Build a metric dictionary before your BI dashboard
By Mrunal Walke2 min read
Define each business metric's calculation, population, timing and owner so dashboard users can compare figures with confidence.
On this page
A business intelligence dashboard is hard to trust when the same label means different things to different teams. Before building charts, write a metric dictionary: a short, versioned definition of what each number includes, how it is calculated and who can explain it.
Begin with a decision
Choose a decision the dashboard should support, such as identifying orders that need intervention. Then define the metric required for that decision. Starting with every available field usually creates a busy dashboard without a clear operating purpose.
For each metric, record its name, business question, source, calculation, filters, time basis, refresh expectation and owner. Include explicit exclusions. Cancelled records and test transactions can change a total even when the formula itself looks correct.
Make the population visible
| Definition element | Example question |
|---|---|
| Grain | One row per order, line or event? |
| Population | Which statuses and entities count? |
| Time | Order date, completion date or posting date? |
| Calculation | Count, sum, ratio or another rule? |
| Freshness | When did the source last refresh? |
| Ownership | Who approves a definition change? |
Zoho Analytics distinguishes formulas and aggregate calculations. The practical implication is that a ratio calculated from grouped totals can differ from an average of row-level ratios. Decide which result answers the business question.
Use a worked ratio check
Imagine two hypothetical teams: one completes 9 of 10 tasks, and the other completes 10 of 100. Their completion rates are 90% and 10%. Averaging those percentages gives 50%, while the combined completion rate is 19 divided by 110, approximately 17.3%.
Neither calculation should be labelled simply 'completion rate' without explaining the intended weighting. For an overall task completion metric, the combined numerator and denominator describe the full task population.
Test the definition against records
Select a small set of source records and calculate the result manually. Include a cancellation, a late update and an empty period. Decide whether an empty denominator should display no data rather than zero.
Show freshness beside the metric when stale data would affect a decision. A current-looking chart can be misleading if the integration stopped yesterday. Keep failed refreshes visible to the owner.
Publish the dictionary with the dashboard and version material changes. When a definition changes, explain whether historical figures were recalculated. That small discipline makes future comparisons much more defensible than changing a formula silently.
Further reading
Primary reference. The workflow examples above are illustrative implementation guidance, not customer results.
Explore the related Dood resource. To discuss your workflow, contact Dood System.
