Blog / 2026
Questions to ask when you're reviewing session recordings (instead of running a usability test)
A usability test script exists because “just watch the participant” isn’t a method — you need specific questions to keep the session focused and the findings comparable across participants. Reviewing session recordings needs the same discipline. Without a checklist, it’s easy to watch ten sessions, feel like you learned something, and be unable to say exactly what.
This is the review equivalent of a moderator’s script — a fixed set of questions to ask of any recording, or any batch of them, before you decide what it means.
Before you watch anything: what are you actually looking for?
Don’t open the session list and start browsing. Decide the question first — “why did signups drop after Tuesday’s release,” not “let’s see what people are doing.” A vague starting question produces a vague answer, the same failure mode a bad usability-test script has.
For a single session
Did the person get where they were trying to go? Not “did they complete a task you assigned” — you don’t get to assign one — but whether the session has a plausible goal (checkout, a specific page, a search) and whether they reached it.
Where did the pace change? A session that moves steadily and then suddenly slows, backtracks, or shows rapid repeated actions is telling you exactly where to look. Read the transcript rather than scrubbing the video — a timestamped list of events makes the pace change visible at a glance in a way real-time playback doesn’t.
Did anything fail silently? Check for failed network requests or JS errors around any point of hesitation. A slow or missing response with no visible error state is the single most common cause of what looks like “confusing UX” — the UI wasn’t confusing, it just didn’t say anything went wrong.
Would a different label or state have prevented this? For any dead click or rage click, ask what the control looked like right before the click. If it looked identical whether it was enabled or disabled, loading or idle, that’s your answer regardless of what caused the underlying delay.
For a batch of sessions (the step that actually matters)
Does this repeat, or is it one person? A single session with a confusing moment is an anecdote. The same control, same page, same behaviour across multiple distinct sessions in the same week is a pattern. Never act on one recording alone — see what is a rage click for why this specifically matters for rage-click data, where individual double-clickers are a real source of false positives.
Is it new, or has it always been there? Compare the same page or flow before and after a recent release. If the pattern predates the release, you’re looking at existing debt, not a regression — which changes how urgently it needs fixing.
Who is it happening to? New users only, or everyone? A pattern concentrated in first-session users usually points to an onboarding or expectation-setting problem; the same pattern spread evenly across tenure points to something structurally broken.
What did they do right after? Did the session end there, or did the person recover and keep going? A rage click followed by the person finding another path to the same outcome is a worse finding for prioritisation purposes than it looks — the person didn’t leave, they built a workaround, which usually means more people than the ones you can see are quietly doing the same thing.
When the answer to these is “I don’t know”
That’s the point where it’s worth running an actual moderated usability test, not before. If you’ve asked all of the above and still don’t know why people are getting stuck at a specific step — only that they are — that’s a real “why” question, and real session data won’t answer it. Five participants and a script will. See session replay vs. usability testing for where that line sits in general.
Running through this checklist manually across dozens of recordings is what makes session review feel like a chore. UserTapes turns every recording into a searchable transcript, so “did anyone rage click checkout this week, and did it start after Tuesday’s deploy” is a query, not an afternoon of video. Connected over MCP, an agent can walk that same checklist for you — see session replay over MCP.