Experiment

A test designed to create learning and move a KPI. An “optimal experiment” is tied to OKRs, measurable, learning-focused, and executable within a short cycle (e.g. a week).

Last updated: December 12, 2025

Definition

A test designed to create learning and move a KPI. An “optimal experiment” is tied to OKRs, measurable, learning-focused, and executable within a short cycle (e.g. a week).

More context

An experiment is a time-boxed test that answers a specific question about how to improve a KPI. It should have a clear hypothesis, measurable success criteria, and a documented learning outcome.

Why it matters

Experiments create compounding knowledge: even failed tests improve future prioritization and expose execution bottlenecks.

How to use it

Write the hypothesis, pick one measurable KPI, define the time window, assign an owner, execute, then document the learning and next step.

Common pitfalls

Running “experiments” that are actually projects with no measurement, or running too many variables at once to learn anything.

Related terms

  • HypothesisA specific, testable statement about what change could improve a KPI. In the process, you typically form 1–3 hypotheses per cycle.
  • KPI (Key Performance Indicator)A measurable number used to guide decisions and evaluate progress. The first step in the process is choosing a KPI that can be measured periodically.
  • OwnerThe accountable person for a metric, area, task, or experiment. Clear ownership prevents work from stalling.
  • Learning experienceA mindset where outcomes (including failures or execution issues) are treated as learning that improves future decisions and process quality.

Related reading