Face Blurring for User-Research Session Highlights

For user-research highlights, show the evidence the team needs without carrying unnecessary participant information into every presentation. A face-blurred webcam tile can still sit beside a real email address, account balance, or personal story.

Start by checking the participant agreement and the intended audience, then edit the smallest useful excerpt.

Choose a clip that supports one finding

A highlight should let the viewer understand the observation without a long setup. Keep enough before and after the moment to avoid misrepresenting what the participant did.

Write a neutral caption describing the task or issue. Avoid turning one person's experience into an unsupported statement about every user.

If the useful evidence is entirely on screen, the participant's face may not be needed.

Inspect the shared screen

Look for names, email addresses, account identifiers, private documents, notifications, bookmarks, and browser tabs.

Crop to the task area where possible. If too much real data appears, recreate the interaction using a test account and clearly label it as a demonstration rather than the original observation.

Unseen does not currently redact arbitrary screen text. Use the research team's approved editor or sanitization workflow for those details.

Treat the webcam separately

For an authorized clip, use Select faces in Unseen and set the participant Hide. Check small tiles carefully; a face obvious to a human viewer may be difficult for automatic detection.

Inspect layout changes when screen sharing stops or the speaker becomes full screen. The same participant may appear at different scales and in multiple face entries.

Use the processing mode permitted by the team's policy. Local avoids a source upload for face processing, but approval still depends on the research arrangement.

Review the spoken content

A distinctive job, personal event, or organization can identify the participant. Decide whether the actual voice and full quotation are needed.

A transcript excerpt may be sufficient, but it needs the same context review. Removing the picture does not remove identifying information from the words.

Do not alter the clip or quote in a way that changes the finding.

Build the presentation from the reviewed copy

Create screenshots, thumbnails, and slide embeds from the approved derivative. Keep source recordings in the restricted research repository.

Attach the participant code and source interval only where internal reviewers need them. Avoid exposing a research tracking identifier in a public filename.

If the highlight moves from a product-team meeting to a public talk, review that new use separately. The research-presentation guide covers that audience change.