The direct answer
Show the evidence behind the decisions the team recurs to: whether to continue or stop a campaign, where to put the next unit of budget, which content to refresh, whether a pipeline problem is collection or quality. Each decision gets one panel with four things on it: the current figure, its source and period, the comparable baseline, and the threshold that would change the decision. A panel missing any of the four is decoration.
This produces a shorter dashboard than the data-source approach, and a more demanding one. Every panel has an owner, because a decision has an owner. When a decision stops recurring, its panel is removed, which keeps the dashboard at the size of the team's actual attention.
Build it decision-first
List the recurring decisions
Write down the calls the team makes weekly or monthly, in the words used in the meeting where they are made.
Name the evidence each one needs
For each decision, the figure that would change it, and the source that figure can honestly come from.
Set the cadence and the threshold
How often the panel is reviewed, and what value, or what change, triggers action rather than discussion.
Remove panels nobody decides from
Keep the underlying data accessible, but off the dashboard. Attention is the scarcest input the dashboard consumes.
Anatomy of a panel that earns its place
- The decision it informs, stated on the panel, not implied by its title
- The owner of that decision, who is accountable for reading it
- Source and reporting period printed on the panel itself
- A comparable baseline: the previous period, the target, or both
- The threshold or change that would trigger a different decision
- A path from the panel to the underlying records, for when the number is challenged
Panels that usually do not earn their place
Lifetime totals that only grow: they cannot change a decision, because they cannot go down. Real-time panels feeding monthly decisions: the cadence mismatch guarantees they are either ignored or over-read. Averages that hide the segment where the decision actually lives, such as a blended conversion rate across audiences with nothing in common. And any metric whose definition nobody in the room can state, which will be argued about instead of used.
The test is simple to run. For each panel, ask: what decision did this change, or confirm, in the last quarter? Silence is the answer that removes the panel.
Growth Engine: dashboards around decisionsThe decision list comes first
A dashboard that starts with the tool ends up showing what the tool can show. A dashboard that starts with the decisions ends up showing what the team needs to decide. The method is to list the decisions first — the weekly operating decisions and the monthly strategy decisions — and then to give each decision exactly one panel. A panel that does not serve a decision is decoration, and decoration is the first thing to remove.
Each metric needs a definition, a source, and a refresh cadence, written down where the team can see them. A metric is not a number on a screen. It is a definition plus a source plus a cadence, and when two sources disagree, the definition is the arbiter. The most common dashboard failure is not missing data. It is two panels that use the same word to mean two different things, so the team argues about the number instead of acting on it.
The cadence follows the decision. Operating decisions are reviewed weekly, because a week is long enough for a change to show and short enough to correct. Strategy decisions are reviewed monthly, because they need a full cycle of evidence. A dashboard that is reviewed on no schedule is a dashboard that is not used, and an unused dashboard is a cost with no return.
Growth Engine exists to design dashboards around decisions. In every case the decision list comes first and the tool follows it, never the reverse. The dashboard is a record of decisions and their evidence, not a display of data.
The panel earns its place by being used. If a panel has not informed a decision in a month, it is removed, and the space is given to a decision that needs one. A dashboard that only grows is a dashboard that is never reviewed. The discipline is subtraction: fewer panels, each tied to a decision, each reviewed on its schedule.
Common questions
How many panels should there be?
As many as there are recurring decisions with owners, which is usually fewer than teams expect. A dashboard of eight owned panels outperforms one of forty unowned ones.
Should executives and operators use the same dashboard?
The same evidence, different views. Decisions differ by level, so panels differ; the records underneath must not, or the levels end up arguing from different facts.
Which tool should we build it in?
The one the team will keep honest. Growth Engine exists to design dashboards around decisions; in every case the decision list comes first and the tool follows it, never the reverse.