Article 7 of 10 · Data & Decision Intelligence · 10 min read
One Number, Three Answers: The Single Source of Truth Problem
Ask three departments for last quarter's revenue and you get three different numbers — each defensible, none identical. The problem is rarely the data. It's that nobody agreed what the number means.
A management meeting, first week of the quarter. The CEO asks a simple question: "What did we do in revenue last quarter?" Finance says €4.12 million. Sales says €4.38 million. The operations report on the shared drive says €3.97 million. Three competent people, three systems, three numbers — for the single most basic figure the company has.
The next twenty minutes are not spent deciding anything. They're spent reconciling. Whose number is right? Finance recognised revenue on invoice date; Sales booked it on the day the deal closed; Operations counted shipped units and multiplied by list price, ignoring the discounts. Everyone is technically correct. And the decision that was supposed to happen in that meeting — do we hire two more people in the northern region — gets pushed to next week, when someone will "clean up the numbers."
This is the single source of truth problem, and it is one of the quietest, most expensive dysfunctions in mid-market companies. It doesn't show up as a crisis. It shows up as friction — meetings that debate figures instead of decisions, reports nobody fully trusts, and a slow erosion of confidence in the company's own data.
It is not a tooling problem
The instinct is to buy something. A better BI platform, a data warehouse, a new dashboard. But a company can own the best tools on the market and still get three answers to one question. The tools faithfully report whatever they're told. If Finance, Sales, and Operations feed them three different definitions, they will produce three different, perfectly-rendered truths.
The root cause is upstream of any software. It lives in five places.
Notice that only one of these — transformation — is even remotely technical. The other four are decisions the organisation never made on purpose. That's the good news: problems you created by omission, you can fix by decision.
Where the number diverges
- Definition — "Revenue" isn't one thing. Booked, invoiced, recognised, collected — four different numbers, all called revenue.
- Source — The CRM, the accounting system, and a spreadsheet each hold a version. None was named the official one.
- Timing — One report refreshed this morning, another is a week old. Same metric, different snapshot in time.
- Transformation — Somewhere a manual export, a filter, a pasted column quietly changes the result before anyone sees it.
- Ownership — No single person is accountable for the number. So everyone maintains their own, and defends it.
What a "single source of truth" actually is
The phrase gets misused. A single source of truth is not one database that holds all your data. Most companies will always have data in several systems, and that's fine. It is also not a mandate that everyone use the same dashboard.
A single source of truth is a single agreed definition for each metric that matters, plus a single governed place where that definition is calculated. Revenue is defined once — say, recognised revenue net of discounts and returns, on invoice date — and every report that shows "revenue" pulls from the one calculation that implements that definition. Not three implementations that happen to agree on a good day. One.
This layer has a name in modern data architecture — a semantic layer, or a governed metrics layer. But you don't need the vocabulary to build the discipline. You need agreement, one calculation per metric, and the rule that no report is allowed to redefine a metric on its own.
The goal isn't one database. It's one definition, calculated in one place, that every report is required to use.
The real cost of not having one
The obvious cost is wasted time — the twenty minutes of reconciliation in every meeting. The larger cost is decision latency. When a number is contested, the decision that depends on it waits. In a fast quarter, "let's confirm the figures and revisit" can mean a two-week delay on a hire, a price change, or a supplier decision that had a real window.
The deepest cost is trust. Once leaders learn that the reported number might be wrong, they stop trusting reports and start trusting instinct — or the loudest person in the room. At that point you have paid for a data platform and reverted to running the company on opinion. Every dashboard in the building becomes decoration.
How to get to one number
This is not a two-year data-warehouse programme. For a mid-market company it's a focused exercise, usually a few weeks, and it follows a clear sequence.
1. Inventory the metrics that actually drive decisions
Not all 200 fields in your systems — the fifteen or twenty numbers that show up in real decisions. Revenue, margin, cash position, pipeline, churn, headcount cost, on-time delivery. If a metric never changes a decision, it doesn't need governing.
2. Write one definition per metric — and get it signed off
Plain language. "Revenue = recognised revenue, net of discounts and returns, dated on invoice." The point of writing it down is to force the argument now, in a document, instead of live in every meeting for the next three years. When Finance and Sales disagree on what revenue means, that disagreement is the whole problem. Resolve it once.
3. Name the governed source for each metric
For each number, one system is the official source. Others may hold copies, but the official value comes from one place. Where systems must be combined, the combination logic lives in one governed calculation — not in three analysts' spreadsheets.
4. Automate the calculation; remove the manual steps
Every manual export, filter, and pasted column is a place where two reports quietly diverge. The fewer human hands between the source and the report, the fewer ways the number can drift. This is where good BI engineering earns its cost.
5. Audit, and keep the definitions living
Definitions change — a new product line, a new revenue rule, a new market. The single source of truth survives only if there's an owner for each metric and a habit of updating the definition on purpose, not by accident in someone's spreadsheet.
The quiet advantage
Companies that fix this don't get a flashy dashboard out of it. They get something better: meetings that spend their time on decisions instead of on reconciliation, reports people actually act on, and a leadership team that trusts its own numbers enough to move quickly. In a market where everyone has roughly the same tools, the company that agrees on what its numbers mean makes faster, more confident decisions than the one still arguing about them.
See what your data is actually telling you.
Upload any dataset into PrismIQ and an AI agent walks you through the analysis, surfaces the patterns and anomalies that matter, and produces a clear report in business language — a fast way to see whether your numbers agree before you invest in a full BI platform.
The 7 Metrics That Actually Predict Business Growth · Data Quality: The Silent Tax on Every Decision