Chief Information Officer / India
Where the technical result needed another question
- Technology colleagues
- 21
- Enterprise technology function
- Branches supported
- 32
- Karnataka and Kerala
- FY2026–27 technology envelope
- ₹16 crore
- ₹12 crore operating expense; ₹4 crore capital expenditure
Decisions & consequences
01A restarted application with unfinished servicing tasks
A scheduled recovery exercise restored our loan-administration application within the technical window agreed for the test. The initial report described the exercise as successful. When servicing colleagues attempted sixty selected tasks, seventeen could not be completed because linked documents remained unavailable or their references did not resolve in the recovered environment. The application was running, but the sampled work had not fully resumed.
I asked the service team to preserve the result and reopen the acceptance discussion. The infrastructure provider argued that its contracted restart measure had been met. That was a relevant fact, but it did not answer the business question. We mapped the failed tasks to the document service and reference dependencies, distinguished faults in the recovery arrangement from incomplete test instructions, and assigned owners for both. No live borrower instruction was changed during the exercise.
Management approved ₹23 lakh of remediation and an additional test window. We corrected the document-service recovery sequence, updated reference checks and involved authorised servicing staff in the repeat exercise. Fifty-nine of the sixty selected tasks completed; one remained open because the source attachment itself was unreadable. That item was assigned to the business owner for investigation rather than described as a recovery fault we had solved. The acceptance paper stated the sample and its limitations.
The additional work displaced a planned internal interface enhancement by one quarter. I included that consequence in the decision, since the recovery repair did not create free capacity. My mistake had been accepting a test plan organised primarily around infrastructure components. I now ask a business owner to define the task evidence before the exercise begins. The lesson is not that a technical measure is useless; it is that its meaning must remain narrower than the service conclusion people may draw from it.
Paying for the records needed after a supplier exit
We planned to leave a hosted workflow provider whose service no longer fitted our support needs. The proposed exit package contained account tables and current status fields. A trial extraction showed that historical attachments, reference explanations and some earlier approval context would not be included in the standard package. The files could be opened, but colleagues could not reliably reconstruct selected past decisions from them.
I recommended delaying the final exit while the contract and operating teams agreed a more usable handover. The project sponsor wanted to retain the original date to avoid another period of dual service. I showed twelve reviewed historical cases, identified the missing material and asked the relevant business owners to state what they needed for ongoing work and retention. We did not assume every historical item should be retained forever; the scope followed the approved requirements and qualified advice.
The revised plan added six weeks and ₹18 lakh for extraction, mapping and verification. We tested a defined sample of 240 records across different periods and workflow states. Two hundred and thirty-four could be reconstructed from the package; six required additional source clarification before sign-off. We retained a record of the final unresolved limitations and the business decisions about them. Access to the extracted material was limited to the authorised transition team and later service owners.
The exit completed after the revised acceptance process, with dual-running expense included in the total transition account. I would have reduced that cost by testing the package earlier in the contract term, when the supplier relationship was less focused on departure. The case changed our procurement review: an export promise now needs an agreed practical test. A file format is only part of an exit capability; the business also needs enough context to understand and responsibly use the material it receives.
Repairing three decisions before buying a reporting platform
A proposed reporting-platform replacement carried an estimated ₹1.9 crore initial cost. Its business case promised a common view across branches, finance and servicing. During preparation, I asked the sponsors to identify three recurring decisions that the existing reports made difficult. The examples concerned overdue-account classification, branch productivity and the timing of fee recognition. In each, teams were using different definitions or cut-off conventions rather than simply lacking a modern display.
The platform sponsor argued that a new tool would force consistency. I disagreed that software alone could resolve choices the business had not made. I proposed a ₹31 lakh programme to agree ownership, reconcile inputs and rebuild the three reports within the existing environment. Finance and operations each nominated decision owners, while technology documented the calculation and change process. Accounting judgements and customer-treatment decisions stayed with their authorised functions.
The smaller programme ran for four months. In the final reviewed cycle, the three reports reconciled to their agreed source populations, and twelve recurring manual adjustments were either incorporated into the approved logic or removed with an explanation. We did not describe the difference between ₹1.9 crore and ₹31 lakh as realised savings, because the replacement had never been committed and some future capability needs remained. The board received both the completed work and a list of limitations that could justify later investment.
My recommendation gave up the broader interface improvements and self-service features in the original proposal. Some managers remained dissatisfied with those limitations, which was a reasonable consequence rather than evidence that they had misunderstood the decision. I would now establish the decision-owner group before commissioning a replacement estimate. The case reinforced a distinction I want technology leaders to make: a smaller scope is useful when it addresses the actual uncertainty, not merely because it costs less.