Building Dashboards Executives Actually Use
Every BI project starts the same way: a workshop, a wall of sticky notes, and a wishlist of forty metrics. Six months later there is a dashboard with forty tiles, and the CEO still asks the CFO to WhatsApp him the sales number. We have rebuilt enough abandoned dashboards to recognise the pattern — and it is almost never a technology problem.
Dashboards fail for three predictable reasons: they answer no specific question, their numbers cannot be trusted, or they arrive slower than a phone call. Fixing those three things matters more than any visualisation technique.
Start from decisions, not data
The wishlist approach — what would you like to see? — produces inventory. The decision approach produces tools. For each executive, ask instead: what decisions do you make weekly, and what number would change your mind? A sales director deciding where to focus this week needs pipeline coverage by region against target, not eleven views of historical revenue. A useful heuristic: if no plausible value of a metric would change anyone's behaviour, it does not belong on the dashboard.
- One screen, one question: each dashboard page should answer a single decision-shaping question a specific person actually has.
- Five to seven numbers on the summary view, each with target and trend — detail lives one click down, not on the front page.
- Show comparison, always: a number without a target, prior period, or threshold is trivia.
- Design for the reader's cadence: a daily operations view and a monthly board view are different products, even if they share data.
Trust is built in the pipeline, not the chart
The fastest way to kill a dashboard is one meeting where its number disagrees with finance's number. From that day, every figure is double-checked against a spreadsheet, and the dashboard is decoration. Trust problems are almost always definition problems in disguise: revenue with or without returns, headcount including contractors or not, order date versus invoice date.
The unglamorous fix is a definitions catalogue — every metric documented with its exact formula, source, and owner, signed off by finance before anything is visualised. Behind it sits an automated pipeline with data-quality checks that block refreshes when source data looks wrong, because an obviously stale dashboard is embarrassing, but a subtly wrong one is fatal.
Executives do not abandon dashboards because the charts are ugly. They abandon them the first time a number cannot be defended in a meeting.
— 7x Technologies BI practice
Make freshness visible
Every dashboard should state, on its face, when its data was last refreshed and what period it covers. Readers forgive latency they can see; they do not forgive discovering that yesterday's decisions used last week's numbers. If a metric genuinely cannot be fresher than daily, say so on the tile.
Adoption is a rollout, not a launch
- Put the dashboard inside an existing ritual: the Monday trading meeting runs from it, on screen, from week one — usage follows habit, not enthusiasm.
- Give every executive a fifteen-minute walkthrough of their own view; those who feel fluent return, those who feel lost quietly revert to asking for reports.
- Watch usage analytics and prune ruthlessly — a tile nobody has opened in a month is clutter earning distrust.
- Ship the mobile experience properly: executives check numbers between meetings, and a dashboard that needs a laptop loses half its moments of use.
A dashboard programme succeeds when the phrase check the dashboard replaces ask someone to pull the numbers. That takes a narrow, decision-focused scope, a pipeline whose numbers survive hostile questioning, and a rollout that embeds the tool into how the leadership team already works. None of it is glamorous. All of it is why some dashboards run companies while others just run on the office TV.