Compare application crash frequency against actual engagement time

Source facts

Endpoint analytics calculates application mean time to failure from engagement duration divided by crash count over 14 days. Engagement includes active interaction and keep-alive activity, such as presenting. The report limits collected crash events to ten per app, device, and day. Its app-performance population also excludes applications without meaningful foreground usage and requires a sufficient active-device population. Microsoft Learn.

Applicability

Use this comparison for desktop application reliability, with the reported application identity and period recorded. Do not treat the report as a complete inventory of every background or rarely used application.

DSE recommendation

Compare crash frequency alongside usage and the number of affected devices. Ask the application owner to identify business impact before prioritizing a widely used app over a small but critical one. Inspect app-version and device detail for the affected population, and keep conclusions about the cause separate from the observed rate. Avoid ranking reliability using raw crashes alone.

Verification

Choose representative reported failures and reconcile their device timelines with the user’s experience. Check whether the compared applications had adequate usage and whether a severe recurring failure may be affected by the event cap. Test any proposed update or repair with the application owner and record the actual workflow result. Preserve the original period and population so a lower count after change is not mistaken for improvement merely because usage fell.

Official references

Microsoft Learn: Application Reliability Report in Endpoint Analytics.