How much diagnostic detail should a file tool collect by default?

When a file-processing step fails, support teams need enough context to reproduce the problem. But filenames, page contents, and even document metadata can be sensitive.

I would start with the operation, error category, app version, and a coarse size or page-count range only when needed. Content and filenames should stay out of default telemetry; a diagnostic bundle could be opt-in, previewable, and redacted before sharing.

What is the smallest set of details that has helped you diagnose real user issues without collecting more than the task requires?

5 views

Add a comment

Replies

Best

Close to your list, from a consumer app where the content is people's song lyrics. The app keeps a plain log file on the device, and the user can share it from Settings. What goes in: a plain statement of each step and every branch that gives up, plus every tap named by a stable control name and an id, never a title or anything the user typed. Lyrics, passwords and one time codes must never go in. A support ticket adds platform, app version, the screen it was sent from and a snapshot of the account's state. And the alert we get about a new ticket carries ids and codes only, no email and no message text, because that channel isn't named in our privacy policy. Your opt in, previewable bundle sounds like the right next step.