Four perspectives, one causal chain
Robert Kaplan and David Norton introduced the scorecard in the early 1990s, and its four perspectives are designed as a chain, not a list: capable people and systems (learning and growth) improve the processes that matter (internal), which deliver what customers value (customer), which produces the financial result (financial). A scorecard whose measures do not connect in that causal logic is a dashboard, not a strategy tool.
| Perspective |
The question it answers |
Illustrative measures |
| Financial |
What must we deliver to owners for the strategy to count as working? |
Revenue mix shift, margin by segment, return on capital |
| Customer |
What must customers experience for those financials to happen? |
Retention in target segments, share of wallet, service-level attainment |
| Internal processes |
Which processes must excel to deliver that customer experience? |
Cycle time on the critical path, first-time-right rates, cost per transaction |
| Learning and growth |
What capability, data and culture must we build to sustain it? |
Critical-role coverage, capability build-out against plan, systems availability |
When a scorecard justifies its overhead
- When strategy has genuinely changed and legacy KPIs still reward the old model. The scorecard is the mechanism that makes the new strategy operational.
- When the board sees healthy financials but leadership suspects the franchise is eroding underneath: lead indicators are the early-warning system.
- When business units each optimise their own numbers and the strategy needs them to trade off in a common direction.
A brief illustration of the causal test
A business-services firm shifts strategy from transactional work to multi-year partnerships. Its old scorecard (utilisation, monthly revenue, new logos) actively rewards the old model. The rebuilt scorecard follows the causal chain: learning and growth measures account-management capability; internal measures the proportion of delivery running on standardised platforms; customer measures multi-year retention and expansion in named accounts; financial measures the revenue mix shift towards contracted recurring work. Every measure earns its place by answering one question: if this improves, do we believe the next perspective improves? Any measure without a defensible answer is deleted, however available its data.
How scorecards go wrong
The failure modes are well-worn, and most of them are visible on the scorecard itself.
- KPI sprawl: forty measures across the four boxes, which is not balance but abdication. A working scorecard holds a handful of measures per perspective, each tied to the strategy's specific bets.
- Measures chosen because the data exists, not because the causal chain needs them. The scorecard then faithfully reports things that do not matter.
- Targets negotiated to be achievable, so the scorecard glows green while market share slides. A scorecard on which nothing is ever red is measuring the negotiation skill of managers, not the strategy.
- The scorecard is decoupled from resource allocation: budgets and bonuses still follow the old financial logic, and everyone learns which set of numbers actually counts.
- Lead indicators get replaced by lags everywhere, recreating the exact blindness the framework was designed to remove.
Independent challenge on the causal claims
Every scorecard encodes hypotheses (improve this measure and that outcome follows), and those hypotheses are set by the same leadership whose strategy they flatter. Operators who have run comparable businesses can challenge the links themselves: whether service levels genuinely drive retention in this market, whether the capability measures actually predict delivery, which green metric is masking decline. Pressure-testing the causal chain is worth doing before a year of management attention is steered by it.
