Skip to content
Module 09 of 1040 min readMixed

Dashboards that get used

Why most dashboards die: too many charts, no clear question, no maintainer. The five-question dashboard discipline.

90%

Listen along

Read “Dashboards that get used” aloud

Plays in your browser using on-device text-to-speech — nothing leaves the page.

Learning objectives

By the end of this module, you should be able to:

  • 01Use the five-question dashboard discipline to scope a dashboard before building
  • 02Design dashboards that get used weekly, not abandoned after one demo
  • 03Recognise the three reasons dashboards die

The graveyard of corporate BI tools is full of dashboards. Every analyst has seen them: built with care, demoed to applause, used twice, then forgotten while the data quietly rots. The reasons are always the same. The dashboard answered no specific question. It tried to show everything. No one owned it.

The five-question discipline

Before building anything, force the stakeholder to articulate the dashboard in five questions. If they can't, the dashboard isn't ready to be built.

  • Who is the audience? (Specific role, not 'the team'.)
  • What decisions will they make from it? (If 'just to know', it'll go unused.)
  • What are the 3-5 metrics that matter? (More than 5 = unfocused.)
  • How often will it be looked at? (Weekly = needs current data; quarterly = a memo would do.)
  • Who maintains it when the source schema changes? (No owner = death warrant.)

The three-chart rule

The best operational dashboards have three charts. A summary number with trend, a breakdown by the most important dimension, and an exception flagger ('what's anomalous this week'). More than three and the reader doesn't know where to look. A dashboard with twelve charts is a report disguised as a dashboard.

Layout discipline

Top-left is the most important real estate. The dashboard's headline metric goes there. Eyes travel top-to-bottom and left-to-right; place charts in order of decreasing importance along that path. End the dashboard with the most operationally useful detail — the table or list someone scans for action items.

text
┌─────────────────────┬─────────────────────┐
│ HEADLINE METRIC │ TREND OVER TIME │
│ (number + delta) │ (sparkline / line) │
├─────────────────────┴─────────────────────┤
│ BREAKDOWN BY KEY DIMENSION │
│ (bar chart, sorted) │
├────────────────────────────────────────────┤
│ EXCEPTIONS / ACTION ITEMS │
│ (table or list, top 10) │
└────────────────────────────────────────────┘
A working dashboard skeleton. Three sections. Each answers a different question. Total of 3-5 visualisations.

Owners matter more than tools

A dashboard without a named owner is dead the moment a source schema changes — and source schemas change. The owner doesn't have to be the original builder; it has to be someone who answers a 'why is the number wrong' question within 24 hours. Stakeholders who insist on dashboards but won't name an owner are asking for furniture; build them a memo instead.

Dashboard sprawl

Many organisations have hundreds of dashboards, of which 20 are actively used. The fix is not better dashboards; it's a quarterly audit that retires unused ones. Build the audit into the workflow when you set the dashboard up — log views, alert when usage drops to zero, retire after 90 days.

Exercise

Pick a dashboard you've built or use regularly. Walk through the five-question discipline. Does it pass? If you redesigned it as a three-chart dashboard, which three would you keep — and why are the others not on the list?

Key takeaways

  • Dashboards exist to answer specific questions. If they answer none, they die.
  • Most dashboards have too many charts; the best have three.
  • An owner who maintains the dashboard determines whether it survives.
Loading progress…
LeadAfrikPublic Economics Hub