Skip to content
EncryptTechBD
Home Industries Our Work Blog Pricing About Contact Discuss Your Project
বাংলা (BN)
Data & Analytics 6 min read

Business metrics definition explained: from measure to decision

What a metric definition actually contains, how it travels from source system to dashboard, and why the same number can be correct and useless.

E
EncryptTechBD · 0 comments
Laptop keyboard in front of a bar chart on screen
Photo: Negative Space / StockSnap

Two managers can look at the same revenue figure and disagree without either being wrong about the arithmetic. One is counting invoiced value, the other collected cash. One includes returns, the other does not. The disagreement is not about data quality; it is about a missing definition.

A metric definition is the written contract that makes a number repeatable. It says what is counted, what is excluded, where it comes from and who owns it. This explainer covers the components, the path a definition travels and the points where it usually breaks.

Measure, metric and KPI: three different jobs

A measure is a raw quantity: an order value, a login count, a support ticket duration. A metric is a calculation with a formula and a unit: average order value, weekly active users, median resolution time. A key performance indicator is a metric attached to a target and a decision: the resolution time the service team is accountable for this quarter.

The distinction matters in meetings. People often ask for a new dashboard when they need a target, and they ask for a target when they first need a definition. Naming the job correctly shortens the conversation.

The anatomy of a metric definition

A definition that survives an audit usually contains the following fields. Short definitions can omit some, but the omissions should be deliberate.

  • Name: the label used everywhere, without synonyms.
  • Question: the business question the metric answers in plain language.
  • Formula: the calculation, written so a reader can reproduce it.
  • Numerator and denominator: the exact populations included in each.
  • Grain: the level at which the metric is valid, such as order, customer, branch or day.
  • Filters and exclusions: test orders, internal accounts, cancelled records, duplicate entries and out-of-period adjustments.
  • Source: the system, table and field of record, including the timestamp used.
  • Owner: the person accountable for the definition and for approving changes.
  • Refresh: how often the value updates and how late-arriving data is handled.
  • Version: a date and change note so historical reports can be interpreted correctly.

How a definition travels through the stack

Most metric failures happen between systems, not inside a formula. Trace the path:

  1. Source system: a transaction is recorded with a status, a timestamp and an identifier.
  2. Ingestion: rows are extracted, sometimes filtered, and land in a staging area.
  3. Model: tables are joined, duplicates removed and business rules applied.
  4. Semantic layer: the metric is defined once and exposed to reporting tools.
  5. Visualisation: a chart applies filters, time zones and aggregation.
  6. Decision: a person interprets the number and acts.

Each step can change the population. A status filter in ingestion, a join that drops unmatched rows or a dashboard default that excludes the current day will all produce a number that looks correct. The defence is to store the definition once, test it at each boundary and reconcile a sample back to the source.

A worked example: on-time delivery rate

Begin with the question: of the orders we promised to deliver in a period, what share arrived by the promised date? A simple formula is:

On-time delivery rate = orders delivered on or before the promised date ÷ orders due for delivery in the period

The formula is easy; the decisions are not. Are cancelled orders in the denominator? What happens when a customer changes the delivery date after the order is placed? Is a partial shipment on time if one item arrives late? Which timestamp proves delivery: the courier scan, the customer confirmation or the warehouse return sheet? Each answer changes the number.

Write the answers down and test them against last month's orders. Then test the reverse case: find an order the dashboard calls late and verify it in the source. If the two checks agree, the definition is usable. If they do not, the gap is usually a filter, a timestamp or a duplicate record.

Five misconceptions that cause disputes

  1. "The dashboard is the definition." A dashboard is one implementation. The definition must exist outside it so it can be reused and audited.
  2. "One source of truth means one database." It means one agreed definition, even when the data comes from several systems.
  3. "Averages are neutral." An average can hide a long tail, mix different grains or weight a small branch the same as a large one.
  4. "More decimal places mean more accuracy." Precision is not validity. A precise number built on a wrong population is still wrong.
  5. "A finance metric works everywhere." Revenue, margin and cost often have different meanings in sales, operations and accounting.

When a definition matters, and when it does not

Exploratory analysis can tolerate loose definitions. If you are looking for a pattern in a spreadsheet, a rough measure is enough. A definition becomes mandatory when a number is reported repeatedly, compared across teams or periods, used to pay a commission, trigger an alert, satisfy a regulator or support a board decision.

The test is simple: would two reasonable people need to ask a follow-up question before acting? If yes, the definition is not finished.

Running a definition workshop

Bring the people who calculate the number today, not only the people who consume it. Ask each function to produce the value for the same period using its current method. Compare the results and write down every reason for the difference. Then agree the formula, the exclusions and the timestamp, and assign an owner.

Finish by testing three edge cases: a cancellation, a late correction and a record that exists in one system but not another. Publish the definition in a place people can find, and version it when the business rule changes. A definition that lives only in a meeting note will be reinvented within a quarter.

Where definitions sit in our work

We build reporting and data foundations inside our Business Automation division. The business operations automation service covers the data flow and the reporting layer. The pricing page explains how discovery, modelling and support are scoped. For the dashboard design that follows, read the KPI dashboard explainer.

Sources

Services that deliver this

Want this implemented properly?

Pricing →

Every engagement starts with a written scope. If a requirement is outside it, we quote it separately before starting.

Responses (0)

No comments yet. Add the first one below.

Leave a response

Your email address will not be published. Required fields are marked *.

Keep reading

More field notes