Five to seven numbers. At least one leading indicator per major function. Each metric has one owner, one threshold, and one expected action when it breaches. Reviewed in 10 minutes at the top of the weekly. Anything more is a dashboard, not a scorecard.
Why most scorecards become dashboards nobody reads
Most companies build scorecards the wrong way: they ask "what data do we have" instead of "what would we need to see to change a decision". The result is a 30-metric dashboard that is exhaustive, accurate, and operationally useless. Nobody reads it because reading it does not change anything.
A scorecard is built backward from decisions. If a metric does not have a clear decision attached to a clear threshold, it does not belong on the scorecard.
Four principles before you pick a metric
- One owner per number. If two people own a metric, neither will move it.
- One decision per number. Each metric must answer "what do we do if this breaches".
- One source per number. If the metric is calculated three different ways in three different systems, you have not picked a metric, you have picked a debate.
- One review cadence per number. Weekly numbers go on the weekly scorecard. Quarterly numbers belong in the QBR.
The structure: five to seven numbers, paired
A healthy leadership scorecard is five to seven numbers, deliberately paired so the team can see cause and effect:
- One revenue number (lagging).
- One pipeline or demand-gen number (leading).
- One delivery or fulfillment quality number (leading or lagging depending on cycle).
- One customer retention or NPS-style number (lagging).
- One financial health number, typically cash or gross margin (lagging).
- One people number, typically open roles or attrition (leading on capacity).
- Optionally, one strategic bet number tied to the quarter's commitments.
For more on the leading/lagging logic, see leading vs lagging indicators.
How to select the actual metrics
The selection exercise is two hours with the leadership team, structured this way:
- 30 min - brainstorm. List every metric currently tracked anywhere in the business. Expect 40 to 80.
- 30 min - prune. For each, ask "if this number changes, does a leadership decision change". If no, cut it.
- 30 min - pair. For each lagging metric still standing, identify one leading metric that should predict it.
- 30 min - assign. Pick five to seven. Name an owner, a source, a threshold, and an expected action for each.
Setting thresholds the team trusts
Every metric on the scorecard has a green-yellow-red threshold. The threshold is set by the owner, agreed by the team, and reviewed quarterly. Three rules:
- Yellow is meaningful. If yellow is just "a bit off green", it gets ignored. Yellow must mean "we are still on plan but trending wrong".
- Red triggers an action, not a discussion. Every red metric has a pre-agreed first move. The weekly is for confirming the move, not inventing it.
- Thresholds are not aspirations. Green is what good looks like in this quarter, not what you wish were true.
Operating the scorecard in the weekly
The scorecard lives at the top of the weekly leadership meeting. Ten minutes, in this order: each owner reads their number, declares the color, and either confirms the pre-agreed action or surfaces it as an issue for the issues block. See the weekly leadership meeting agenda for how it slots in.
The scorecard is also the first page of the QBR pre-read. The 13-week trend by metric tells the story of the quarter without anyone narrating it.
Evolving the scorecard each quarter
The scorecard is a living artifact, not a fixed one. Each QBR, the team reviews the scorecard itself with three questions:
- Which metric did not move a single decision this quarter? Replace it.
- Which decision was made repeatedly without a metric backing it? Add one.
- Which threshold proved wrong? Re-set it.
Expect the scorecard to be 70 percent stable quarter over quarter, with one or two metrics rotating in and out. A scorecard that never changes is probably no longer predictive. A scorecard that completely changes every quarter is probably reactive.
Five common scorecard mistakes
- All lagging. Revenue, gross margin, retention, NPS. Reads the past, predicts nothing.
- Too many metrics. Fifteen numbers becomes zero numbers. Attention is finite.
- Vanity numbers. Total users, total downloads, total pageviews. Number goes up; nothing decides.
- Composite scores. "Customer health index" hides the input. Show the inputs, not the average.
- No owner. A scorecard metric without one named owner becomes a shared inability to move.
See how your scorecard maps to the six pillars.
The ASCEND assessment scores the Data pillar on metric clarity, leading-indicator coverage, and decision throughput from the numbers you already track.