p95 Latency

p95 latency is the response time below which 95% of requests complete, a tail-latency metric that captures the slow experiences a simple average would hide.

Reporting a percentile rather than a mean exposes the worst common cases: a good average can still mask a poor p95 that frustrates a meaningful share of users.

For agents, p95 latency is a standard term in SLOs and SLAs because the slowest responses often shape perceived reliability.