The hire that never happened is absent from the transaction file

The two totals looked ready for a decision. Over the same twenty-eight days, one depot completed 120 hires and another 80. Colleagues wanted to allocate the next equipment investment to the first. The count was higher, the period matched and the chart was easy to understand. Then Harish asked how many days the second depot had anything available to hire. It had cycles available on sixteen days and none on twelve. The transaction file had recorded what happened within those service conditions, not what every customer might have wanted.

That does not prove the second depot had stronger demand. It makes the proposed ranking less certain. The distinction matters because analytical corrections can create a new confident story as quickly as they dismantle the old one. Once availability enters the conversation, it is tempting to make every low count a supply problem. We had no basis for doing that either. Customer interest, equipment availability, the timing of requests and other local conditions needed a more suitable account.

Follow the path into the completed record

A completed transaction is the end of a path through the service. Someone must want an arrangement, the business must have something suitable to offer, and the parties must complete it under the relevant terms. The file contains those who reached the end. Depending on how the service records enquiries, it may say little about those who encountered no availability or decided not to continue. Treating the final count as a complete demand measure silently assumes that the missing paths do not change the comparison.

Our cargo-cycle rental setting made the issue tangible. A depot with no cycle available could not produce the same type of completed record as one able to offer a hire. But the absence of a record did not tell us whether one customer, ten customers or nobody had wanted a cycle that day. It also did not tell us whether the available equipment would have suited the work requested. Operating colleagues retained those judgements. Data analysis could identify the missing evidence without pretending to supply it.

The argument for the first depot was still reasonable. It had demonstrated completed business within the observed period. That fact might matter in an allocation discussion, especially if the company needed a near-term commitment. I did not ask colleagues to ignore it. I asked them to distinguish evidence of completed activity from a conclusion about relative market demand. A board or executive team can choose to act under uncertainty, but the account should say which uncertainty it is accepting.

The first proposed repair was arithmetic. Divide the second depot's eighty hires by sixteen available days, then extend that rate over twenty-eight. The result would look more comparable because each depot would have a full-period figure. Yet the calculation assumed that unavailable days would behave like available ones. We had not established that. The timing of availability might itself be related to demand, and the unobserved days could differ in other ways. A smoother table would not make those assumptions disappear.

Do not repair missing demand with a convenient rate

I declined the extrapolated demand claim as well as the original ranking. That decision was frustrating because it removed two quick routes to an answer. It also protected the discussion from a false choice between treating eighty as the whole market and inventing a larger total from a simple rate. The useful next step was to ask what evidence would distinguish the explanations that mattered to the investment. We commissioned ₹32 lakh of observation work and deferred the expansion conclusion.

The programme connected appropriately collected request evidence with availability. Its purpose was the defined depot decision, not a broad invitation to collect every customer movement or assemble unrelated personal records. Qualified owners needed to agree the collection and use. The research also required operating context: which requests the service could reasonably consider, and which equipment conditions or terms limited what could be offered. A number of enquiries alone would not settle the suitability of an eventual investment.

I do not claim a demand increase or a utilisation benefit from that choice. The case records a decision to obtain evidence, with an explicit expense and a delayed allocation. No later result has been invented to make the story feel complete. That may be less satisfying than a before-and-after improvement, but it is a legitimate executive outcome when the original question cannot be answered responsibly from the available record. The company still has to decide how much uncertainty it can afford to reduce.

There are circumstances in which further observation will not be worth its cost. A small reversible allocation may justify a bounded operating trial sooner than an expensive research programme. An urgent service need may require a decision before the preferred data exists. My argument is not that every transaction comparison must wait. It is that management should state what the record establishes, what it leaves unknown and why the selected next commitment is proportionate to that uncertainty. Those are business judgements, not automatic consequences of a missing field.

Match the research to the investment choice

The research design should begin with the alternatives management is actually considering. If the choice concerns equipment at two depots, the relevant evidence must help compare those options within their operating conditions. A general customer-interest survey may be interesting without resolving that allocation. Equally, a technically complete availability history may still leave the request pattern unclear. The data function should connect the observations to the decision rather than accumulate detail simply because it can be collected.

My early survey work made me attentive to who reached a questionnaire. Rental analytics added a different version of the same concern: who reached a completed hire. The analogy is useful, but it should not flatten the differences. A depot's equipment and service choices actively shape which transactions become possible. That means a demand account can contain the effects of earlier management decisions. Ignoring those conditions may cause the next allocation to reinforce the pattern the company itself created.

I would now bring availability into the first specification of a market comparison. I would ask operating colleagues which stages of a request are recorded and which are not, and agree how the limits should appear in the management account. The aim is not a report so cautious that it cannot support action. It is a report that helps management choose the right kind of action: commit, observe, test or defer, with the reason made explicit.

For directors, the case offers a practical question about growth evidence. Does the transaction total describe customer demand, service capacity or an interaction between them? The answer may change the investment argument without changing a single recorded hire. A completed file can be accurate and still be insufficient for the conclusion attached to it. Recognising that limit allows the board to examine the next commitment without asking a familiar chart to supply customers and journeys it never observed.