Balanced Scorecard

The executive dashboard has forty-seven KPIs and most of them are green while the company misses its year. Ask which measure is supposed to drive which and the room goes quiet. The dashboard is a list, not an argument.

Kaplan and Norton's insight in the early 1990s was that financial measures are lagging indicators — they tell you what already happened — so a scorecard should state the causal hypothesis instead: what you build in learning and growth drives internal process performance, which drives what customers experience, which drives the financials. The strategy map is that hypothesis drawn out as arrows, and a scorecard is worth having only if the hypothesis is testable, which requires measures scarce enough that a broken link actually shows.

The strategy literature supplies the discipline the scorecard needs to avoid becoming what it replaced. Joan Magretta, distilling Michael Porter, is precise about the endpoint: superior profitability has exactly two sources — relative price or relative cost — and both trace back to a distinct set of choices and the fit among them. A scorecard whose measures cannot trace to those choices is measuring activity, not strategy. Rumelt is blunter: a list of performance goals is not a strategy, and a scorecard assembled without a diagnosis reproduces the KPI wall with better graphics. Leinwand and Mainardi's Strategy That Works adds the coherence argument — companies close the strategy-execution gap through a few mutually reinforcing capabilities — which is the case for measure scarcity: measure the handful of things your identity actually depends on, not everything that moves.

The honest limit is that the causal links in most strategy maps are asserted, not demonstrated. Kaplan and Norton themselves framed the scorecard as a hypothesis to be tested against results; in practice the arrows rarely get revisited, and the moment executive pay attaches to a measure, the measure starts being managed. Both failure modes are design inputs, not footnotes.

The service builds the map with the causal links explicit and no orphan objectives, enforces the two-measures-per-objective ceiling in code rather than by discipline you must supply, dispositions your existing metrics keep-or-retire, and tells you honestly which measures are robust enough to attach executive pay to — and which would be gamed.