Chief Operating Officer / India
What the service promise required in practice
- Records centres
- 4
- Controlled-access service locations
- Operations colleagues
- 228
- Within a company of 285 employees
- Operating budget
- ₹27 crore
- Approved annual remit; not client loan value
Decisions & consequences
01Accepting the record set rather than the parcel count
A lender's regional archive transfer involved 24,600 mortgage files entering our contracted custody service. The commercial timetable assumed that the sending register would become our opening inventory once the containers arrived. During preparation, staff found examples where the listed reference existed but the enclosed set did not agree with the register. Continuing on the original basis would have made it difficult to distinguish an incoming difference from a later handling problem.
I asked the transition lead and the client's appointed representative to agree an exception procedure before final acceptance. We continued receiving containers, but separated disputed entries from the routine inventory. The review identified 386 differences, including absent listed enclosures, duplicated references and material assigned to the wrong file. Our staff recorded what was physically present; we did not determine the legal effect of a missing document or decide whether a loan was enforceable.
The client initially preferred acceptance followed by later reconciliation. I recommended a nine-working-day extension for the affected population and ₹6.4 lakh of additional checking effort, funded through an agreed service variation. We assigned each difference to the party able to provide evidence. Of the 386 entries, 301 were reconciled during the extension; eighty-five entered a separately agreed unresolved schedule with restricted service instructions. The remaining 24,214 entries could enter routine service without waiting for those exceptions.
The outcome was a more explicit opening position, not a claim that every original record was complete. The client's lending and legal teams retained responsibility for consequences outside our service. In retrospect, we should have agreed the difference categories during the sample review rather than after the first containers arrived. I now include that work in the transition estimate. The lesson was commercial as well as operational: if acceptance means more than receipt, the additional examination needs time, authority and a price.
When a completed retrieval still needed work
Our monthly service report showed strong retrieval completion, yet a client team continued to report that requested sets were not ready for use. I asked for a review of 1,140 requests from a defined six-week period rather than debating the average turnaround figure. The review found ninety-two cases marked complete at dispatch that subsequently required further document work, clarification or an authorised replacement enclosure.
The existing measure was not fabricated. It recorded a real event, but the label encouraged readers to infer a later stage. I brought the centre manager, commercial lead and client service owner together to define separate events for preparation, authorised dispatch and confirmed usable handover. The client also had to state what confirmation it could provide and within what process. We could not unilaterally claim that a set was legally sufficient or suitable for every future purpose.
We piloted the revised reporting for eight weeks. Forty-eight requests required additional work after dispatch during that period, against a different volume of 1,208 requests. Those counts were reported with the changed process and cohort, without presenting them as a controlled estimate of improvement. We added two trained checking positions at an annualised ₹14 lakh cost and repriced the relevant service element after the pilot. The original dispatch measure remained visible, allowing the parties to see where extra time was being spent.
I had underestimated how strongly staff associated the old completion figure with recognition for good performance. We explained the new measure before changing individual review discussions and examined cases where the source material, rather than staff conduct, caused the delay. The decision made the report less flattering at first, but more useful. My learning was that a service metric changes behaviour through its label as well as its calculation; both deserve an explicit review.
Dividing an intake before the centres were overwhelmed
A prospective client proposed moving 68,000 files into our centres during one week, followed by immediate retrieval availability. Storage space was available, so the opportunity initially looked feasible. The operating estimate showed a different constraint: our trained indexing and checking capacity was already committed to routine work and another transition. Hiring temporary handlers would not resolve that bottleneck.
I recommended three releases of 22,000, 23,000 and 23,000 files across six weeks. Each release required an agreed inventory, a completed exception review for the preceding population and a capacity decision by the transition lead. We budgeted ₹11 lakh for temporary preparation support and extended shifts in defined areas, while keeping checking and release authority with trained staff. The commercial team explained the revised timetable and its price before signing the final service schedule.
The client accepted staging but asked for the second release to move earlier. We declined that change after the first release produced more indexing corrections than the sample had suggested. The second release proceeded four working days later than originally planned, and the third completed within the revised six-week window. Routine retrieval work at the receiving centres remained within the agreed service range, although one lower-priority internal reorganisation moved into the following month. That displacement was included in the review rather than hidden as a consequence nobody owned.
The case did not prove that smaller releases are always better. The client carried an extended period of parallel arrangements, and our own temporary-support cost was real. It showed why physical capacity and usable service capacity should be considered separately. I would improve the initial sample by including records from more source locations, because the first sample understated variation. The operating promise became more credible when the release decision could respond to what the previous intake had actually required.