Insights / Article / Data foundations and advanced analytics
ArticleDashboards leaders actually use
A dashboard earns its place when it answers a question a leader asks every week. Start from decisions, show few measures and make every number trustworthy.
Most organisations have more dashboards than anyone can name. A few are opened every Monday morning. The rest were built for a project, admired at launch and quietly abandoned. The difference is rarely the software. It is whether the dashboard answers a question a leader genuinely asks every week.
This article sets out how to build the first kind: start with decisions, keep the measures few, and make the numbers trustworthy enough that nobody needs to check them in a spreadsheet before the meeting.
Why good-looking dashboards go unused
Dashboards usually start from the data that happens to be available. A team connects the finance system, the CRM and the warehouse feed, then arranges whatever comes out into tiles. The result is comprehensive and visually tidy, but it answers no particular question, so nobody returns to it.
The second failure is trust. If the sales director’s number differs from the one in the finance pack, the meeting turns into a debate about whose figure is right. After two or three of those, leaders go back to the spreadsheet they control, and the dashboard becomes decoration.
The third is overload. A page with forty measures tells a leader that everything matters equally, which means nothing stands out. People skim, see mostly green and move on, even when the one red tile is the one that matters.
Start with the decisions, not the data
Before any design work, sit with the people who will use the dashboard and list the decisions they make on a weekly or monthly rhythm. Be specific. “Understand sales performance” is not a decision. “Decide whether to move field sales effort from region A to region B” is.
For each decision, ask three questions. What would make you act? What would you need to see to be confident? How often do you make this call? The answers tell you which measures belong on the page, the level of detail required and how fresh the data must be.
- Interview four to six leaders for 30 minutes each and capture the recurring decisions they make.
- Group the decisions by rhythm: daily, weekly, monthly or quarterly.
- For each decision, note the measure that would trigger action and the threshold that matters.
- Discard measures that no decision depends on, however interesting they are.
- Sketch the page on paper and test it against last month’s real decisions before building anything.
A typical mid-sized distributor running this exercise might find that its operations director makes four recurring decisions: staffing levels for the coming week, which overdue orders to escalate, whether to authorise overtime and when to raise a supplier performance issue. That is a short, clear brief for a dashboard.
Five clear measures beat fifty
Once the decisions are clear, the measures usually fall into a small set. Resist the urge to add more “for context”. Context belongs one click away, in a drill-down or a supporting page, not on the front screen where it competes for attention.
Each measure on the front page should pass a simple test: if this number moved sharply, would someone do something different this week? If the honest answer is no, it does not belong there.
A practical limit
Aim for five to seven measures on the main view. If a leader needs more than that, the dashboard is probably serving two audiences or two decision rhythms, and should be split.
Show each measure with the comparison that makes it meaningful: against target, against the same period last year or against a rolling average. A bare number invites the question “is that good?” and the dashboard should answer it before anyone asks.
Make every number trustworthy
Trust is built by showing your working. Every measure should carry its definition, available on hover or in a short glossary, written in plain language. “Gross margin: invoiced revenue less landed cost of goods, excluding freight recharges” settles arguments before they start.
Show the time of last update prominently. If the warehouse feed failed overnight and the figures are a day old, leaders need to know that before they act. A visible timestamp turns a silent failure into an obvious one.
- Publish one definition per measure and use the same logic in the dashboard, the board pack and the source system.
- Display the data refresh time on every page, not buried in a settings menu.
- Flag incomplete periods clearly, such as a month that is only half finished.
- Reconcile key totals against the finance ledger at least monthly and record the result.
Give every measure an owner
A dashboard without owners becomes a list of questions nobody answers. Assign each measure to a named person who understands how it is produced and can explain a movement within a day. That person is also responsible for raising data problems and proposing changes to the definition.
Ownership changes the conversation in the weekly meeting. Instead of “why is this number down?” followed by silence, the owner arrives with a short explanation and a proposed action. The dashboard becomes the agenda rather than a backdrop.
We cut our weekly pack from thirty pages to one screen. The meeting got shorter and the decisions got faster, because we stopped arguing about the numbers.
Finance director, wholesale client
Keep it alive after launch
Decisions change as the business changes, and the dashboard should follow. Review it every quarter with its main users. Remove measures nobody has looked at, and add new ones only when they are tied to a decision. Most reporting tools record how often each page is viewed; use that evidence rather than opinion.
Watch for the warning signs of decline: leaders bringing their own spreadsheets to meetings, requests for “just one more tile” every month, or a growing list of known data issues that never get fixed. Each one is a prompt to go back to the decisions and start again.
Where to start this week
Pick one leadership meeting that runs on a regular rhythm. List the three or four decisions it makes, choose one measure for each, write its definition and agree who owns it. Build that single page, use it for a month and refine it. One dashboard that is genuinely used will do more for your data culture than a library of impressive ones that are not.
Source: Intelliversal analysis.
