Your click-through rate is falling because you are growing

Here are two numbers from the same site over the same eight months.

Click-through rate: down from 29.25% to 2.72%. A tenfold collapse.

Clicks: up from 31 to 16,410. A 529× increase.

Both are true. Neither is a mistake. And the first one caused a genuine near-miss: the kind of moment where a founder looks at a falling line, concludes the strategy is broken, and changes something that was working.

This is the most commonly misread pattern in Search Console, so it is worth understanding properly rather than being reassured about.

Why the average has to fall

Average CTR is not a measure of quality. It is a weighted average across every query you appear for, and the weights are impressions.

When you launch, you appear for almost nothing. Typically that is your brand name and a handful of very specific long-tail phrases. Both convert extremely well, someone searching your name and finding you clicks at a high rate, because you are exactly what they asked for.

Then you start ranking for more things. Broader queries, more competitive queries, queries where you sit at position 9 rather than position 2. Every one of those arrives with a lower CTR than your existing average, and every one of them drags the average down, while adding clicks.

Here is the arithmetic, simplified:

StageQueriesImpressionsClicksCTR
Launchbrand only1063129.25%
Month 4brand + long tail24,3852,0528.42%
Month 8broad coverage604,34816,4102.72%

The CTR line fell every single month. The business grew every single month. The falling line is a description of the growth, not a warning about it.

There is only one way to keep a 29% CTR: never rank for anything new. That is not a strategy, it is a ceiling.

The test that tells you which one you have

None of this means a falling CTR is always fine. It means CTR alone cannot tell you. You need to look at it against impressions and clicks together, and the four combinations mean four completely different things.

CTRImpressionsClicksWhat it means
↓↑↑Healthy expansion. You rank for more, at lower positions. Keep going.
↓↑flat/↓Wasted visibility. You are appearing but not earning the click. Fix snippets and intent match.
↓flat↓A real problem. Same exposure, fewer clicks. Check for SERP feature changes and competitors.
↑↓↓Contraction. You lost broad coverage; only the strong queries remain. Investigate indexing.

That last row is the one people misread as good news. A rising CTR alongside falling impressions is the classic signature of losing coverage. You are down to the queries you were always going to win. We watched exactly this happen when a site was deindexed from one engine and came back with a CTR of 10.68% where it had been 46.56%; the 46% was the unhealthy number, because it meant almost nothing but the brand name was indexed.

Split brand from non-brand before you conclude anything

Aggregate CTR mixes two populations that behave nothing alike, and until you separate them the number is close to meaningless.

In this dataset:

That is a 4.9× difference. Which means your headline CTR is largely a measure of what proportion of your traffic is brand, and a falling headline CTR often just means non-brand grew faster than brand, which is precisely what acquisition looks like.

If you track one CTR number, track non-brand CTR. It is the one that responds to your titles and descriptions. The blended number mostly responds to your mix.

Position is the other half

CTR without position is uninterpretable, because position explains most of the variance.

The trap is that average position hides its own distribution. An average of 8.42 sounds like you are nearly there, five places from the top. It usually means a scatter across many queries, most far from page one, with a handful of good ones pulling the mean up. You cannot fix an average; you can only fix the queries inside it.

The other trap is that published CTR-by-position tables are averages across categories that behave nothing alike. In this dataset the same site, the same pages, on the same day:

One curve cannot describe both. Derive your own from your own export. It is the only version that can price a ranking change for you, and it is usually very different from the table you were about to use.

Zero-click is real, and it is not the whole story

A share of the decline is genuinely external. More queries now resolve inside the results page (AI overviews, featured snippets, direct answers) so an impression is worth less than it was.

Two honest observations:

It hits informational queries hardest. If your page answers a question that can be answered in a sentence, expect the summary to take it. If your page performs a task, the click still has to happen, because a synthesised answer cannot merge someone's files for them.

It is not a reason to stop measuring. "Zero-click" has become the universal explanation for any CTR decline, which makes it useless as a diagnosis. Run the four-row table above first. If impressions and clicks are both rising, you do not need an industry-wide explanation for your falling average. You have an ordinary arithmetic one.

What to do instead of panicking

Stop watching blended CTR as a health metric. Watch clicks, and watch non-brand CTR separately.

Give it a quarter before you judge. Search results settle on a one-to-four month delay. In this dataset the single best growth month, up 61%, was a month with zero commits; it was being paid for work shipped in the previous two. Changing strategy during that lag resets the clock on a payout about to land.

Know how big your noise is. In the last sixty days of this dataset the day-to-day swing was 187% of the mean: a five-fold range with nothing changing. If you check on a Saturday and again on a Tuesday you can watch traffic "collapse" and "recover" without a single thing having happened.

Then fix the queries, not the average. Pull the queries with high impressions and low CTR at positions 5–15. That list is where snippet work actually pays, and it is short enough to act on this afternoon.

Questions

Is a falling click-through rate bad for SEO?

Not by itself. If impressions and clicks are both rising alongside it, the decline is arithmetic: you are ranking for more queries at lower average positions, and each new one dilutes the average while adding traffic. It only signals a problem when clicks are flat or falling too.

What is a good CTR in Google Search Console?

There is no useful universal figure, because it depends almost entirely on your position mix and how much of your traffic is brand. In this dataset brand queries converted at 16.02% and everything else at 3.29%, a 4.9× gap on the same site. Compare yourself against your own position→CTR curve rather than a published benchmark.

Why did my CTR drop after my traffic increased?

Because the new traffic arrived at lower positions than your existing traffic. A brand query at position 1 might convert at 30%; a competitive query at position 9 might convert at 1%. Adding the second lowers your average and raises your clicks at the same time.

How do I separate brand from non-brand traffic?

Filter queries containing your brand name and any common misspellings, then compare the two groups' clicks, impressions and CTR separately. Brand traffic is your most valuable and least scalable; non-brand is the one that responds to on-page work. Tracking them blended hides both.

Does a low CTR hurt my rankings directly?

The evidence for CTR as a direct ranking factor is weak and heavily contested. What is not contested is that a higher CTR at the same position means more traffic, which is a sufficient reason to work on titles and descriptions without needing the ranking argument at all.

How long should I wait before acting on a CTR change?

At least a month, and preferably a quarter. Day-to-day variance in this dataset was 187% of the mean, which is large enough to manufacture any story you want from a two-day comparison. Search results also settle on a one-to-four month delay, so a change you make today is being judged against work you did last quarter.

Run this on your own data