Resources
Building a dashboard that gets used
A dashboard that gets used answers a question asked by someone who has to decide. It therefore starts from that question, chooses its representations accordingly, and sets thresholds that trigger action.
One question, one decision, one indicator
Before choosing a chart, you write the question and the decision it drives: « are our delivery times holding, and do we need extra staff this week? » The indicator follows the decision, the representation follows the indicator. Change over time calls for a line, comparison between categories calls for bars, a share of a whole calls for a breakdown. That downward path produces short dashboards recipients actually open.
- The question written at the top, in the recipient's own words
- One primary indicator, completed by two contextual ones
- A representation chosen by the nature of the question
- A visible threshold separating normal from action required
- An update frequency that is stated and held
Choosing the right representation
A trend
A line, with an axis starting at zero. A truncated axis turns a small variation into a collapse.
A comparison
Bars sorted by value. Sorting does half the reading work, and it is a decision, never a default.
A share of a whole
A segmented breakdown, with counts displayed. The share is computed at render time, never by hand.
A point-in-time status
A single large figure, with its threshold and its date. Often the best visualisation available.
The business intelligence programme, block by block
From query to indicator, through decision modelling. The final block produces a complete dashboard for a real organisation, defended before a panel.
| Teaching block | Year | Hours | Credits |
|---|---|---|---|
| Databases and querying | 1 | 240 h | 24 |
| Data quality and preparation | 1 | 200 h | 20 |
| Analytical modelling | 1 | 160 h | 16 |
| Reporting and dashboards | 2 | 200 h | 20 |
| Steering and indicators | 2 | 160 h | 16 |
| Cross-curricular base | 1-2 | 120 h | 12 |
| Workshop and placement | 2 | 120 h | 12 |
Questions about dashboards
How many indicators should I show?
One primary and two contextual ones cover most weekly decisions. Additional indicators belong in a detail view, opened on demand by those investigating.
How do I know a dashboard is useful?
When a documented decision rests on it. The best check is to ask the recipient what they did differently this month: the answer says exactly which indicators to keep.
Can an indicator produce side effects?
An indicator shapes behaviour, which is its virtue as much as its risk: the business intelligence programme devotes a whole block to building an indicator, knowing its effects and balancing several complementary measures.
Explore next
- Misleading analysisSelection effects, chosen denominators, correlation versus causation, truncated axes: the four checks to run before accepting a figure.
- Machine learningRegression, classification, cross-validation, model selection and interpretability: what a model actually does and how it is validated.
- Models in productionInference service, drift monitoring, retraining, running cost and the trade-off between local and cloud processing.
- Data governanceData ownership, life cycle, hosting, retention, reversibility and vendor dependence: framing governance.
- Personal dataPurpose, legal basis, minimisation, retention, security and individual rights: the processing register in practice.
- Industrial site safetyHSE management system, work permits, co-activity, internal audit, leading indicators and safety culture: a site's complete arrangement.