Analytics

Compare an intervention with a meaningful reference

Editorial edition · 16 November 2026

The idea

A change in a number does not automatically show an improvement caused by an intervention. The reference period may differ in workload, season or measurement method. A baseline makes the comparison explicit enough to examine those differences.

Design choice

Choose a comparison that resembles the operating conditions of interest. Record any simultaneous changes. If a controlled comparison is not possible, describe the result as observational and keep plausible alternative explanations visible.

Worked example

A queue becomes shorter after a software update during a holiday. Compare arrival volume and processing capacity before attributing the whole reduction to the update.

Before you ship

  • Match relevant operating conditions
  • Record concurrent changes
  • Separate observation from causal evidence

Examples are illustrative. Adapt the checks to your own environment.

Further reading

OpenTelemetry metrics

Search the library