The Dashboard That Everyone Trusts
A retail group's finance team publishes a monthly gross margin dashboard to 40 regional managers. The dashboard pulls from a cloud data warehouse refreshed nightly. Last month a regional manager noticed that the 'Gross Margin %' tile showed 31.2% while the finance pack showed 29.8%. Investigation found the dashboard computes margin as (Revenue - Cost of Goods Sold) / Revenue using a COGS figure that excludes inbound freight, while the finance pack includes freight. The dashboard has been live for 14 months and no one raised the discrepancy before. The tile has a small 'i' tooltip that says 'Margin %' with no definition. Two regions have already made pricing decisions based on the 31.2% figure. The data team says fixing the definition will take three weeks because freight is captured in a separate system.
1. Identify the visualisation and communication failures in this dashboard and explain how each contributed to the wrong decisions.
Read the worked answer
Three distinct failures compound here.
- Undefined metric with a false sense of authority. A single-number tile labelled 'Margin %' presents a derived measure as if it were unambiguous. The tooltip gives no formula, no inclusions/exclusions, and no owner. Viewers reasonably assume the number is the margin. A definition should be visible on the face of the tile, not hidden behind an 'i'.
- No reconciliation or variance signal. The dashboard and the finance pack are two views of the same underlying reality with different definitions. There is no visible bridge, no note on the source, and no automated check that the two agree. A simple variance indicator ('differs from finance pack by 1.4pp due to freight treatment') would have surfaced this in month one.
- Precision without context. 31.2% implies two-decimal precision. Presenting a number to one decimal place signals it is measured, not estimated or defined; readers anchor on the digits rather than the definition.
The 14-month silence is itself evidence: when a number looks authoritative and no one is accountable for its definition, disagreement is attributed to the reader's misunderstanding, not the dashboard's. The two regions acted because the dashboard was the most convenient number, not because it was the right one.
2. The data team proposes a three-week fix. What interim controls would you put in place, and how would you communicate the correction to the 40 managers?
Read the worked answer
Interim controls (this week):
- Relabel the tile to state the definition explicitly: 'Margin % (excl. inbound freight)'. Change is cheap and removes the ambiguity immediately.
- Add a prominent banner linking to the finance pack definition and stating the known 1.4pp difference.
- Suppress the tile for the two affected regions until their pricing decisions are reviewed, or annotate those regions directly.
- Add a manual reconciliation step: finance signs off the dashboard's margin against the pack before publication, recorded in a simple log.
Communication: Send a short, factual note from the CFO, not the data team, covering: what was wrong, which number is authoritative, the size of the difference, which decisions are affected, and what managers should do next. Avoid blame language; the failure was systemic. Offer a 30-minute walkthrough for the two affected regions.
Permanent fix: ship the freight-inclusive definition with a documented metric dictionary, a visible formula, a named metric owner, and an automated reconciliation check that blocks publication if the dashboard and pack diverge by more than a set tolerance. Treat the metric dictionary as a controlled artefact with version history.