Why use percentiles instead of an average for latency?

An average hides the tail. If most requests are fast but a few are very slow, the mean looks fine while real users hit the slow ones. p95 and p99 answer the question that matters - what is the experience for the slowest 5 or 1 percent of requests - which is why latency SLOs are written against percentiles, not averages. p50 (the median) tells you the typical case; the gap between p50 and p99 tells you how bad the tail is.

How is the percentile computed?

The values are sorted and the percentile is found by linear interpolation between the two nearest ranks - the same method used by NumPy and spreadsheet PERCENTILE functions. For large request volumes this is exact; monitoring systems often approximate p99 from histograms, which can differ slightly, but for a pasted sample this gives the true value.

No. All computation runs in your browser - nothing you paste is sent anywhere. You can analyze production latency numbers without them leaving your machine.

Free tool · Runs in your browser · Nothing uploaded

Latency percentiles.

Paste a list of response times (or any numbers) and get the p50, p90, p95, p99, min, max, mean, and standard deviation — the percentiles that actually describe latency, because the average hides the tail.

Values (one per line, or comma/space separated)

The average lies; the tail tells the truth

Report latency as an average and you will be blindsided: a mean of 50ms can hide a p99 of 600ms, and it is that slow 1% — often your most active users, making the most requests — who feel the pain and churn. Percentiles put numbers on the experience at the edges, which is why every serious latency SLO is written against p95 or p99. Watch the spread, too: a wide gap between p50 and p99 means an unpredictable tail worth chasing down.

Related free tools

Production-ready checkerScore your deploy Deploy error decoderDecode any deploy error CORS config generatorFix CORS the right way Cache-Control builderBuild a Cache-Control header

All free tools →

Free tools for the backend you run.

One of a set of free developer tools from Infraveil — the control plane for running, securing, and watching your backend on servers you own. Pairs with the SLA calculator and capacity calculator.

See all free tools

Get the backend toolkit

Practical tools and guides for shipping and running a backend you own. No spam.