Plan a test that can answer one question

A small test limits scope so the team can observe the work and learn before considering a larger change.

Write down the question, the setting and period, who coordinates and observes, what evidence will be recorded, and when the team will review it. For any regulated, safety-critical, or sensitive work, use approved local material and involve the appropriate trained roles before proceeding. This page is a coordination aid, not clinical instruction or approval.

Planning sequence

Complete these coordination decisions before the test begins.

  1. 1

    State the learning question

    Example: “When the facilitator uses a named-owner field in the meeting note, is that field present in the sampled entries afterward?”

  2. 2

    Set the boundary

    Name one workflow and a short observation window. The fictional example uses one weekly service-coordination meeting for four weeks.

  3. 3

    Assign roles

    Name a cycle coordinator, a workflow owner, and a person responsible for checking the measure definition. Responsibilities are roles, not assumed approvals.

  4. 4

    Agree what will be observed

    Use the written definition on the Measures page. Record the source, period, denominator, and any missing observations.

  5. 5

    Check local requirements and set a review point

    Before any sensitive or regulated work, consult approved local procedures and the relevant trained roles. Set a date for the team to review observations and decide what to do next.

Bounded test or broad rollout?

The fictional examples concern note coordination, not care delivery.

Bounded test

Try the named-owner field in notes after one weekly coordination meeting for four weeks, then examine the sampled entries and team observations.

  • One workflow
  • A defined observation window
  • A scheduled learning review

Overbroad proposal

Introduce a new note format across all services immediately, without a local test, agreed measure, or review point.

  • Multiple contexts at once
  • Unclear evidence for learning
  • No explicit decision checkpoint