. Switch on the two metrics that are off by default
Search Console shows Clicks and Impressions when you arrive. Average CTR and Average position are boxes you have to click before their columns appear in the table below. Most people never click them, which is why most people only ever see half the report.
2. Set the date range to three months, not 28 days
Twenty-eight days is too short for a small business site to say anything. A quiet month and a public holiday will swamp any real pattern.
3. Sort the Queries tab by impressions, not by clicks
Sorting by clicks shows you what is already working. Sorting by impressions shows you what Google is already willing to show you for, which is where everything unclaimed is hiding.
4. Put every row into one of three buckets
| Seen, not clicked | High impressions, position inside the top 10, CTR under 2%. Google is offering you the traffic and your result is not persuading anyone. This is a title and description job — the one line of the search result you write yourself. |
|---|---|
| Nearly there | High impressions, position between 11 and 30. You are on page two, where clicks essentially do not happen. This is a content job: the page exists but does not answer the question fully enough. |
| Yours already | Low impressions, CTR above 10%. You own this search. The question is whether there are ten more questions like it that you have never written a page for. |
5. Read the Pages tab the same way
Queries tell you what people want. Pages tell you which of yours they land on. A page with many impressions and few clicks has a title problem; a page with no impressions at all has a different problem, which is the next step.
6. Open Indexing → Pages, which almost nobody does
This report lists every page Google decided not to put in its index, with the reason beside it. A page that is not indexed cannot rank, cannot be clicked, and will never appear anywhere in the Performance report — so it is invisible in exactly the place you were looking.
It is worth saying that this is not only a report for other people's sites. Reading these reports in October is how we found that a set of sitemap files on our own platform had quietly stopped updating — a file-ownership problem on the server, invisible from the websites themselves, which looked and behaved perfectly normally. Nothing on any page was wrong. The only place it showed at all was here. We fixed it the same week.
What to do with what you find
Three findings cover most of what this exercise turns up, and each has a different fix.
A high-impression page with a weak click rate
Rewrite its title and meta description so they contain the fact the searcher is after — a price, a location, an opening time, a straight answer. Not adjectives. In EZsite this is the SEO Setting panel on that page, filled in per language, so your Chinese and English visitors each get a sentence written for them rather than one translated from the other.
A question that converts above 20%
Write the sibling questions. BBBOX earns 27.9% on a permit-cost question; the people asking it will also ask about delivery, about foundations, about how long assembly takes. Each of those is a page, and each page is a new entry point.
A high-impression search that is not your business at all
Leave it alone. You are not going to win it and you do not want it. Note it down so you remember why your average CTR looks the way it does, and judge the site on the searches that belong to you.
Do this today
- Open Search Console and confirm your site is verified. If it is not, that is the first job — the data only starts accumulating from the day you verify, so a site added today will have nothing to show you until next month.
- Performance → 3 months → switch on Average CTR and Average position.
- Sort Queries by impressions and read the top twenty rows.
- Mark each one: seen-not-clicked, nearly-there, or already-yours.
- Fix the titles and descriptions on the first group this week.
- Open Indexing → Pages and read the reasons listed there.
None of this needs a tool you have to pay for, and none of it needs a developer. It needs somebody to open the report and look past the four numbers at the top.
If your site is on EZsite, your sitemap is generated and submitted for you, and every page has its own title and description fields per language. If you would like us to go through your Search Console with you and tell you which of these three buckets your pages are in, send us your website address.