One household can appear in three successful channels

A household hears about a lender at an event, asks a referral partner a question and later calls a branch. Each interaction may be useful. Each channel may also report the household as an enquiry it generated. By quarter end, the marketing report can describe three successful acquisition routes while the company has one prospective relationship. The problem is not that only one interaction deserves credit. The problem is that the report has not decided what it is counting. Until it does, a comparison of channel costs can reward the route best at recording activity rather than the route that helped the household progress.

Choose the unit before judging the channel

An interaction, an enquiry, an applicant and a borrower answer different questions. The distinction should be settled before a dashboard turns them into a league table. Interactions can help us understand workload and reach. Distinct enquiries can describe a population expressing interest under an agreed definition. Applications and subsequent lending decisions belong to a further process with additional conditions. Marketing can influence that journey without controlling every result. I want the report to preserve those stages so a team is not encouraged to make an early count look like a later commercial outcome.

Our quarterly review began with 8,600 enquiry entries. Under the agreed matching method, 2,666 were repeated entries, leaving 5,934 distinct enquiries. The reduction was 31% of the original entries. It was not evidence that demand had fallen by that amount during the quarter. We had changed our understanding of the records, not observed thousands of households withdraw interest. That explanation mattered because a correction can otherwise create a false story of business deterioration. The original number and the revised number needed to remain visible with a clear description of why they differed.

The exercise also challenged ownership. A regional programme manager could reasonably say that an event introduced the household to Avirneli. A partner might have explained the next step, and a branch colleague might have answered the question that led to an application. Giving the entire relationship to the last recorded contact would simplify reporting while ignoring the work that preceded it. Giving every channel a separate new enquiry would exaggerate the population. We therefore separated the first recorded contact from later assistance, allowing the report to describe different contributions without multiplying the household.

That distinction does not create a perfect account of causation. We cannot assume the first recorded contact was the first exposure, or that a later interaction made no difference because it did not receive the primary count. People do not organise their decisions around our reporting system. They may forget an earlier encounter, use a shared telephone number or return after a long interval. A useful measure should be honest about those limits. Its purpose is to improve the spending decision, not to claim that the company has reconstructed every influence on a household's choice.

Keep uncertainty in the matching method

Matching records requires care because a false merge can be as misleading as a missed duplicate. Similar names or shared contact details do not always identify the same enquiry. Families may share a number; a person may enquire about different circumstances at different times. We kept weak matches outside the confirmed repeated-entry count and documented the method used. That left some uncertainty unresolved. I preferred an explainable result to a cleaner number produced by treating every resemblance as identity. The work should not silently remove people because an analytical rule makes the spreadsheet easier to reconcile.

The method also needs a stable reporting period and a clear definition of a renewed enquiry. Otherwise, teams can disagree about whether an older relationship should count as new simply because a quarter has changed. The appropriate rule depends on the business question. I want the definition agreed with the people who will use the result, including operations and finance, before it becomes a performance measure. Changes to that definition should be explained alongside the trend. A chart can appear to improve even when the only change is the boundary used to count its population.

Access to the underlying records should remain proportionate to the task. Marketing does not need every detail collected later in a lending process to understand repeated enquiries. The review should use authorised fields and a controlled method, with the appropriate teams handling sensitive records. This is partly a question of disciplined analysis: collecting more data does not automatically resolve ambiguity. It can create additional obligations while distracting the team from the definition it actually needs. A report should be able to explain its unit and limitations without circulating household financial material through a campaign discussion.

People whose results change need a chance to examine the method. That is different from allowing each channel to choose the rule that makes it look strongest. We reviewed contested cases, corrected genuine errors and kept a common basis for the final report. The discussion became more useful when colleagues could see that later assistance remained visible even when it no longer created another distinct enquiry. Recognition and accountability should not depend on pretending that every valuable interaction is a new person. A commercial organisation can value collaboration while still maintaining a credible population count.

Use the corrected count for a real decision

The review supported moving ₹42 lakh in the following quarter's channel allocation. We did not describe that movement as a saving or claim that it immediately increased funded loans. The money remained within the programme, with a different allocation based on a better account of reach and assistance. The decision also considered channel cost, regional capacity and the next questions we needed to test. A corrected count is an input to judgement. It does not replace the commercial work of deciding whether a channel reaches households the company can serve responsibly.

A useful follow-up asks whether the revised allocation produced the kind of evidence expected. Did the programme reach a different population? Did branch workload change? Were households clearer about the next step? Those questions can be examined without promising that every outcome will improve. Credit assessment and the household's circumstances still affect whether an application becomes a loan. I want the marketing team accountable for the contribution it can substantiate, while working with the other functions to understand the whole journey. Assigning all later outcomes to the campaign would be as misleading as ignoring them entirely.

At board level, a large enquiry total should invite a small set of precise questions. What is the unit? How are repeated contacts treated? Has the definition changed? Which decision is management asking the number to support? Directors do not need to inspect individual records to ask those questions. They need enough confidence in the measure's meaning to judge the investment proposal. Management should be able to explain why a lower corrected total can support a stronger decision, and why a channel that assists many people may deserve funding even when it introduces fewer new enquiries.

The most productive result of our review was not the smaller number by itself. It was a more useful conversation between teams that had been claiming overlapping success. We could discuss introduction, assistance and progression without insisting they were the same event. That made the allocation more deliberate and the report more candid. A household should not have to become three people in our system for three teams to receive credit for helping it. When the count respects the relationship, marketing can make a clearer case for the money it asks the company to commit.