I examine the conditions that make an operating promise usable.

In 2021 we examined twelve disruption scenarios involving customer operations and their technology and supplier dependencies. The exercise had been scheduled, staffed and completed. That made it tempting to present completion itself as a positive result. Eight scenarios met the conditions agreed for the exercise; four did not. In those four, the relevant component could be available while a subsequent customer action remained blocked or required authority that the exercise team had not secured.

I asked for the four failures to remain prominent in the management account. The trade-off was practical and reputational inside the organisation. A simpler completion message would have supported confidence in the programme and allowed the team to move on. A more accurate account required additional work, expenditure and discussion with colleagues whose local recovery measures had appeared satisfactory. I owned the operating coordination, while the technology and risk leaders retained their separate responsibilities.

We assigned an operating owner to each failed condition, traced the unresolved handoff and agreed the evidence needed before a retest. Management funded AUD 780,000 in follow-up work, including process changes, additional rehearsal time and support arrangements. We did not aggregate unlike recovery times into one reassuring average. Each scenario retained its own starting assumptions and completion conditions, so a later result could be compared with the question the original exercise had actually asked.

The result was an accountable correction programme rather than a claim that the bank had become immune to disruption. The four scenarios were scheduled for retesting after their work was completed; they were not retrospectively marked successful. The enterprise gained a clearer distinction between an exercise that happened and a capability it had demonstrated. I would involve the customer-operations participants earlier in setting the exercise conditions next time. Waiting until the result discussion to expose the difference between a restarted component and a completed task created avoidable disagreement about what success was supposed to mean.

A 2023 review covered twenty-four supplier arrangements approaching a commercial decision. Nineteen had sufficient support for their planned renewals. Five required a different discussion: the bank wanted a transition, but the receiving operations had not demonstrated all of the capacity, record-handling and reconciliation steps needed to take the work. Contractual provisions gave us routes to act. They did not establish that those routes could be carried out on the proposed timetable without leaving customer work stranded.

I recommended six-month extensions for those five arrangements, at an additional cost of AUD 1.4 million. Procurement colleagues were understandably concerned about the commercial position. Operating teams also worried that an extension might reduce the urgency to prepare. The alternative, however, was to let a desired end date stand in for evidence of readiness. I put the recommendation to the appropriate decision-makers with its expense and remaining dependencies visible.

For each arrangement we set out the work to be received, the responsible operating team and the evidence required from a rehearsal. Legal and information-security specialists reviewed the matters within their remits. We separated those assessments from the question of whether colleagues could perform the actual transition sequence. The extensions carried funded work and progress reviews; they were not described as an unlimited permission to defer. We also avoided assuming that a supplier’s willingness to help would remain unchanged under a severe disruption.

The immediate outcome was continuity through an explicitly costed preparation period. It was not five completed exits, nor a saving against a hypothetical future arrangement. That distinction changed the way management read the programme: expenditure bought time for evidence, not the evidence itself. Looking back, I would have begun operating rehearsal much earlier in the commercial cycle. Raising the concern close to a decision date narrowed our choices and made an otherwise sound operating argument unnecessarily expensive to act on.

In a fourteen-day review during 2024, we followed 312 mortgage settlements through to their expected completion dates. Of those, 264 completed on time and 48 were late. The lending-administration dashboard had given a more comfortable impression because it emphasised completion of our internal processing stage. That stage mattered, but it was not the event the borrower was waiting for. A file could leave one queue while a missing instruction or unresolved handoff prevented the settlement from finishing.

I required the report to show the customer completion result alongside the internal milestone. The decision made our performance look worse in the short term and created work to establish the reasons behind individual delays. Credit decisions and legal judgements remained with the people authorised to make them. We did not instruct administrators to bypass a control in order to improve the date measure. The aim was to distinguish a necessary unresolved decision from an avoidable operating delay.

Teams traced the forty-eight late cases, established the next responsible participant and corrected the associated communication. We recorded AUD 107,000 in corrective fees and expenses, separately from the value of the mortgages. The case review preserved the different causes rather than assigning every delay to the last team that touched the file. Where a customer or specialist response was required, that condition remained visible. A completion date could not honestly be promised simply because we had finished our own checklist.

The useful outcome was a changed management account and a clearer follow-through on unfinished settlements. I do not claim that the review established a permanent improvement rate or removed every source of delay. It showed that our chosen milestone had been too narrow. I should have challenged that definition earlier, especially because my career began with precisely this kind of gap between local completion and the customer’s experience. The corrective lesson was to make the end condition part of the operating design before it becomes a performance debate.

In 2015 a distribution proposal included closing eight branches. The property and staffing plans were sufficiently developed to make the programme look ready. Yet the operating account of replacement access varied considerably. Some customer tasks could move to the proposed channels with workable assistance. At three locations, important arrangements were still assumptions rather than demonstrated alternatives. I led branch and central operations at the time, so I could not reasonably treat this as a question for the property team alone.

I supported postponing those three closures while proceeding with five whose arrangements had been tested. The delay and additional access work cost AUD 920,000. That was an unwelcome departure from the original timetable, and we had to explain why the bank could not apply one identical decision to all eight locations. I wanted the discussion to concern the tasks people needed to perform, rather than stereotypes about which customers would or would not adopt a particular channel.

The operating teams documented the affected tasks, trialled the proposed alternatives and identified where additional support was needed. We worked with the relevant customer and distribution colleagues on communication and exceptions. The five closures that proceeded were tracked separately from the three deferred locations. By 2018 the five completed changes showed AUD 1.1 million in measured annual premises and operating expense reductions. That figure was not the original forecast for all eight sites, and it did not establish that every affected customer preferred the new arrangements.

The enterprise gained a more discriminating decision: proceed where the evidence supported it and keep unresolved access visible elsewhere. I would now insist on the task-level examination before a closure timetable becomes an internal commitment. Introducing it late made the delay feel like a failure of execution, when part of the problem was that the proposal had asked operations to execute an insufficiently tested assumption. The cost was real, and so was the responsibility to explain why incurring it was preferable to pretending the assumption had been resolved.

As lending-administration manager in 2008, I inherited a growing concern about aged records and initially accepted the combined backlog as the main measure of the problem. A detailed review of 1,040 entries showed why that was inadequate. There were 740 records with work we could complete, 210 awaiting a customer response and 90 duplicate entries. The categories reconciled to the reported total, but they did not call for the same action. Treating them alike encouraged repeated handling without an explained route to resolution.

I proposed a nine-week correction programme costing AUD 320,000. It meant diverting experienced colleagues from normal improvement work and acknowledging that my earlier reporting had overstated the usefulness of a single age measure. We did not remove a genuine customer obligation by calling it an exception. The purpose of classification was to establish the next action and its owner, with a review route where the status was uncertain.

Teams completed the work on the 740 actionable records through the existing controls. The 210 awaiting customer input were placed into an explained follow-up process; they were not represented as completed transactions. The ninety duplicates were reconciled and removed from the administrative count after verification. We kept the source record and reason for that correction so that a lower total would not conceal an unexplained deletion. Credit approval remained outside my remit throughout the exercise.

The programme produced a more useful account of work and a completed subset that could be evidenced. It did not mean that 1,040 customer cases had been resolved, and the AUD 320,000 expense was not a reduction in credit losses. That precision changed my approach to operating reports. A total becomes useful when it points to decisions colleagues can make. I would have commissioned the category review earlier instead of allowing an age distribution to become the whole story. The discomfort of correcting my own report was less costly than continuing to manage several different problems as though they were one queue.