Write a definition sheet for each dashboard metric

01

Define the question and population

Write the decision the metric supports, followed by the unit being counted: an order, customer, invoice or completed job. Name the population and reporting period. For a percentage, define both numerator and denominator, including how duplicates, cancellations and incomplete records are treated.

For example, an on-time completion measure might cover distinct orders due in a calendar month. Its denominator includes orders still unfinished at month end. Decide whether to use the original or revised promised date, who may change it and how the change is recorded. There is no useful comparison until the rule is consistent.

02

Specify the calculation and source

In a hypothetical month, 45 orders are due. Three are canceled before their due dates under the documented exclusion rule, leaving 42 eligible. Of those, 36 are completed by their agreed due dates. The rate is 36 divided by 42, or about 85.7%. Six eligible orders are late or unfinished. Show 36 and 42 beside the percentage. A single changed order moves this small sample noticeably.

List the system and fields used for each order ID. Include the due date, completion date and cancellation status. State the time zone, extraction date and refresh schedule. If a date is missing, put the record in an exception count and resolve it. Do not silently treat it as on time or remove it from the eligible population.

03

Assign ownership and check the output

Name a metric owner who can explain its definition and a source owner who can correct data. Before publishing, compare a small sample of dashboard records with the underlying orders. Reconcile totals where feasible and record missing data, failed refreshes and manual adjustments.

For each review, show the data date and mark a stale value. If two systems disagree, trace the difference to their fields and cutoffs before choosing a figure. A dashboard can display an unresolved discrepancy with a note. It should not imply precision that the source records cannot support.

04

Manage changes deliberately

Give the definition a version and effective date. When the rule changes, retain the old definition and label historical results. Recalculate earlier periods only if the necessary records exist and the new rule can be applied consistently. Otherwise show a break in the series.

Review whether the metric still supports its decision and pair it with a relevant countermeasure. For on-time completion, rework or customer complaints may reveal a quality trade-off. Differences in workload mix or staffing can affect both results, so use the sheet to frame questions rather than assign a cause automatically.

05

Show the effect of missing dates

Suppose 42 orders are eligible and 36 have documented on-time completions. Three are documented late, while three have missing completion dates. Do not silently exclude the unknown three or count them as on time. Until the records are resolved, the on-time rate could range from 36 ÷ 42, or 85.7%, to 39 ÷ 42, or 92.9%. Report the three unknown records beside the result and investigate them.