DRAFT — structure and argument are ours, but this is published under a byline and should not go live until the named author has read and approved it.
A familiar pattern: a cost program lands, the run rate falls, the board is satisfied, and within three years the cost base has returned to roughly where it started. The savings were not imaginary. Something else happened.
Removing work is not the same as removing demand
Most programs remove capacity — roles, contractors, a layer of management. Far fewer remove the demand that made the capacity necessary: the exception handling, the manual reconciliation, the reporting that exists because someone once asked for it and nobody has asked since.
When capacity is cut and demand is not, the work does not disappear. It queues, degrades, or reappears somewhere less visible. The cost returns because the underlying need never went away.
Three tests before committing
- What generates this work? If the answer is a process defect or a product
design choice, the durable saving is upstream of the team being cut.
- Who will notice if it stops? Work nobody can name a consumer for is
usually the safest to remove — and is rarely where programs start.
- What happens at the next peak? A cost base sized for the median month
will be rebuilt, informally and expensively, at the first stress event.
The uncomfortable version
Structural cost reduction usually requires deciding not to do something — exiting a product, accepting a service level, standardising something that is currently bespoke for good historical reasons. Those are executive decisions, not program decisions, and programs that avoid them produce savings with a three-year half-life.