Hotjar

Definition

Hotjar is a behavior tool for session recordings, heatmaps, on-page surveys, and feedback. In an A/B test it answers why a variant moved the metric. It does not assign the variant or decide the winner.

What Hotjar means in A/B testing

Hotjar stores a session and can attach event tags to it. Filters then keep only the sessions that carry a tag. That is the join to an experiment: if the converting session is tagged with the test and the variant, you can replay that slice instead of a random sample of the whole site. Heatmaps and surveys can use the same tag. The tag is not a sample size, and a handful of replays is not a significance test.

Why Hotjar matters

A conversion lift does not say which part of the variant did the work. Replays of people who converted on the winner, or surveys shown only to the loser, are how a team picks the next hypothesis. They also catch a broken layout that made a variant lose for a reason that has nothing to do with the idea. Without a variant tag, those replays are mixed and the comparison is a guess.

Example of Hotjar

A checkout-summary test loses on completion rate. Mida tags converting sessions as mida_conversion_{test_id}_{variant} . On a Hotjar Plus plan, the filter for the losing variant's tag shows the sessions that did convert there. The first ten replays show the new summary hiding the shipping cost until the last step. The metric said the variant lost. The recordings show the mechanism. The test is not "won" or "lost" inside Hotjar.

How to use Hotjar

Install Hotjar and Mida, enable the integration, and convert once on each variant in a test session. Confirm the two event names appear on that Hotjar session. Use them to filter recordings after the experiment has a real sample, not instead of the sample. If you need the filter, confirm the workspace is on Plus or above before you promise the team a per-variant heatmap.

Common mistake

A common mistake is expecting a tag on every exposed visitor. The integration deliberately tags conversions only. Another is calling a variant broken because three recordings looked bad. Recordings pick the next test. They do not replace the conversion count.

How Mida works with Hotjar

The Hotjar integration calls hj('event', ...) when a visitor converts. It applies two tags: mida_conversion_{test_id} and mida_conversion_{test_id}_{variant}. Inside Hotjar those tags filter recordings and heatmaps to people who converted on a test or on one variant. Surveys can be targeted at the tagged session, for example a question shown only to the losing variant.

Both snippets have to be on the page. The event call works on every Hotjar plan. Filtering by the event needs Plus or above. Exposure is not sent, so the tag quota is not spent on every page view. Microsoft Clarity is a separate integration with its own tag, documented at Microsoft Clarity, not a mode of this Hotjar connection.

Related A/B testing terms

• Microsoft Clarity
• Heatmapping
• User Experience (UX)
• Conversion Rate

FAQ

Can I watch everyone who saw variant B, including people who did not convert?

Not with the standard integration. Mida sends the Hotjar event on conversion only, because Hotjar plans cap how many event tags a session can carry. Exposure-level tagging is not part of that integration. The integration page points at Mida's webhook if you need to call Hotjar yourself for exposure.

Why can I see the tag call but not a filtered heatmap?

Calling hj('event') works on every Hotjar plan. Filtering recordings and heatmaps by that event is a Plus or Business feature. Until that filter exists, the tag is stored and you still cannot isolate the variant inside Hotjar.

Is a Hotjar recording the experiment result?

No. The recording is how a converting visitor behaved. The result is still the pre-registered metric and the sample in Mida. Use the recordings after the quantitative read, to see what the winning or losing variant actually did.