Guides 13 Sep 2026 5 min read by Kalum

Rage clicks, dead clicks and quick-backs: reading Microsoft Clarity properly

Clarity is free, unlimited, and shows you the three things Analytics never will — where people get angry, where they click nothing, and where they give up and go back to Google. Here is what each signal actually means and what to change.

Analytics tells you a page got 4,000 visits and people stayed 48 seconds. That is a number, not information. It does not tell you that 6% of those people jabbed at a screenshot four times because it looked like a button, or that a fifth of them were back on Google inside eight seconds.

Microsoft Clarity tells you both, for free, with no sampling and no session cap. It is the single most useful free tool most site owners have never installed.

What Clarity is, in one paragraph

You add a small script to your site. Clarity records what real visitors do — every click, every scroll, every moment of hesitation — and gives you heatmaps, session recordings and a set of frustration metrics. It is free with no session limit, which is unusual enough to be suspicious; the honest answer is that Microsoft gets aggregate behavioural data out of it. On WordPress there is an official plugin, and if you use our tool it can install and connect it for you.

The recordings are compelling and will eat your afternoon. The metrics are what actually change decisions, so start there.

Rage clicks

What it measures: someone clicking the same spot repeatedly in quick succession.

What it means: they expected something to happen and nothing did. Almost always one of four things:

• An image that looks clickable — a screenshot of a button, a product photo, a chart people want to enlarge.
• A real link or button that is broken, or covered by something invisible like a sticky header or a cookie banner.
• Something slow enough that the first click looked like it failed.
• A form field that will not take focus, usually on mobile.

What to do: do not guess — watch one recording. Clarity lets you filter recordings to sessions with rage clicks, so this takes about ninety seconds. You will see the exact element. Then either make it do what people expect (link the screenshot, open the image), or make it stop looking clickable (drop the border, drop the button styling).

Where to worry: anything above about 3% of sessions on a page. Below 1% is background noise.

Dead clicks

What it measures: a click that produces no reaction at all — no navigation, no change on the page.

What it means: the gentler cousin of rage clicks. People are exploring and finding edges. Usually images, headings styled to look like accordions, and table rows people expect to sort.

What to do: dead clicks on images are the commonest, and the fix is usually to make the image open full size. Dead clicks on a heading mean people expect it to expand — either make it expand or change how it looks.

Where to worry: around 8% of sessions and up. Dead clicks are less urgent than rage clicks but they are free information about what people wanted.

Quick-backs — the one that touches rankings

What it measures: someone arriving, leaving within a few seconds, and — crucially — going back where they came from.

Why it is different from the others: rage clicks and dead clicks cost you conversions. Quick-backs cost you conversions and tell Google something. When a searcher clicks your result, returns to the results page within seconds and clicks a competitor, that is about the clearest signal available that your page did not answer the query.

Google does not publish how it uses this, and anyone who tells you the exact weighting is guessing. But a page with a high quick-back rate and a slipping position is a pattern you will see repeatedly once you start looking.

What it means: nearly always a mismatch between the promise and the page. The title said one thing, the page opened with something else. Or the answer is there but it is under 600 words of introduction.

What to do:

1. Look up what the page actually ranks for in Search Console — not what you meant it to rank for.
2. Read your first screen as if you had just searched that. Is the answer there?
3. Put the direct answer in the first paragraph. Two or three sentences. The background can follow.
4. If the title over-promises, change the title rather than the page.

Where to worry: 15% and up on a page with meaningful search traffic. This is the signal worth fixing first.

Scroll depth and excessive scrolling

Average scroll depth below about 40% on a long page means most readers never reach the second half. Either the intro is too long, or the thing they came for is buried. Move it up.

Excessive scrolling — people scrolling up and down repeatedly — means they know the answer is on the page and cannot find it. That is a structure problem, not a content problem. Add a table of contents, make the H2s say what is under them rather than being clever, and put the most-searched section nearer the top.

JavaScript errors

Clarity flags sessions where the browser threw an error. This matters for two reasons: broken scripts can stop content rendering for some visitors entirely, and they drag on Core Web Vitals, which Google does use directly.

The usual culprits are an abandoned plugin, an ad script, and a third-party embed whose host has changed something. Anything above roughly 5% of sessions is worth an hour with the browser console open.

A twenty-minute routine

Once a fortnight:

1. Open the dashboard and sort pages by quick-back rate, looking only at pages with real search traffic. Fix the worst one.
2. Sort by rage clicks. Watch one recording of the worst page. Fix what you see.
3. Check the errors count. If it moved, find out what you installed.
4. Write down what you changed and the date, so next fortnight you can tell whether it worked.

That last step is the one everyone skips and it is the one that makes the whole thing worth doing. Clarity updates daily, so a fix you make on Monday shows up in the numbers by the weekend — which is faster feedback than almost anything else in SEO, where you normally wait a month to find out whether you were right.

Read next

A/B testing titles on a WordPress site without upsetting Google

Split-testing headlines is standard practice everywhere except SEO, where everyone is quietly afraid of cloaking. Here is what is actually risky, what is not, and how to run a test on a small site that gives you an answer rather than noise.

Keyword research without Ahrefs or Semrush

Paid keyword tools sell you two things: a volume number and a difficulty score. Both are estimates. Here is how to get better answers for free, using sources that are closer to the truth than either.

A free SEO audit checklist you can run in one afternoon

Paid audits hand you 600 issues sorted by a severity score somebody else invented. This one has 24 checks, in the order that actually affects traffic, and every tool it uses is free.