A meeting can spend its first twenty minutes trying to reconcile two numbers that share a name. The experience creates an understandable appetite for one definition, one formula and one accepted answer. I share the frustration when teams disagree because no one has explained the population or period. I am more cautious when the proposed remedy assumes that every difference is a mistake. Before standardising a management measure, I want to know whether the teams are actually answering the same question. A shared label is not sufficient evidence that they are.
The fictional Nivralen Bank reviewed seventy-six management measures in 2021. Forty-seven could adopt shared definitions. Twenty-nine required contextual distinctions because their purposes, populations or periods differed. The programme used ₹2.7 crore for the review and associated work. We did not describe the result as complete formula uniformity. We described the agreement that had actually been reached: common definitions where the question was common, and explicit differences where a single formula would hide the reason a team needed the measure in the first place.
This was not a defence of every local preference. A department can become attached to a number simply because it is familiar or flattering. The review needed to challenge those habits. It also needed to recognise that a real business purpose might require a different denominator or observation window. The data office could help compare those choices and explain their effects. It could not determine the purpose without the relevant business owner. A technically consistent measure that fails to answer the intended question is not a useful victory for data governance.
Start with the decision the measure serves
I begin by asking what a recipient is expected to do differently after seeing the number. The answer helps distinguish a descriptive comparison from a signal intended to trigger action. It also reveals whether the period, population and level of aggregation fit that use. Two teams may both discuss activity while one is interested in a current operational workload and another is examining a longer-term relationship. If the questions differ, forcing an identical calculation can move the disagreement out of the meeting without resolving the underlying need.
The name should then be tested against the definition. Does it tell a reasonable reader what is included and excluded? Does it suggest a whole population when only a subset is observed? Does it make a current state sound like a completed outcome? These questions connect to my branch-average mistake. The formula produced an equal-weight average across branches, but my words described customer experience. The error was not repaired by insisting that the calculation followed a standard rule. The rule needed to match the claim being made from it.
In the 2021 work, the forty-seven shared definitions were valuable because they removed avoidable ambiguity. Agreement meant colleagues could discuss the result without repeatedly reconstructing its basic meaning. Yet even a shared definition does not establish that every source record is correct. It clarifies the intended calculation and interpretation. Source quality, timeliness and the suitability of a particular decision still require their own examination. I want those questions to remain visible so that the success of a definition exercise is not used as evidence for an unrelated assurance claim.
Agreement can preserve a legitimate difference
The twenty-nine contextual measures required names and explanations that made their differences understandable. That was a more demanding communication task than choosing one winner. It required us to explain why two values could differ without either being wrong for its purpose. Where comparison was useful, the account needed to identify what changed between the populations or periods. Where comparison would mislead, the presentation needed to say so. The result should help a management reader avoid a false reconciliation rather than make disagreement seem permanently unavoidable.
Some colleagues worried that allowing contextual definitions would provide an excuse for continued fragmentation. I took that concern seriously. A contextual distinction should be supported by a stated business purpose and an accountable owner, not merely by a preference to keep a local spreadsheet unchanged. The review should make the reason inspectable. If the purpose later disappears, the distinction can be reconsidered. This creates a different form of discipline from blanket uniformity: each difference has to explain itself, and each shared definition has to remain appropriate for its agreed use.
There is also a cost to carrying several meanings. Readers need orientation, and teams need to maintain the explanation when their work changes. We accepted that cost where the alternative would conceal a genuine difference. The decision gave up the easiest programme headline. It would have been simpler to announce that all seventy-six measures now followed one formula each across the institution. Instead, the account retained the split. I considered that a more useful statement because it told management what had been resolved and what had deliberately remained distinct.
Keep the definition open to an accountable change
A definition can become obsolete without anyone making an obvious technical error. The business may change the process, the population may broaden or the recipient may begin using the number for a new purpose. I want the institution to notice that shift before continuity of the label creates a false sense of continuity in meaning. A change should have an owner and an explanation. The historical account may also need to show which definition applied at the time, rather than silently presenting a new meaning as though it had always been used.
That concern connects definition work with versioned evidence. If a management report changes its calculation, can the reader still understand an earlier decision made under the previous one? The answer may require more than retaining old values. It may require preserving the population and rule that produced them. I would not ask every director to inspect the detailed calculation. I would ask whether management can explain a material change in terms that match the decision before the board, and whether the responsible function has reviewed the consequences of comparing across it.
For a CDO's public profile, this subject offers a way to show judgement beyond a list of platforms or analytical methods. The leader can explain the disagreement, the business purpose considered and the uniformity that was consciously declined. A real account should use permitted information and avoid publishing internal definitions that are restricted. The professional value lies in making the reasoning clear. A visitor should understand why the leader chose a qualified agreement rather than assuming that every number with the same label needed to become identical.
I want consistency to mean that the institution knows what it is saying. Sometimes that requires a common formula. Sometimes it requires clearer language for a legitimate difference. The seventy-six-measure review ended with both. Its value was not that it removed every future debate about data; it gave those debates a more precise starting point. When a number is questioned, the owner can explain the business question, the population and the agreed interpretation. That is a stronger basis for management judgement than a uniform surface that leaves the original question unanswered.