Forty-Two KPIs and No Plan

Symptoms: flat wait time, falling engagement, turnover, conflicting measures. Mechanisms:

1. Translation without negotiation. Forty-two KPIs were pushed down; no one negotiated up. Targets arrived as assignments, so departments complied with the reporting and not with the intent. That is why data entry doubled while wait time did not move.
2. No prioritization. With 42 equally weighted measures, everything is a priority and nothing is. Effort spread thinly across 60 scorecards produces no breakthrough anywhere.
3. Internally contradictory measures. Volume rewards seeing more patients; cost per visit rewards spending less per encounter. Each department optimized locally and the system lost - classic suboptimization. Nobody owned the interaction.
4. Lagging indicators only. Wait time and cost per visit are results. There were no measures of the drivers (room turnover, provider scheduling templates, registration cycle time), so no one could tell what to change this month.
5. No deployment of resources. The budget ignored the plan, which told every manager the plan was decorative.
6. No review cadence with authority to change course. Monthly scorecards are reporting; deployment requires a forum that reallocates people and money.

The wait-time goal failed not because the measure was wrong but because nothing in the system connected a clinic's daily work to it.

Architecture (hoshin-style, adapted):

• 3-4 breakthrough objectives for the year, no more. Wait time and cost per visit can both survive, but as one system objective ('reliable, affordable access') with two result measures and a stated trade-off rule.
• Catchball, not cascade. Each clinic's leadership negotiates its contribution upward in two rounds. The output is a small matrix linking each breakthrough objective to annual targets, improvement initiatives, and named owners - typically 8-12 initiatives total, not 60.
• Two measures per objective: one lagging result (wait time), two or three leading drivers (exam room turnover minutes, third-next-available appointment, registration cycle time). Leading measures are what staff can actually move.
• One owner per objective, with authority over cross-department trade-offs. Without that, the volume/cost conflict has no referee.
• Monthly operating review (90 minutes): owner reports on drivers, not on the result, and the room makes resourcing decisions. Quarterly, the executive team reallocates budget against the same objectives.
• Budget linkage: no initiative appears on the matrix unless it appears in the budget, and vice versa.
• Retire measures. Explicitly kill the 42-KPI scorecards; leaving them alive re-creates the old system underneath the new one.

Expect resistance from corporate planning - the 42 KPIs are their product. Give them a new product: the matrix and the review.

Reframe: this is not a measurement problem, it is a design problem. Two departments were given measures that are individually rational and jointly destructive, so the conflict is in the scorecard, not in the people.

Resolution steps:

1. Name the suboptimization explicitly in the operating review, with the arithmetic: if a clinic adds visits to fill capacity, cost per visit falls, but if those visits are unscheduled walk-ins, wait time rises and so does cost per completed visit. Both departments were right about their own number.
2. Replace the pair with a system measure. Cost per completed visit with an access standard attached, or visits per clinical hour at the wait-time target. A measure that cannot be improved by damaging the other objective is the goal.
3. Pair measures rather than merge them. Where two objectives genuinely trade off, report them together in the same review with a rule: neither may improve by more than X% while the other degrades. That forces the conversation into the room instead of into the scorecard.
4. Change incentives if any exist. If a manager's bonus rides on cost per visit alone, no amount of reframing will work.
5. Watch for the third variable: some of the wait-time problem is likely provider scheduling templates and room turnover, which neither KPI touches. Check the drivers before assuming capacity is the answer.

Judgement point: the exam-relevant move is refusing to pick a winner between two locally sensible metrics and instead redefining the level at which the measure is taken.