The meeting where nobody agrees on the number
A familiar scene: two managers bring different sales totals to the same meeting. One report was exported on Monday, the other on Wednesday. One includes returns and the other does not. Half an hour goes on working out why the figures differ, and the decision that was supposed to be made is postponed.
The problem here is rarely arithmetic. It is that the organization has several versions of the same fact and no agreed way to say which one counts. Pulling data into one managed place is a common response, and it can help, as long as expectations stay realistic.
Where conflicting versions come from
Versions multiply in ordinary ways:
- A report is exported to a spreadsheet, then edited locally.
- Someone adds a column to adapt the file to their team’s needs, and that copy is then emailed onwards.
- Two systems hold the same customer or product, and each is updated by a different team.
- Definitions differ quietly: “revenue” may mean invoiced, paid or recognized under accounting rules.
- Data is pulled at different times, so each version is correct for a different moment.
Spreadsheets are not the villain. They are flexible and familiar, and the limits of spreadsheet systems are a topic in their own right. The trouble starts when a spreadsheet becomes the unofficial record for something many people rely on.
What a single source of truth changes
A single source of truth means that for each important measure there is one place designated as authoritative, and everyone agrees to use it for shared decisions. Done well, it has practical effects:
- Fewer reconciliation arguments. When everyone pulls from the same place, differences point to a genuine question rather than a copy that fell out of date.
- Shared definitions. The effort of writing down what a measure means is often more valuable than the platform itself. Careful metric definitions make it clear what is included and excluded.
- A visible audit trail. It is easier to see where a number came from and when it was last refreshed.
- Less duplicated effort. Analysts stop rebuilding the same extract in separate files.
The effect is on the quality of the conversation. Teams can still disagree about what to do, but they are less likely to disagree about what happened.
An illustration
Imagine, purely as an illustration, a company where the sales team’s report shows 120 orders for the month and the finance team’s report shows 114. The 6-order difference could come from orders placed but later cancelled, or orders entered on the last day of the month that finance books in the next period. With no agreed rule, each team defends its own figure. With a documented rule, for example “an order counts once it has shipped”, both reports can be rebuilt from the same record, and the difference either disappears or becomes a short, explainable reconciling item. The numbers here are invented to show the mechanism, not drawn from any real business.
The caveat: garbage in, garbage out
Putting data in one place does not make it correct. A central store inherits whatever errors exist in its sources, and because many people now rely on it, mistakes can spread further and look more authoritative. Several risks are worth naming:
- Bad inputs. If source systems contain duplicates or missing fields, the central version will too. Checking for common data quality issues at the point of entry is usually cheaper than correcting them later.
- False confidence. A tidy dashboard can discourage people from questioning a figure that deserves a second look.
- Stale data. A single source of truth that updates monthly is not much help to a team that needs yesterday’s position.
- Hidden judgment calls. Choosing which source wins when two disagree is a decision, and it should be written down and owned rather than buried in a script.
For these reasons the central dataset should come with a named owner, a stated refresh schedule, and a clear way for users to report suspected errors.
Making it work in practice
A few habits tend to matter more than the choice of tool:
- Start with the handful of numbers that cause the most disagreement and settle those first.
- Write each definition in plain language, with an example of what is in and out.
- Make the official source easier to use than the copy. If people export to spreadsheets because the central tool is slow or hard to filter, versions will return.
- Keep a record of changes to definitions, so that a shift in a trend can be traced to a rule change rather than a real change in the business.
- Allow local analysis, but ask people to label it as working analysis rather than the official figure.
When one source is not the right goal
For a very small team, the overhead of a formal central dataset may exceed the benefit, and a shared file with a clear owner may be enough. In a large organization with distinct business units, a single global definition may not fit every unit, and some measures may legitimately differ by region or product line. In those cases the better aim may be a small number of agreed enterprise measures plus clearly labelled local ones.
Centralization also changes who has to maintain things. If the central team becomes a bottleneck, people will go back to private copies. A good test is whether the people who make decisions find the shared numbers faster and more trustworthy than their own, and if not, whether that is a tooling issue, a quality issue or a process issue.
This article is general information and not advice on any specific system or compliance obligation.