Masked session replays: what the recording cannot tell you

You watch someone fill in a voucher field, submit it and leave. The entered value is masked. Was the voucher expired, mistyped or excluded for that item? The replay cannot answer.

The recording may show that an error state appeared after submission. It does not establish what the person entered or why they stopped. Keep that gap in the finding.

Masking and blocking leave different evidence

Depending on the technology, session replays reconstruct recorded page states and interactions. They are not necessarily complete videos of everything the person saw. Inputs, entire regions or embedded content may be absent.

The rrweb documentation distinguishes masked text from blocked elements rendered as placeholders. A mask therefore preserves a different view of the interaction from a completely excluded region. Check which settings the actual integration applies.

Fictional case: The voucher field’s value is hidden, but the error “Not valid for this item” remains visible. You can identify the feedback, but not the particular voucher. If the entire area is blocked instead, even the displayed error cannot be assessed from that replay.

Describe the observation without inventing a cause

Start with what is visible: “An error state appears after submission; the recording ends without another visible attempt.” Then add what is missing: “Input masked; voucher validity and reason for leaving unknown.”

A pause proves neither confusion nor anger. The person could be checking something outside the captured view. The end of a recording need not be the deliberate end of the task either. Check whether collection stopped early or the relevant region was technically absent before describing abandonment.

Recreate the question with example data

You do not need unmasked customer inputs to investigate the voucher form. Prepare synthetic cases in a protected test environment: a valid code, an expired code and a code with an item exclusion. Inspect the feedback, focus after the error and the ability to try again. This checks the functional rules; it does not retrospectively reconstruct the unknown individual case.

If the open question is whether people understand the error, use a focused task with prepared information. A technical reconstruction of clicks does not answer that comprehension question. The article on clear form errors covers the next step.

Preserve the collection limits. Revealing real inputs solely to make analysis easier would create a different collection process. Before another recording, also use synthetic values to inspect dynamically displayed fields and error messages: they can repeat information outside the originally masked input. Masking alone is no assurance of anonymity or legal compliance.

The Analytics data check addresses the plausibility of agreed events. It is not a complete replay or legal audit. When enquiring, state which conclusion the available material currently cannot support.