Latency Through the Day

A single average can hide a changing workload. This fictional dataset compares two invented response-time series across eight points in a day. It is not a measurement of Stage.

Read the shape before the number

Both series rise around the middle of the day and decline later. The second series stays lower, but a chart alone cannot explain why: the difference could come from workload, implementation, or the choice of observations.

Use Load chart to activate the view, inspect both series, and download the small CSV. Every data point has an explicit hour and unit.

Calculate one difference

At noon, the invented baseline is 115 ms and the candidate is 72 ms. The absolute difference is:

The relative reduction is approximately MathematicsRendering this equation locally; the complete source follows.43 / 115 = 0.374. These are descriptive calculations on invented values, not a performance promise.

Ask better questions

• Are the samples comparable?
• Are outliers visible?
• Which units are used?
• Does a lower value change the reader's actual experience?

For categorical comparison, open Choosing a Cache Strategy. For the conceptual request path behind a measurement, return to Inside a Request Pipeline.

Inspection notes

The text, table, and source download remain useful without activating the chart. The mathematics should retain a readable fallback. Change the theme and browser width to compare line visibility and table layout.