A single source of truth in reporting: why it is so hard to create

Almost every company I work with has the same conversation. In a meeting someone says: "According to my data, revenue for the quarter is X." Someone else says: "No, I have a different number." Then comes the investigation - who calculated it, where they pulled it from, which period, with or without tax.

Usually it turns out both are right - each in their own frame of reference. But nobody can say confidently: which number is correct.

This is called the single source of truth problem. And it is not technical in nature.

Why numbers diverge

The divergence is almost never caused by someone making a formula error or exporting the wrong table. Divergence is the result of several systemic things that accumulate over years:

• Different systems record the same event at different times. A sale appears in CRM at the moment of the deal. In accounting - at the moment of payment. In the warehouse system - at the moment of dispatch. All three numbers are different. All three are "correct".
• Different teams use different definitions. "Active customer" in marketing and "active customer" in operations are not the same thing.
• Historical data was edited manually without documentation. Someone once corrected a number in Excel, told nobody, and now two sources have diverged permanently.
• There is no metric owner. Nobody to ask: "is this number correct?"

What does not work as a solution

The typical reaction is to build one "correct" dashboard where all data flows together. Sometimes that helps. More often it does not. The dashboard shows a number but does not explain where it came from or why it was calculated that way. A year later nobody remembers the methodology. Two years later a competing dashboard appears.

Another typical reaction is to buy a BI system or a data platform. That also does not solve the problem by itself. A tool automates whatever already exists. If there is no methodology, the tool automates chaos.

What actually works

A source of truth is created not by a tool, but by agreement and discipline.

Concretely that means:

• For each key metric there must be a documented definition: exactly what is counted, for what period, what is included, what is excluded.
• Each metric must have a single owner - a person or role responsible for its correctness.
• The data source must be recorded: which system, which table, what transformation logic.
• Methodology changes must be documented with a date and a reason.

Technically this can live in a simple spreadsheet, a wiki, a git repository - the form is not critical. What matters is that it exists and is maintained.

A simple readiness test

Ask yourself and your team:

1. Can we right now show where the revenue number in the monthly report comes from - step by step, down to the original record in the system?
2. If the person who assembles this report quit today, who could reproduce it without losing anything and without making calls?
3. When was the last time the definition of a key metric changed? Was it documented?

If any of these questions has no answer - there is no single source of truth, regardless of how many systems the company has.

Creating a source of truth is not an IT project. It is a management decision about who is responsible for what.