What the first-order model typically captures
Direct salary and benefits cost removed, against the licence or infrastructure cost of the replacing automation, projected over a budget cycle — a straightforward, defensible calculation on its own terms, and the reason these proposals clear board approval relatively easily.
The gap isn't that this calculation is wrong; it's that it's treated as the complete picture rather than one input to it.
What the second-order model needs to include
The cost of rehiring and retraining if the automation underdelivers against its business case, the value of institutional knowledge that leaves with the reduced headcount and the cost of reconstructing it later, and the effect on remaining staff morale, retention, and productivity in the surrounding team, all have real, estimable costs — rarely modelled, not because they're unmeasurable, but because they're uncomfortable to put a number next to.
A board paper that includes a worst-case second-order scenario alongside the first-order saving gives genuinely informed consent to the decision, rather than an incomplete one dressed up as thorough.
A board approving a workforce reduction on a first-order cost model alone isn't approving a complete business case. It's approving half of one.MNB Strategy Practice
Key actions
- Model rehiring/retraining cost if the automation underdelivers
- Estimate the value of institutional knowledge leaving with the reduction
- Account for remaining-staff morale and retention effects explicitly
- Present a worst-case second-order scenario alongside the first-order saving