Single Source of Truth: Why Business Numbers Disagree

A single source of truth means each fact the business relies on lives in one named place, with one person who answers for it. Each report then reads that fact from the same place. When business numbers disagree, the cause can be two systems answering slightly different questions. Neither figure has to be wrong for that to happen. The aim is one version of the truth that sales, finance and operations all accept. For background, see how data ends up in separate systems. This explainer sits under the Signisys use case on integrating business data across systems.

Single Source of Truth in Plain Words

In Signisys’ working description, it works fact by fact. For revenue, customers and headcount alike, decide which system owns the fact and who answers for it. Each report then reads from that place.

Two vendor pages help separate the terms. IBM’s system of record explainer describes a system of record as an authoritative source of data for a business domain or business process. A source of truth, in IBM’s wording, aggregates and harmonises data from several systems of record into a view across domains. IBM calls these two roles that a system plays, not two kinds of system. It adds that the word “single” emphasises an approach in which one source of truth is the one that each stakeholder and app references. However, IBM also notes that some organisations might have multiple sources of truth.

Workday’s guide ties the “truth” half to standardisation and agreed definitions. In its example, the departments must agree on what constitutes a new customer. Both pages come from software vendors, so read them as vendor wording.

Three terms come up from here on:

• System of record: the place where a fact is first entered and kept.
• Source of truth: the place where people bring facts from several systems together and read them for decisions.
• Master data: the shared core records, such as customers and products, that several systems use.

Why Business Numbers Disagree: Four Gaps a Single Source of Truth Closes

Each gap below comes with a question that finds it. One disagreement can involve more than one gap. Run all four questions before deciding where the fault lies. Closing these gaps is what a single source of truth is for. If business numbers disagree and no gap fits, the figures may hold plain errors, which is a separate problem.

Single Source of Truth Gap 1: The Same Word Means Two Things

Words like customer, order and revenue can carry a different meaning in each system or team. For example, sales may count a customer from the first signed order. Finance, by contrast, may count one only once payment arrives. Both figures are correct by their own definition. Yet the two totals then differ.

Workday’s guide makes the same point with its new-customer example, where the departments must agree on what constitutes one. Until the departments agree, a report may show one team’s meaning and label it as the company’s.

The question that finds this gap: ask both owners to write the definition in one sentence. Do the two sentences match? If they do not, the disagreement is about words, not data. In that case, the fix is an agreed definition, written down where both teams can read it.

Single Source of Truth Gap 2: Reports Read at Different Times

Two reports can describe the same month and still be read on different days. Someone takes one extract on the 1st and the other a week later, after late orders and returns arrive. In addition, systems update at different speeds. IBM notes that some systems of record capture data in real time, while others use batch updates. A batch here means a set of changes loaded together at a set hour. A report read before the batch runs would lack those changes.

The question that finds this gap: what date and time does each report show it was read? If the two differ, compare them at the same moment before going further. Then note the read time on each report. The next meeting then starts from the same point.

Single Source of Truth Gap 3: Numbers Joined by Hand

In Signisys’ working description, this gap opens when one person builds the report by hand. That person copies figures from several exports into a spreadsheet, adjusts them and sends the result. The spreadsheet then becomes the report, and the steps live in one person’s head. A second person repeating the work may reach a different total. This article makes no claim about how widespread the practice is.

The question that finds this gap: who builds the report, and could someone else rebuild it from the same steps? If the answer is no, write the steps down before trusting any figure. For removing the manual assembly itself, see the management reporting use case.

Single Source of Truth Gap 4: Copies Edited After Export

Someone spots an error in an exported file and corrects it there. Nobody makes the same fix in the system the figures came from, though. The next export can bring the error back, while the corrected copy drifts away from the original. Copies that people update separately need a rule for which one wins when they differ. Without that rule, one version of the truth splits into several copies.

The question that finds this gap: where does a correction happen, and does it flow back to where the fact was first recorded? If corrections stay in copies, agree that each fix goes into the original system.

Wrong, missing or duplicate values are a different problem. That belongs to data quality work. For help on that side, see the use case on improving data quality.

What a Single Source of Truth Is Not

It is not a dashboard, a warehouse or one system that does everything. In Signisys’ working description, a dashboard is a screen where people read numbers. Who owns each of those numbers is a decision that people make elsewhere.

A data warehouse is a store that collects data from several systems for reporting. Workday’s guide describes the warehouse or lake as the storage, and the source of truth as the governed layer on top. IBM, by contrast, lists data warehouses and data marts, master data management platforms and data lakehouses among the forms a source of truth can take. Taken together, the two pages suggest a place and a decision. Storing data is the place, while agreeing which version counts is the decision.

Nor is it one system for everything. IBM says each system of record is scoped to a single domain, and it notes that some organisations might have multiple sources of truth. A single source of truth, then, means one agreed answer per fact. One version of the truth for revenue can sit next to a different owner for headcount.

This article names and ranks no product.

Single Source of Truth: Who Owns Each Fact

Use the grid below in a meeting with the owners of the two reports. Fill in one row per fact. IBM’s explainer says that when questions or discrepancies arise about a piece of information, the system of record’s version is treated as the correct version. Agreeing that rule fact by fact is how a single source of truth becomes one version of the truth.

How to Tell Which Gap You Have

Run this in one meeting with the two report owners and both reports on the table. Each step points to one of the gaps above.

Record each answer in the grid above, against the fact in question.

Joining the systems up, with figures flowing without hand work, is a separate subject.

What This Article Does Not Cover, and When to Get Help

This explainer covers why business numbers disagree and who should own each fact. It does not cover how to connect the systems, how to score an existing set-up or how to choose an approach. Wrong or duplicate values belong to data quality work, and the effort of building reports belongs to the management reporting use case linked above. In addition, this article gives no share-of-cause statistics. The IBM page was published on 12 November 2025 and the Workday page carries no date; both were read on 2 November 2026.

Bring in specialist help when one of these is true:

• The people who own the facts have no access to the systems that record them.
• More than two departments dispute the same definition.

Signisys provides data integration and analytics services through its business intelligence and analytics solution and its data and AI services. Read this paragraph as a description of a Signisys service, not neutral advice.

Frequently Asked Questions