Chief Financial Officer / India
The corrections we made before committing
- Annual service revenue
- ₹124 Cr
- Mirelva; year ended March 2026
- Institutional clients
- 18
- Loan administration and reconciliation support
- Finance colleagues
- 47
- Led through four managers
Decisions & consequences
01Repricing the accounts the contract would actually bill
In February 2025 a proposed loan-administration contract was presented using 90,000 accounts at ₹120 per account per month. That produced ₹12.96 crore of annual revenue against ₹11.4 crore of recurring direct cost. The commercial team expected a contribution of ₹1.56 crore before central overheads. During the finance review, we compared the client's account file with the draft billing definition. Only 74,000 accounts qualified: dormant accounts and records still awaiting migration acceptance were excluded. The larger population was useful for planning some operating work, but it was not the population the client had agreed to pay for.
At 74,000 accounts, the same rate produced ₹10.656 crore annually, leaving a ₹74.4 lakh shortfall against the recurring direct-cost plan. Operations confirmed that the proposed staffing and supervision could not fall in proportion to billable volume. Commercial colleagues argued that reopening the price would damage a relationship developed over several months. I asked them to distinguish that genuine relationship risk from the certain weakness in our own model. Finance could not approve a margin calculated from accounts outside the contract definition.
The client accepted a revised ₹132 monthly rate after further negotiation. At the same planning volume, recurring revenue became ₹11.7216 crore and the contribution before central overheads ₹32.16 lakh. A separate ₹25 lakh mobilisation fee covered agreed transition work; we did not include it in recurring contribution. These were approval assumptions, not a claim about revenue already earned. The final contract retained an explicit account-count reconciliation and a process for reviewing material scope changes.
We accepted a much smaller recurring contribution than the original paper suggested and delayed the planned start while the terms were settled. My decision was to make that trade visible before commitment. It did not establish that every future exception would be profitable, but it gave the board and operating team a common description of the business they were accepting.
Removing client balances from a distribution discussion
A November 2025 planning dashboard showed ₹25.6 crore as cash available across accounts visible to our reconciliation teams. The report had been assembled from a broad operating extract. It included ₹7.2 crore in accounts legally owned by clients alongside ₹18.4 crore in Mirelva's own bank accounts. Our teams could view the client records under service arrangements, but could not use those balances to meet our obligations. No client money had entered a company account. The error was the decision report's failure to distinguish ownership before aggregation.
The inflated view appeared in preparation for a proposed ₹5 crore shareholder distribution. Employee shareholders had waited for a return, and the commercial team expected receipts to strengthen the position shortly. I asked treasury to reconcile our own balances to the bank-account register and remove every client account from the company cash view. I then showed the board the corrected starting point and the cash commitments falling due before the next dependable receipts. Expected collections remained in the forecast, with their timing uncertainty stated separately.
The board deferred the distribution pending a refreshed cash and obligations review. That was a real disappointment to colleagues who were also owners. I supported the deferral and explained that their ownership did not make client balances available to us or turn an anticipated receipt into cash already held. We added a checked legal-owner field to the reporting source and restricted the company treasury view to approved company accounts. Client-operation reporting continued in a separate view for its proper purpose.
The result was a corrected basis for a capital decision, not a recovery of ₹7.2 crore or evidence of a missing bank balance. My responsibility was to interrupt the decision while the misleading report could still be corrected. The case reinforced a simple discipline: before discussing what a balance allows us to do, establish whose balance it is.
Accepting the credit before pursuing the remaining invoice
A ₹1.1 crore invoice for an annual service period remained unpaid more than four months after issue. The account team described the delay as a difficult client's collection behaviour. Operations had meanwhile documented processing delays that supported a ₹36 lakh credit under the agreed service terms. The proposed collection forecast continued to include the full invoice while commercial discussions were unresolved. I asked for the service evidence, the contractual calculation and a clear account of which amount we could still defend.
The relationship manager wanted to retain the full claim as a negotiating position. Operations worried that accepting the credit would be read as a judgement on the whole team's performance. I separated the financial adjustment from that broader performance discussion. The supporting record established a specific service shortfall and a corresponding credit; it did not establish that every activity under the contract had failed. Keeping the larger receivable in a collection forecast would conceal the distinction rather than strengthen our position.
After commercial and legal review, we agreed the ₹36 lakh credit and finance recorded the adjustment under the applicable reporting policy. The revised balance was ₹74 lakh. The client paid ₹68 lakh in February 2026, leaving ₹6 lakh unresolved at the review date. I required the report to show the credit, the cash received and the residual amount separately. We did not call the full ₹74 lakh collected, nor describe the credit as an unexpected improvement in working capital. It reduced the amount the company was entitled to pursue.
The decision gave up part of the original billed income and exposed weaknesses in how service disputes reached finance. We introduced a joint monthly review of disputed invoices, with operating evidence attached before a collection estimate was approved. That process does not remove disagreement with clients. It makes it harder for a known service adjustment to remain hidden inside an optimistic cash forecast.
Keeping a software approval separate from its expense profile
A reporting-system proposal came to the board with a total planned cash cost of ₹2.8 crore. Its summary suggested that most of the programme would be capitalised, making the immediate effect on profit appear modest. The supporting budget mixed eligible development work with exploratory work, training and ongoing support. Our technology head wanted a quick approval before a supplier slot expired. The investment case also described staff capacity released as though it would automatically become a payroll saving.
I asked finance to review the cost categories with the external advisers against the company's applicable accounting policy. The resulting planning assessment identified ₹1.1 crore of development expenditure that could qualify for asset recognition if the required conditions and evidence were met. The remaining ₹1.7 crore comprised ₹60 lakh of exploratory work, ₹48 lakh of training and ₹62 lakh of support costs. Final recognition would follow the actual work and evidence, not the approval label. I put the full ₹2.8 crore cash schedule beside that assessment.
The operating review also showed that colleagues expected to gain capacity would remain employed to handle unresolved client exceptions. We removed an unsupported payroll-reduction claim from the return calculation. The board approved a staged programme with a first release of ₹1.2 crore and a review before the remaining commitment. That first release was part of the ₹2.8 crore envelope, not an additional amount. We retained the reporting-quality objective while requiring evidence of usable outputs before expanding the work.
This decision did not reject software investment or prescribe an accounting outcome in advance. It rejected the idea that a favourable expense profile could settle the funding and benefit questions. The technology team accepted a slower release of funds, and finance accepted responsibility for maintaining a transparent cash and recognition record. The company could then assess the programme on the work delivered, the obligations incurred and the resources genuinely released.