NAB has published the job number. It still owes everyone the mechanism

A number with a date on it

A headcount target with a deadline is a promise about costs. It is not a measurement of productivity. National Australia Bank has made the first and let everyone infer the second.

The bank has said 2,500 roles will go by June, 1,500 of them before Christmas, concentrated in corporate functions. That is roughly six per cent of the workforce. Staff told the ABC that entire teams are being made vacant and asked to compete for a smaller number of seats — “spill and fill” — with two-week consultation periods and final days of 2 or 17 December. The chief executive, Andrew Irvine, told staff the changes are “expected to improve outcomes for customers” and encouraged people to check on one another.

The first half of that sentence is the part that should be tested. The second is manners.

What the number does not say

There are three ways a bank gets to 2,500 fewer people, and they are not interchangeable.

The first is automation: the work genuinely stops being done by anyone. That saving is permanent, and the risk moves into systems and models that need owners, monitoring and someone who understands what they do when they fail.

The second is relocation. The work still exists and is still done by people, just under a different employment regime, often through a captive or a vendor. The saving lasts exactly as long as the wage gap does, and it brings handover risk, concentration risk and a longer distance between the person doing the work and the person accountable for it.

The third is doing less. That is not a saving. It is a transfer to customers, settled later in hold times, processing errors and remediation.

All three land on the same line in the accounts. Only one of them is efficiency. Staff who spoke to the ABC described both an AI push and work moving offshore — the bank itself has not, on the public record, apportioned the number between them. Until it does, “improve outcomes for customers” is an assertion with nothing underneath it.

The deadline is the tell

If the driver were automation, the sequence would be tooling first, headcount second. You would know the work had gone because it had stopped arriving.

A December date does not work that way. Capability arrives when it arrives; 17 December arrives on 17 December. Where the two do not meet, the gap is closed by whoever is still sitting there, or by contractors in February at a higher daily rate than the person who left. That is the pattern worth watching in the half-year numbers, and it is precisely the pattern an aggregate expense figure hides.

Spill and fill does not test for the thing banks need

One worker quoted in the reporting has spent a decade in NAB’s technology team and now has a fortnight to compete against colleagues for a reduced number of roles. A process like that selects for people who interview well under duress. It does not select for the person who knows why a particular overnight batch behaves oddly at the end of the month, because that knowledge is usually undocumented and lives in a head rather than a runbook.

A bank is a licensed operator of critical infrastructure, supervised by a prudential regulator precisely because its failures are other people’s failures. A restructure of technology teams is an operational risk event, not merely a human resources one. Someone should be able to say which systems lost their last experienced owner this quarter, and whether anyone checked before the consultation window opened.

The strongest case the other way

Banks are not employment schemes. Corporate layers that have accumulated over a decade are paid for by customers, through margins and fees, whether or not those customers ever benefit. Announcing a single number with a timetable is more honest than a slow quarterly drip that leaves everyone guessing for three years. A reapplication process, for all its cruelty, at least applies a stated test rather than a tap on the shoulder. And deferring the announcement to February would not spare anyone the uncertainty — it would extend it.

Accept all of that. None of it is an argument for cutting without publishing the basis. Efficiency claimed in advance of evidence is cost reduction with a story attached, and the group with no way to check the story is the customer base that is supposed to benefit from it.

The room this lands in

These are corporate roles in a tight-ish market, not a national tragedy; the same week brought reports of butcher shops closing because they cannot find qualified staff. But the same week also brought a Foodbank report counting 100,000 more households short of food than a year ago, most of them hitting that wall for the first time. Fifteen hundred people being told in a fortnight that they may be unemployed by 17 December is a decision taken inside that economy, not alongside it. It does not make the cuts wrong. It does raise the standard of explanation owed.