Haomai Technology AI Brain

What does “sales” mean? Defining a metric teams can share

Two teams in one company can use “sales” for orders, invoices or payments received. Conversational analytics becomes useful when people agree on what the figure represents.

Data semanticsUpdated November 20264 min read

One name can cover different calculations

In a synthetic business, confirmed orders total 200 this month and cash collected totals 150. The sales team discusses deals signed while finance tracks money received. Both may call their figure “sales.” The ambiguity lies in the concept used in the question, even when both amounts are calculated correctly.

If the system picks the most similarly named field, the answer may look precise and still miss the decision a user is trying to make. Defining business semantics brings these implicit conditions into the open.

Start with a small definition

Choose one familiar metric before trying to model the whole organization. List its business object, reporting time, calculation, exclusions and comparison dimensions. For confirmed order value, use confirmation date, exclude canceled orders, group by region and maintain a consistent unit.

Review the relationships next. How do orders connect to products, customers and regions? Could a join turn one order into several rows? Without a clear grain before aggregation, even a simple sum can count an amount more than once.

Test the definition on boundary cases

Build a small synthetic sample with an ordinary order, a cancellation, confirmation across a month boundary, a duplicate import and a missing amount. Have business and technical teams calculate the metric independently and explain the treatment of each record. Agreement on these cases makes the definition usable.

Reuse the sample for future changes. When field mappings, sources or rules change, test these examples before moving to the full dataset. This helps distinguish a change in meaning from an execution error early.

Version your business definitions

A reorganization or a change to revenue recognition or refund handling can alter a metric. Specify when the new definition applies and whether history will be recalculated. Otherwise, a trend change caused by a rule update can look like a change in operations.

Versioning can start with a clear definition record, reason for change, effective period and approval process. This gives teams a shared reference when they disagree and tells users which interpretation a particular answer applied.

Make clarification part of the conversation

People can keep asking questions in familiar terms. When several definitions are plausible, explain the differences and let them choose orders, invoices or collections before continuing. Both sides then know which metric the analysis uses.

Starting with one metric and adding the relevant objects and relationships turns company knowledge into rules people can check. Use the metric definition card in the documentation to establish a baseline for a figure your team discusses often.

Back to all articles

Try “permissions,” “data source,” “manufacturing” or “API.”