An efficiency claim begins with the unit you choose

## The arithmetic was right and the conclusion was too large

I once presented a lower cost per policy as evidence that our service organisation had become substantially more efficient. The calculation was correct, the source figures reconciled and the result was encouraging. It still needed a challenge. The business had shifted towards more individual policies and fewer family contracts. A policy was not the same unit of human coverage in the two periods. I had moved from a statement about a ratio to a statement about operating performance without examining the bridge between them. The mistake is easy to make because a clean calculation feels like the end of an enquiry rather than the beginning of an interpretation.

The fictional comparison used ₹4.2 crore of annual service expense over 36,000 policy-years, followed by ₹4.32 crore over 48,000 policy-years. That gave approximately ₹1,167 and ₹900 per policy-year. The numerator's scope was held constant and the periods were defined. Those are necessary features of a useful comparison, but they did not remove the effect of mix. Covered life-years in the same account moved from 84,000 to 86,000. On that basis the cost was ₹500 and approximately ₹502. Neither calculation was a trick. They described different relationships within the same operating account, and the difference mattered to the staffing recommendation being considered.

It would have been equally careless to replace the first headline with a claim that efficiency had deteriorated because the second ratio rose slightly. Some administrative work follows the contract, some the person covered and some the service event. A family contract can reduce certain repeated tasks while creating other kinds of work. The correct denominator depends on the question. The actuary also reminded us that an administrative cost comparison is not a conclusion about claims risk. A finance team should be able to preserve these distinctions without making the account so complicated that colleagues can no longer decide anything.

I withdrew the broad claim and asked for a more specific explanation. The change was uncomfortable because I had already used the number to support a positive account of performance. Yet defending the first interpretation would have made the ratio more important than the work it was supposed to explain. I wanted operating colleagues to see that their challenge could change the executive summary, not merely be acknowledged in a footnote. Finance earns credibility when it is precise about the limits of its own conclusion. That can include correcting an attractive claim before anyone outside the company asks the same question.

## Let the operating task challenge the ratio

We funded ₹24 lakh of activity and cost analysis with operations. The purpose was to understand which tasks changed with the contract count, which related more closely to covered people and which depended on specific events. This was a management study, not an attempt to replace statutory accounting or actuarial assessment. We wanted to know what the service team actually had to do before deciding whether four positions could be removed from the plan. A reduction might ultimately be justified, but the initial ratio was not enough to support it. Holding the proposal also had a cost that belonged in the discussion.

The analysis required conversations that did not begin with a preferred efficiency result. A task can disappear because a process improved, move to another team, or become less frequent because the business mix changed. Those explanations have different implications for the next period. If work moved, the company should not claim a saving solely because one function's account fell. If mix changed, the improvement may not recur when the portfolio changes again. If a process improved, management still needs to understand whether the new method can sustain quality and handle unusual cases. The same favourable number can therefore lead to several different operating decisions.

I also wanted the time basis to remain clear. Policy-years and covered life-years describe exposure over a period, while a year-end snapshot answers a different question. Mixing an annual expense numerator with an unexplained closing count can create a further distortion. Our defined comparison kept the periods aligned, but that discipline should not be assumed whenever a familiar ratio appears in a pack. The reader needs to know what entered the count and what did not. That explanation can be short. It simply has to be present before a unit measure is used to justify a consequential change in capacity or spending.

Operating leaders sometimes worry that this kind of enquiry will turn every result into an argument about definitions. That is a fair concern when finance keeps refining a model without making a recommendation. The answer is to connect the definition to the decision. If the proposed action is a staffing reduction, explain which work is expected to disappear or become easier. If the action is an investment, explain which activity it will change and how that will be observed. A ratio is useful when it helps answer that question. It is less useful when it becomes a ceremonial measure that every proposal must make look better.

## Retain more than one useful view

The final management account retained both policy-year and covered-life-year views, with activity analysis alongside them. It did not crown a single denominator as universally correct. That was a more modest result than the first executive headline, but a better basis for reviewing capacity. The board could see that expense had risen, contract volume had changed materially and covered-person exposure had moved much less. It could then assess the recommendation with an understanding of what the figures did and did not establish. We did not need to make one measure wrong in order to explain why another was also necessary.

This approach also changes how a finance team discusses its own contribution. We can claim to have improved the decision account without claiming to have discovered a saving that operations has not delivered. The ₹24 lakh study was a cost, and the held staffing reduction was not cash generated. Any later change would need its own implementation and outcome evidence. That discipline protects the credibility of future recommendations. Colleagues are more likely to trust a favourable efficiency account when they have seen finance retain an inconvenient distinction in an earlier one. The aim is a consistent standard of explanation, not permanent scepticism about improvement.

For directors, a useful question is whether the unit remains comparable as the business changes. They do not need to reconstruct every operational calculation themselves. They can ask management to explain a material shift in mix, show a second relevant view and connect the conclusion to the proposed action. They should also allow a clear answer to remain conditional where the evidence is conditional. A board that expects every paper to end in a single confident percentage may unintentionally reward the omission of the very distinction it needs. Concision should reduce clutter while retaining the facts that change judgement.

My practical rule is to ask what changed in the counted population before naming the saving. The answer may support the original conclusion, narrow it or overturn it. In our case it narrowed the conclusion and delayed a staffing decision while further work was done. Another business may find a clear process improvement that survives several sensible denominators. Both are useful outcomes. What matters is that the company understands the operating change behind the arithmetic and can explain why the chosen measure supports the decision. The number should serve that judgement, not become a substitute for making it.