The pattern is consistent enough to be predictable. A cost programme delivers its target in year one. Reporting shows the savings landing. Eighteen months later the cost base is back within a few per cent of where it started, and nobody can point to the decision that put it there.
What happened in between is almost always the same thing: the roles were removed but the work was not. The tasks that the departed roles performed did not disappear. They were absorbed by people who were already fully occupied, performed less well for a period, generated complaints or errors, and eventually justified a backfill under a different job title.
The test to apply before every cut
For each reduction, answer one question in writing: which specific activity stops, and who has agreed it can stop? If the answer names a report, a review step, a reconciliation or a service level that is being deliberately retired, the saving is likely to hold. If the answer is that the remaining team will absorb it, treat the saving as temporary and plan accordingly.
- Retire the outputs first, then the roles that produce them, in that order.
- Get the consumer of each report or service to confirm they can live without it. Producers are poor judges of their own output's value.
- Watch overtime, contractor spend and professional fees in the two quarters after the cut. This is where absorbed work reappears first.
Track the cost base, not the savings
Savings trackers are the wrong instrument. They measure activity against a plan and tend to show green long after the underlying cost has drifted. The honest measure is the total cost of the function, by quarter, against the pre-programme baseline, with contractors and capitalised labour included. It is a less flattering chart and a far more reliable one.
If the only place the saving is visible is the savings tracker, it is not a saving.
Protect a small number of things
The programmes that rebound least are the ones that named, at the outset, three or four capabilities that would not be touched regardless of pressure, typically the ones tied directly to revenue or to regulatory standing. Making that list explicit stops the salami-slicing that damages those areas quietly, and it makes the rest of the exercise easier to defend internally.