A percentage can be arithmetically correct and strategically misleading. The problem may lie in the conclusion attached to it rather than the calculation itself. I learned that through a freight account in which I used the share of loaded journey legs to describe the share of balanced circuits. The numbers were not fabricated, but the interpretation changed the question they appeared to answer. Correcting that distinction required more than checking the division again.
The fictional 2006 cohort at Navrathin Logistics contained 720 journey legs, 480 of them loaded. Those legs formed 360 two-leg circuits. Of the circuits, 120 were loaded in both directions and 240 were loaded in one direction only. I had used 480 divided by 720 to suggest that two-thirds of the return circuits were balanced. The relevant circuit calculation was 120 divided by 360: one-third. Mihira Dervankar challenged the account, and I withdrew the stronger interpretation while supporting ₹32 lakh review expense.
## Keep the conclusion about the unit being counted
The loaded-leg calculation still had a valid meaning within the stated cohort. Two-thirds of the legs carried a load. The mistake was to allow that measure to become a description of complete circuits loaded in both directions. The units were related but different. Two loaded legs within one balanced circuit and one loaded leg within an unbalanced circuit contribute to the activity count in different ways. The aggregate did not preserve the relationship I was asking it to describe.
Writing out the cohort made the distinction visible. The 120 both-loaded circuits contributed 240 loaded legs. The 240 one-direction-loaded circuits contributed another 240 loaded legs. Together they produced 480 loaded legs across 720 total legs. The same record therefore supported the two-thirds activity measure and the one-third circuit measure without contradiction. I wanted the corrected account to show that structure, because merely replacing one percentage with another could leave readers believing that the underlying data had changed rather than the interpretation.
The error mattered to the strategic question. We were considering the commercial balance of the return journey, not only how often a recorded leg carried something. The loaded-leg share could remain useful as one description of activity, but it could not settle whether complete circuits were balanced. I had let the measure answer a question it was not designed to answer because the more encouraging headline travelled easily into a senior account of the network.
I do not regard this as an argument against aggregation. Senior decisions need summaries. The challenge is to preserve the relationship between the summary and the conclusion drawn from it. A leader should be able to say what the unit is, what period or cohort it covers and why that measure is relevant to the choice. Those questions are often more useful than adding another decimal place to a calculation whose interpretation has already moved beyond its meaning.
## Make the correction visible to people who used the claim
Some colleagues preferred to retain the two-thirds headline because it described a substantial amount of loaded movement. Their concern had a legitimate part: the correction should not make the useful activity disappear. I agreed that the network account needed both views. But keeping the number without correcting the return-balance language would leave the original inference in place. The task was to preserve the valid finding while removing the conclusion I had wrongly attached to it.
That required an explicit explanation to people who had used the earlier account. Quietly changing a heading in the next paper would have left them with different understandings of the same network. I needed to say what I had claimed, why it was too strong and which measure answered the question we were actually discussing. The correction made the circuit account less flattering. That consequence belongs in the leadership record because it shows why acknowledging the distinction could be uncomfortable even when the arithmetic was easy to verify.
I also wanted to keep my responsibility clear. Mihira's challenge exposed the problem, but the interpretation had appeared in material I supported. Describing the response as a general improvement in reporting would remove that ownership and make the case more distant than it should be. A public profile can explain a correction without turning it into a story in which the leader merely discovers a problem created elsewhere. The useful question is how my own judgement changes when a familiar measure no longer supports the account I have given it.
The ₹32 lakh funded review of the analysis. It was not a claimed financial benefit or a price paid for an immediate improvement in the freight network. The consequence was a clearer and less encouraging interpretation, with further work on the underlying account. I do not want a case about restraint in measurement to become an inflated claim about the value of the correction itself. Restoring meaning can be important without establishing a monetary result that the evidence has not measured.
## Do not let the repaired measure become a new overstatement
One-third of the circuits being loaded both ways still did not tell the reader their distances, costs or profitability. A short loaded leg and a long loaded leg both count as one within this cohort. The count also does not establish the commercial value of the goods or the operating requirements of the service. I therefore kept the corrected circuit measure separate from any claim about empty mileage, fuel or financial performance. Fixing one inference should not create permission to make several new ones.
The practical change for me was to ask whether the conclusion remains about the thing being counted. If a paper uses legs, does its claim concern legs? If it speaks about complete circuits, does the evidence preserve their paired structure? The question is simple enough to be used in a senior discussion without turning it into a technical seminar. It also gives the analyst room to explain why an apparently related measure answers a different question and should remain alongside, rather than replace, the one already familiar to the board.
All people, records and expenses in this example are invented. The explicit cohort makes the arithmetic inspectable; it does not describe a real company's operations. A genuine account would need authorised data and consistent definitions. The lesson I would carry into another institution is to check the meaning after checking the calculation. An accurate percentage becomes a useful strategic measure only when the conclusion stays attached to the unit and relationship the evidence actually contains.