How to Read Search Console Properly (clicks, impressions, CTR, position)

Full transcript

1,220 words · script draft, under review.

Intro

Search Console says the average position of a page is thirty-six. Page four. Hopeless, right?

Except the page ranks third for the search that matters. The thirty-six comes from somewhere else, and if you read the number at face value, you'd rewrite a page that's doing fine.

Search Console has only four metrics. They look simple. In this episode we read each one properly, on one of my own sites: what it counts, how the averages trick you, how to compare periods, and what "position" even means on a results page full of boxes.

The four metrics, exactly

First, what each number actually counts.

A click is a click from a Google results page that leads to your site. Not a visit, not a session. Just the click.

An impression is counted when a link to your site was on the results page the user loaded. It doesn't mean anyone saw it. A result at position fifty can get an impression if the user scrolled through more results.

CTR, click-through rate, is clicks divided by impressions. Nothing more.

And position is the position of your site's topmost result for that search. If two of your pages appear, only the higher one counts.

All four are counted per search, then summed or averaged over the period. And that's where it goes wrong, because averages behave in ways that don't feel intuitive.

Average position is impression-weighted

First the arithmetic, with made-up numbers. A page ranks third for query A, and gets a hundred impressions. It also ranks fortieth for query B, a broad search where people scroll deep, and gets nine hundred impressions.

The average position isn't the average of three and forty. It's weighted by impressions. Every impression votes once. So query B, with nine times more impressions, pulls the average to about thirty-six.

Now imagine the page starts appearing for a third query, at position sixty, with a thousand impressions. More visibility. And the average position gets worse.

That's the trap. Average position can fall because you gained new, deep rankings, not because you lost anything.

Now the same thing on real data. This comes from one of my own sites, through the Search Console API, and it's anonymized: the domain, the page URLs and the search terms are replaced, so we stay focused on the principles and not on the keywords. The numbers are untouched.

Each bar is one of the eight biggest queries for a single page. The dashed line is the page's average position: eight point six. The best query sits at about three. On real data the gap is smaller than in my made-up example. On this site, I found no page with a gap of more than ten positions. But the direction is the same. And look at the box: these eight queries cover only about a fifth of the page's impressions. The rest is spread across queries outside this chart, and they carry most of the weight in the average. So never read average position for a page on its own. Break it down by query, and look at position together with impressions.

CTR averages hide the same thing

CTR has the same problem. A page's CTR is total clicks divided by total impressions, so it's dominated by whichever queries have the most impressions.

On the same page, the top query, at about position three, gets around eighteen percent. Most of the others, between positions six and eleven, get close to nothing. The page total lands at under five percent, and looks like a problem.

One detail worth noticing: there's a second query at about position three with no clicks at all. The average would never show you that. It's a question for the results page itself, which we'll get to in a minute.

It's not a problem. It's a mix. CTR only means something when you compare it at a similar position. In episode eleven we'll build your site's own CTR curve, per position, and use that to find titles worth rewriting.

Filters and dimensions

Most reading mistakes are fixed by one filter.

Search type. The default is web search, but images, video and news are separate. Make sure you know which one you're looking at.

Country and device. A site can rank well on desktop in one country and poorly on mobile in another. The total hides both.

Query and page filters. Filter on one page, then look at its queries. Or filter on one query, then look at which pages appear for it. That second view is how you'll spot two pages competing for the same search, which is episode twelve.

And one warning before you start adding numbers together. The totals in the chart are counted per site. The rows in the page table are counted per page. Some queries are also hidden for privacy. So the table rarely adds up to the chart. That's next episode's whole topic.

Comparing periods

Comparing periods is where many false alarms come from.

Compare equal lengths, and whole weeks. This is my site, day by day, for four weeks. Look for a weekly rhythm in your own chart. On this site it's weak, the days bounce around without a clear weekend dip. But many sites have one, and then a period that starts on a Monday and one that starts on a Saturday aren't comparable.

Leave out the last two or three days. Search Console data isn't final yet, and a period that ends yesterday will almost always look like a drop.

And think about season. Last twenty-eight days against the twenty-eight before is fine for spotting a sudden change. For a trend, compare with the same period last year. Episode thirteen goes deeper into separating seasonality from real change.

What "position" means on a page full of boxes

Finally, position on a modern results page. Search Console counts some features as positions and ignores others. Ads are not counted. Some elements, like a "People also ask" box, can take a position of their own.

So "position three" can be the first blue link on the page, sitting below ads, a map and an AI-generated summary. On a phone, that can be well below the first screen.

This is why the same position can give very different CTR on different searches. Before you judge a position or a CTR, look at the actual results page. Search Console tells you where you are in the count. Only the results page tells you where you are on the screen. That's what the SERP part of this series is for.

A reading order

To put it together, here's the order I read Search Console in. It keeps the averages from fooling me.

Start with totals, in whole weeks, without the last few days. Then split by search type, country and device, to see where a change actually happened. Then go to pages, and find the few pages that explain most of the change. Then, for each of those pages, look at its queries, with position and impressions side by side. And last, open the actual results page for the most important queries.

Every step narrows the question. By the time you reach the results page, you're not browsing anymore. You're checking one specific suspect.

Wrap-up

Four metrics, two averaging traps, whole weeks, and a look at the real results page. The Search Console reading checklist is linked in the description. Next episode: why query reports never add up, and when to trust page-level data instead. See you there.