A department hits its target. The number is green. The business is worse off.
This is not a failure of measurement. Each of the metrics below is legitimate, correctly calculated, and improving:
- Procurement lowers unit cost 6% by consolidating suppliers — and lead time rises from two weeks to five, so delivery holds more inventory and misses more dates
- Sales grows bookings 20% by discounting into a segment the delivery organization has never served, so realization falls and rework rises
- Operations raises utilization from 74% to 86%, removing the slack that used to absorb variation, so every unplanned absence now slips a deadline
- Finance cuts discretionary spend 15%, including the training that kept certification current, so a capability lapses eighteen months later
In every case the metric moved in the intended direction. The cost moved somewhere the metric could not see.
Why this is structural rather than careless
Metrics are assigned to functions because accountability requires ownership. Ownership requires a boundary. The boundary is what makes the metric measurable — and it is also what makes the displaced cost invisible.
So the failure mode is not that people gamed the number. It is that the system was designed to reward exactly what happened.
The question that catches it
Before accepting a local improvement, ask one thing:
What cost, delay, workload, or risk moved somewhere else?
Most of the time the honest answer is "we don't know," and that answer is useful. It identifies precisely where the enterprise view is missing.
What connected measurement looks like
You do not need to measure everything. You need to know, for each important metric, which other metric it trades against — and to look at the pair.
Illustrative: utilization trades against delivery predictability. Unit cost trades against lead time. Bookings growth trades against realization. Cost reduction trades against capability retention.
A management view that shows utilization alone invites a bad decision. A view that shows utilization beside on-time delivery and rework makes the trade-off visible, and lets a leader decide it deliberately rather than discover it two quarters later.
That pairing is not a technical problem. It requires knowing how the business actually works — which functions depend on which, and where performance propagates. That model usually does not exist in written form anywhere in the organization.
Why this survives management review
The displaced cost is usually invisible to the review that would have caught it, for a specific reason: it lands in a different reporting period from the decision.
Procurement's saving is immediate and appears next month. The inventory cost and the missed delivery dates accumulate over the following two quarters, by which point they read as an operations problem. The two events are never placed side by side, because no report contains both, and the person who would connect them does not exist.
This is why the fix is not more discipline in the review. It is a view that contains both numbers.
What to do with an existing metric set
You do not need to redesign the scorecard. For each metric that carries real behavioral weight — the ones people are measured on — write down two things:
- What it trades against. The metric that moves in the opposite direction when this one is pushed hard.
- Where that trade-off would show up, and after how long.
That second column is what turns an abstract concern into something checkable. If procurement's saving should show up as lead time within one quarter, then lead time belongs in the same review, on the same page, from the start.
Most firms find that five or six metrics carry nearly all the behavioral weight, and pairing those is a short exercise with a disproportionate effect.
Building the model of how those functions actually interact is the first phase of a 360° Enterprise Intelligence Blueprint.