How to Run a Programme Health Check

Before you read this guide, score your programme on these five statements. Answer honestly. One: more than 80% of our experiments have a falsifiable hypothesis with a 'because' clause. Two: of experiments completed in the last two quarters, more than 70% had a decision recorded. Three: a new team member could search our experiment history and find everything tested in a given product area in under five minutes. Four: we can name three patterns that have emerged across multiple experiments. Five: if our programme lead left tomorrow, less than 20% of the programme's knowledge would be lost. If you answered yes to all five, your programme is healthy. If you answered no to two or more, or if you could not answer because you do not have the data, that is itself a health signal.

What to Assess

1. Evidence integrity.

Checkpoint: Pull up your last ten completed experiments. How many had a decision protocol written before launch? How many reached full sample size? If you cannot answer without opening each one individually, you are not tracking experiment quality at the programme level.

2. Evidence synthesis.

Checkpoint: Ask a team member to find all experiments related to a specific product area from the last two years. Time them. If it takes more than five minutes, your experiment history is not reliably searchable.

3. Decision influence.

Checkpoint: Produce a list of decisions made because of experimentation in the last six months. Not results. Decisions.

4. Programme operations.

Checkpoint: When was your last backlog grooming session? Your last brainstorming session? Your last review board meeting? If any of these rituals have been skipped more than once in the last two quarters, the governance structure is eroding.

5. Organisational resilience.

Checkpoint: Name every aspect of the programme that only one person understands. Now ask: what would happen to each of those aspects if that person left with two weeks' notice?

Running the Assessment

Gather evidence (1 to 2 weeks). Score each dimension (Strong, Adequate, Weak, Critical). Identify root causes: infrastructure gap, process gap, knowledge gap, resource gap, priority gap.

Checkpoint: When you identify the root cause for a weak dimension, is it 'infrastructure gap' more than once? If the same root cause appears across multiple dimensions, that is the underlying problem.

Build an improvement plan. Three to five specific actions with owners and deadlines.

What Good Looks Like

A healthy programme scores strong or adequate across all five dimensions. If the checkpoints repeatedly pointed at infrastructure gaps, that is the finding. The question is whether the data needed for the assessment is already structured and accessible, or whether running the health check is itself a months-long research project.