MELLORCATalk to Mellorca
Patterns

Data, Reporting & Management Visibility

When Every Department Reports the Same Metric Differently

Sales, operations and finance each report a different figure under the same name, and the monthly meeting becomes a reconciliation exercise. Here is why that happens and how to settle it.

Mellorca Patterns2026-10-08

Observable condition

The monthly management pack is open on the boardroom screen and the first ten minutes go to an argument nobody planned. Sales says it closed 142 new customers last month. Operations says 118 were activated. Finance says 126 were invoiced. Customer support counts 131 accounts with a login. All four figures are called "new customers", all four come from a real system, and each of them is defensible to the person who produced it.

The same thing happens with other measures. "On-time delivery" is measured from order date by one team, from the promised date by another, and from the date the customer was told by a third. "Active client" means a login in the last ninety days for one department and a paid invoice in the last twelve months for another. Staff turnover, average resolution time, conversion rate and utilisation all show similar drift.

An illustrative example: a Cape Town managed-services company with about ninety staff reviews performance every month. Each head of department arrives with a slide that is internally consistent and externally contradictory. The managing director has learned to ask, before any discussion, "whose number is this?" The first third of the meeting is spent reconciling instead of deciding.

Realisation

The disagreement is rarely about arithmetic. Each team is answering a slightly different question with the same name. Sales counts a signature, operations counts a go-live, finance counts an invoice. None of them is wrong, but the business has never said which question the name "new customer" is meant to answer, so every department quietly answers the one that suits its own work.

That is why the problem survives so many good-faith attempts at cleaning up. Re-running the reports does not help, because the reports are correct against their own definitions. Asking people to be more careful does not help either, because the carelessness is not in the calculation. It is in the absence of a shared, written meaning.

The UK government's service manual, written for public services, makes the underlying point plainly: a performance measure needs to be defined clearly enough that it can be collected the same way each time and compared over time. A number that changes meaning depending on who reports it cannot show whether things are improving.

Realisation prompt: pick the single measure you quote most often to customers, lenders or staff. Can three people in different departments write down its definition and arrive at the same sentence?

Identification

This is a pattern of divergent metric definitions. The visible symptom is competing numbers in meetings. The common assumption is that one team has the wrong data, or that the business needs a better reporting tool. The structural cause is that each department has defined the measure locally, in the system and the vocabulary it works in, and nobody has been given the job of fixing one business meaning and applying it everywhere.

The published Pattern when nobody owns the definition of a business metric covers the ownership gap. This Pattern concentrates on what the divergence looks like day to day, and on the quiet ways departments drift apart even when a definition once existed. The Discovery note executives get different answers to the same KPI question describes the same condition from the executive side.

Diagnosis

  • Metrics are born inside tools. The CRM, the billing system and the service desk each ship with their own idea of a customer, a closed case or a completed job, and teams adopt the tool's default without noticing that it differs from the next tool's.
  • Different moments are measured. A sale has a quote date, an acceptance date, a start date and a first-invoice date. Each department measures from the event closest to its own work.
  • Inclusions and exclusions are unwritten. Are internal test accounts counted? Credit notes? Customers on a free trial? Public holidays in turnaround time? Every unrecorded choice is a place for numbers to split.
  • Time periods do not align. One team reports calendar months, another reports four-week cycles, and finance reports on the date a transaction was posted rather than the date it happened.
  • Local fixes become permanent. A manager adjusts a formula to make the report match what they know to be true on the ground. The adjustment is reasonable, undocumented, and never reaches the other departments.
  • Systems are not connected. Where data has to be re-keyed or exported between tools, small differences in timing and filtering creep in. IBM's overview of data silos describes how isolated stores of data lead to inconsistent views of the same business.

Commercial impact

The commercial value at stake is decision quality, management time, credibility with outside parties and the ability to improve what is measured.

Management time is the visible cost. Reconciling four versions of a number in every monthly meeting is hours of senior attention that produces no decision. The less visible cost is that people stop trusting the pack. Once a manager has been embarrassed by a figure that turned out to be calculated differently, they begin to keep their own private numbers, and the business gains a fifth version.

The external cost can be larger. A lender, an investor, a prospective buyer or a large corporate customer who asks for headcount, recurring revenue or service levels and receives two different answers draws a conclusion about management, whether or not it is fair. In tenders, a service-level figure that cannot be reproduced the same way twice is a liability if the customer audits it.

Targets and incentives suffer too. If a bonus depends on "customers acquired", the definition decides who gets paid, and a disagreement about the definition becomes a disagreement about people. The Pattern the difference between having data and having visibility explores why plenty of data does not translate into a shared view.

Common misidentification

The usual response is to appoint someone to "clean the data". That helps for one month and then the numbers drift again, because the cause is the definition, not the data entry. Another response is to buy a dashboard platform so that everyone looks at the same screen. A shared screen showing a contested number only makes the contest more public. Microsoft's documentation on scheduled refresh in Power BI is a useful reminder that even with one tool, a dashboard shows the data as of the last refresh and according to the model behind it. The model still has to encode a decision somebody made.

A third response is to declare one department's figure "the official one" by seniority. This ends the argument but not the problem: the other departments continue to run their work against their own numbers, and the official figure is quietly distrusted.

Possibility

In a better state, a short list of the measures the business actually steers by, perhaps eight to twelve, each has a written definition on one page: what is counted, from which event, from which system, over what period, what is excluded, and who owns the definition. Reports across departments draw on that definition rather than rebuilding it. When someone proposes a change, it is made once, dated, and communicated.

The monthly meeting then starts from agreed facts. If two numbers differ, the difference is explained by a named reason, such as "activated versus signed", instead of discovered live. Where it is useful to show both, the business labels them clearly and gives each its own name.

Intervention

  1. Collect the contested measures. Over one reporting cycle, note every number that was challenged or reconciled in a meeting. This is the real list of measures that need definitions.
  2. Put the versions side by side. For the three most contested, ask each department to write how it calculates the figure, including source system, start event, inclusions and period. Differences usually become obvious within an hour.
  3. Decide what question each measure answers. Choose the definition that serves the purpose, whether that is billing, capacity planning or customer experience. Where two purposes are real, keep two measures with two names.
  4. Write and own the definition. Keep it short, assign an owner, and store it where report builders will see it. The Insight how to decide which system owns which data helps to settle which source is authoritative.
  5. Rebuild the reports against it. Change the departmental reports so that they use the agreed definition, and record the date of the change so trends can be read correctly.
  6. Review quarterly. New products, branches and systems create new edge cases. A short review keeps the definitions alive.

Diagnostic questions

  • Which numbers were queried or reconciled in your last management meeting?
  • Is there a written definition for each measure in the monthly pack?
  • Do departments pull the same measure from the same system?
  • When a formula changes, who is told, and is the date recorded?
  • If a bank asked for your customer count tomorrow, whose figure would you give?

Bottom line

When every department reports the same metric differently, the business is not short of data. It is short of a decision about what the words mean. Settling a small number of definitions, owning them, and building reports from them turns the monthly pack from an argument into a tool, and gives outsiders one answer when they ask.

Sources and further reading

Related Mellorca reading

Article summarySales, operations and finance each report a different figure under the same name, and the monthly meeting becomes a reconciliation exercise. Here is why that happens and how to settle it.
Explore our solutionsStart a conversation