How I Use Clarity Session Recordings to Find Conversion Problems

← Back to My Microsoft Clarity Overview

Disclosure: This is not a paid promotion. I have no affiliate or commercial relationship with Microsoft or Microsoft Clarity. It is a tool I genuinely use across my own businesses and for clients. Views are my own.

What session recordings actually show — and what they do not

Session recordings are the closest thing to watching a real visitor use your site. They are also easy to misread if you do not understand what you are looking at.

A Clarity session recording is a reconstruction of a visitor’s session — not a video of their screen. Clarity records the DOM state of your page and the visitor’s interactions (mouse movements, clicks, scrolls, keystrokes in non-sensitive fields) and replays them as an animation. This means recordings look like video but are actually a replay of data.

What you can see in a recording: where the visitor’s mouse moved and paused, where they clicked, how far they scrolled, which pages they visited and in what order, how long they spent on each page, and where they left the site. For mobile visitors, you see taps and swipes rather than mouse movement.

What you cannot see: what the visitor was thinking, what they looked at specifically (eye tracking is not captured), what they did in other browser tabs, or anything from pages outside your site. A visitor who spent 30 seconds on your contact page and then left may have been interrupted, satisfied, or frustrated — the recording alone cannot tell you which.

The recording review workflow

Treating recordings as an investigation rather than entertainment makes the process faster and the conclusions more useful.

The workflow that works across all the business sites follows the same sequence every time. Start from heatmap data — not from recordings. Heatmaps show you aggregate patterns across hundreds of sessions; recordings show you individual behaviour. It is much easier to find meaningful recordings when you already know which page has a problem and roughly where on that page the problem is occurring.

The workflow keeps the review process from becoming open-ended. Watching recordings without a structure tends to produce a long list of things you noticed without a clear priority order. Watching with a question produces an answer — or a reason why more investigation is needed.

How to use filters to find the most useful recordings

Filters are what separate a useful recording review session from an hour of watching people scroll. Use them before you watch a single recording.

In Clarity’s recordings section, the filter panel lets you narrow sessions by a wide range of criteria. The most useful filters for conversion investigations are below. Apply them in combination — for example, filtering for sessions on a specific page that also included a rage click will quickly surface recordings most likely to show friction.

The five friction signals that matter most

Most of what matters in a recording review comes down to five behaviours. Learn to spot them quickly and you will rarely need to watch more than a few minutes of any session.

When you are watching a recording, you are looking for moments that tell you something went wrong — or something the visitor expected did not happen. These five signals cover the vast majority of meaningful friction you will find in recordings on a typical business website.

Friction signal

Rage clicks

Multiple rapid clicks on the same spot. Flagged automatically by Clarity with a visual indicator in the recording. Almost always means the visitor expected something to happen and it did not. Check what element they were clicking and why it might have failed to respond.

Friction signal

U-turn behaviour

Visitor navigates to a page, spends a few seconds, and immediately goes back to the previous page or leaves entirely. Often indicates the page did not match what the visitor expected based on where they came from — a navigation label or ad that promised something the page did not deliver.

Friction signal

Form abandonment

Visitor reaches the contact form, begins interacting with it — moving the cursor to fields, sometimes starting to type — and then stops and leaves. Clarity masks form field content by default, so you see the interaction without the data. Common causes: too many fields, a required field that seems unreasonable, or a question that creates hesitation.

Friction signal

Excessive scrolling

Visitor scrolls up and down the page repeatedly rather than in one continuous direction. This usually means they are searching for something — a price, a phone number, a specific piece of information — and cannot find it easily. The fix is often making that information more prominent or adding it to a section the visitor was returning to.

Intent signal

Deep engagement before exit

Visitor spends significant time on the page, scrolls deeply, reads carefully — and then leaves without converting. This is not a friction signal in the traditional sense. It suggests the content is compelling but something at the conversion point itself is failing: the form is not visible, the CTA label is unclear, or the ask feels premature given the content.

Contextual signal

Repeated visits to the same page

Visitor returns to the same page or section multiple times within a session. This can indicate genuine interest and consideration — or confusion about where to find something. Context matters: is it a pricing page they are returning to (likely interest), or a navigation element they keep clicking (likely confusion)?

How many recordings to review before drawing conclusions

One recording is an anecdote. Ten recordings that all show the same behaviour is a pattern worth acting on.

The number of recordings you need to review depends on what you find in the first few. If you watch five recordings filtered for rage clicks on a specific page and all five show the rage click happening on the same element, that is already a strong enough signal to act on. You do not need to watch fifty more to be certain.

If you watch five recordings and each one shows something different — different exit points, different behaviour patterns, no consistent signal — then either the problem is not concentrated enough to identify from recordings alone, or the filter you applied was too broad. In that case, tighten the filter (add an additional criterion) or return to the heatmap data to narrow the investigation.

Decision points: act, investigate further, or leave alone

Not every friction signal in a recording needs fixing. Here is how to judge what is worth acting on.

The same principle that applies to heatmaps applies here: the goal is to find concentrated, repeatable friction — not to fix every imperfect visitor journey. Some visitors will always navigate in unexpected ways. What you are looking for is behaviour that appears consistently across multiple sessions and has a plausible connection to a lower conversion rate.

Real example: finding the ad landing page problem

Here is how session recordings — filtered by paid traffic — identified a specific problem that Google Ads data alone would never have surfaced.

Session recordings and heatmaps are two sides of the same investigation. Heatmaps show you the aggregate pattern; recordings show you the individual moment. Used together — as described in Guide 02 — they give you a complete picture of where your site is losing visitors and why.

For the full context of how session recordings fit into a website sales strategy, see my Microsoft Clarity overview. If you are running Google Ads and want to take the traffic-source filtering further, Guide 04 covers the full Google Ads integration setup.

Disclosure: This is not a paid promotion. I have no affiliate or commercial relationship with Microsoft or Microsoft Clarity. It is a tool I genuinely use across my own businesses and for clients. Views are my own.