The announcement that normal service has resumed can arrive before the operating consequences of an interruption are understood. New work begins moving again. Attention shifts to the cause of the event and the measures intended to prevent recurrence. Meanwhile, a queue remains. It may contain instructions received during the disruption, exceptions created by temporary arrangements and records whose status is uncertain. I regard that queue as part of the recovery, because the customer obligation does not disappear when the system becomes available.
There is an understandable desire to reduce the total quickly. An accumulated queue consumes capacity and can prolong uncertainty for customers and colleagues. But the total may combine work that requires very different treatment. Some instructions can be completed normally. Others need a decision, a missing fact or reconciliation with activity that occurred through another channel. Treating every entry as an identical item to clear can create new errors while making the headline number look better. Speed needs an operating account behind it.
An aged item is a question before it is a unit of work
My concern with this distinction predates my appointment as COO. In 2008 I accepted a lending-administration backlog report that relied too heavily on the combined count and its age. Reviewing 1,040 records revealed 740 actionable items, 210 awaiting a customer response and ninety duplicates. The numbers reconciled, but the work did not have one common next step. My earlier report had described the scale of an administrative problem without giving colleagues enough help to resolve the different conditions inside it.
That experience matters during recovery because disruption can make status harder to establish. A record may appear unfinished in one place even though the instruction was completed through an alternative process. A temporary note may identify an action but omit the authority required to take it. An item may be old because a necessary response is missing, rather than because nobody has worked on it. Age is useful evidence of a possible problem. It does not establish the explanation or authorise the next action.
The classification should be designed around decisions people can make. I want to know what can proceed, what needs review and what requires an external response, with a route for uncertainty rather than a forced category. The categories must also reconcile to the original population. Otherwise a falling queue can conceal records that have simply become difficult to find. Reconciliation does not mean preserving every duplicate as active work. It means retaining an intelligible account of why the administrative count changed and where the underlying obligation now sits.
There is a cost to doing this properly. Experienced colleagues may have to investigate records instead of processing new work, and additional support may be required. In our 2008 correction, the nine-week programme cost AUD 320,000. That expenditure needs a clear purpose and a review of progress. Yet comparing it only with the apparent cost of immediate bulk clearance is misleading. The comparison should also consider the consequences of repeating a completed instruction, overlooking a genuine obligation or making an unsupported correction.
Do not purchase speed by losing the reason
Operating teams need authority to act, but that authority should be specific. A broad instruction to remove the backlog can encourage people to interpret ambiguity in whichever way reduces the count fastest. I prefer an explicit account of the permitted actions, required referrals and evidence to retain. This is especially important when temporary staff or colleagues from another function are helping. Their willingness to contribute does not automatically give them the context or decision rights needed for every item they encounter.
Controls can themselves become a source of delay if nobody knows how to obtain the necessary decision. Naming a control without providing an accessible referral route leaves the team with a prohibition rather than a process. I want the responsible specialist involved in designing how difficult items will be handled. The answer may be a scheduled review, a dedicated contact or a more precise distinction between routine and exceptional work. It should not be an informal bypass justified by the urgency of the recovery programme.
The record of a correction is part of its quality. If an entry is removed as a duplicate, somebody examining the account later should be able to understand the basis for that conclusion. If an item awaits a customer response, the team should know what has been requested and how follow-up will occur. A cleaner dashboard is not enough. The explanatory trail helps the organisation avoid recreating the same uncertainty when another colleague takes over or when a customer asks about the history of the request.
I also separate processing output from customer resolution. Completing work on the 740 actionable records in our earlier case did not mean that all 1,040 entries represented resolved customer transactions. The 210 requiring input remained in a follow-up process. The ninety duplicates were administrative corrections. Reporting those outcomes distinctly prevented a larger but unsupported success claim. During a disruption, when pressure for good news is stronger, that discipline becomes more valuable rather than less necessary.
Explain what remains when normal service returns
Management needs to know whether the recovery effort has enough capacity to handle outstanding work without creating a second problem in current service. That requires a view of incoming demand as well as the older queue. Moving colleagues temporarily may be appropriate, but the consequences should be visible. An apparent improvement in the recovery count can be purchased by allowing new obligations to accumulate elsewhere. I want the operating account to show that movement instead of rewarding one team for exporting the problem to another.
Customers need communication that corresponds to their circumstances. A general restoration notice can be accurate about the service and still insufficient for someone whose earlier instruction remains unresolved. The organisation should explain the relevant next step without asking the customer to reconstruct its internal process. It should also avoid implying that contacting the bank repeatedly is the only way to keep a request visible. The appropriate communication approach will vary, but its purpose is to reduce uncertainty with an honest account of what can presently be done.
The later review should examine why the queue became difficult to interpret. Sometimes the interruption exposed an existing weakness in record ownership or status definitions. Sometimes the temporary response created new ambiguity. Learning from those conditions is different from simply requesting more staff for the next event. Additional capacity may help, but it cannot compensate indefinitely for instructions whose state and authority are unclear. The bank needs to improve the work’s intelligibility as well as its capacity to process it.
I do not regard a zero administrative count as the only possible sign of a responsible recovery. Some matters may legitimately remain open with an owner, a reason and a next action. What I reject is an unexplained remainder, or a disappearance that cannot be reconciled. Recovery becomes credible when the organisation can account for the obligations it still carries. The queue is therefore not an afterthought to the restart. It is one of the places where the bank demonstrates whether the recovery has reached the people who were waiting.