Checklists
Three checklists for the programme team: one for reviewing a single metric, one for launching a dashboard, and one for auditing a metrics programme. Use them as prompts for a conversation with the teams, not as a form to be filled in and filed. Tailor them, and drop items that do not fit your context.
Metric review checklist
Use this before a metric is adopted and at each periodic review (see topic 1.4, Metrics governance and ownership).
Purpose
- The metric answers a question that a named audience asked.
- The decision it should inform is written down.
- It measures an outcome, or it is clearly paired with one (see topic 1.3, Outcomes over output: choosing what to measure).
Definition
- A written definition, formula, unit, and data source exist.
- The clock or boundary rules are explicit (when counting starts, stops, and pauses).
- An owner is named, and a person to ask when the number looks wrong.
- Known limits of the data are recorded, including gaps and changes in how it was entered.
Honesty
- The most likely gaming vector is written down.
- A paired guardrail metric or practice is in place.
- The metric describes a team, service, or system, and never ranks individuals.
- Time measures show a median and a high percentile, with the sample size (see topic 1.6, Statistical literacy for service metrics).
- A baseline of several periods exists, so changes can be read against ordinary variation.
Use
- The team that does the work sees the number first and helped to define it.
- The last review led to a recorded decision, including a decision to wait.
- A condition for retiring the metric is written down.
Dashboard launch checklist
Use this before a dashboard is shown to its audience (see topic 11.1, Designing a service metrics dashboard).
Audience and content
- The audience and its questions are written down, and the audience helped to review a sketch.
- The view has no more than about nine measures, with detail one step down.
- The set is balanced across flow, quality, reliability, experience, and cost.
- Every measure has a companion that guards against its likely distortion.
Clarity
- Every value shows trend and context, with events noted.
- Every measure has a definition, source, owner, refresh time, and known gaps within one click.
- Targets are shown as references, and any status colour follows a stated rule that allows for variation.
- Decoration that adds no information has been removed.
Trust and operation
- The charter has been agreed, and individual-level views are off (see topic 11.3, Rolling out metrics without breeding fear).
- Teams have seen their own data and had time to correct it.
- Data is pulled automatically, with freshness checks and reconciliation with the source (see topic 11.2, Tooling landscape).
- A named owner, a review date, and a challenge route exist.
- The reports this dashboard replaces have a retirement date.
Programme audit checklist
Use this twice a year, with the teams, alongside the self-assessment (see topic 12.5, Maturity self-assessment).
Purpose and trust
- The programme has a sponsor and a written charter that leaders follow.
- Reported breaches of the charter were handled and the outcomes published without names.
- Team surveys and near-miss reports do not suggest rising fear.
- No league tables of teams or individuals exist anywhere in the organization's tools or slides.
Measures and data
- Each team can name the decisions its measures informed last quarter.
- Definitions are kept in a portable catalogue, with owners.
- Source coverage, pipeline reliability, and reconciliation are checked and visible.
- An export of definitions and history has been tested in the last year.
Balance and breadth
- Outcome measures sit beside activity measures in each view.
- AI-related measures follow baselines, sampling, and verified outcomes, not volume alone (see topic 10.4, Metric inflation and quality dilution).
- Cost, risk, and people measures are present where they matter, and none are used to rank individuals.
Improvement
- Improvement actions from the last review are completed or dropped with a reason.
- Measures and reports were retired in the last year.
- The effort asked of teams for measurement has been estimated and is acceptable to them.
- The programme itself has been reviewed with the teams, and the plan was changed as a result (see topic 11.5, An incremental adoption roadmap).