The balanced scorecard and strategy maps

Lessons 4 to 8 dealt with financial measures of divisional performance: ROI, residual income, EVA and the transfer prices that feed them. Financial measures are indispensable, but on their own they have well-known limitations. They report the results of past decisions rather than the drivers of future results, they can be improved in the short run by cutting investments in customers, people and processes, and they say little about whether a division is carrying out its strategy. The balanced scorecard, introduced by Robert Kaplan and David Norton in 1992, was a response to those limitations. This lesson explains the scorecard's four perspectives, the strategy map that links them, how measures are chosen, weighted and cascaded, and what research and practice have found. BUS-450 covers strategy formulation; here the focus is on translating a strategy into measures, and ACCT-421 takes strategic performance management further.

Cormorant's strategy

Cormorant's board has adopted a strategy for 2026 to 2030 built around the shift from gas water heaters to electric heat pumps. Appliances will grow heat pumps from 30 percent to half of its sales by 2030, competing on reliability and energy efficiency rather than price. Components will develop variable-speed motors and control boards that make Cormorant's heat pumps quieter and more efficient than competitors'. Services will build a network of certified technicians and a growing base of maintenance contracts, so that customers who buy a Cormorant heat pump stay with Cormorant for service. Head office wants a measurement system that shows, every quarter, whether each division is executing its part of the strategy, not only whether it earned its profit.

The four perspectives

The balanced scorecard organizes measures into four perspectives. The financial perspective asks how the business looks to shareholders: profitability, growth, return on capital and risk. The customer perspective asks how customers see the business: satisfaction, retention, acquisition, market share and the value proposition the strategy promises them. The internal business process perspective asks what the business must excel at to deliver that value proposition: operations, innovation, after-sales service and regulatory and social processes. The learning and growth perspective asks how the business will sustain its ability to change and improve: employee skills, information systems and organizational culture and alignment.

The word balanced refers to several balances: between financial and non-financial measures, between lagging indicators of outcomes and leading indicators of future performance, between internal and external perspectives, and between short-term results and long-term capability. A scorecard is not simply a list of key performance indicators. Its measures are chosen because they describe the strategy, and they are linked by hypothesized cause-and-effect relationships.

Strategy maps

In later work, Kaplan and Norton developed the strategy map, a diagram that shows the strategic objectives in each perspective and the cause-and-effect links between them. The map reads from the bottom up. Investments in people, systems and culture improve internal processes; better processes deliver the customer value proposition; satisfied, loyal customers produce revenue growth and productivity, which create shareholder value. Each objective on the map has one or more measures, a target and initiatives intended to achieve it. Drawing the map forces managers to make the strategy explicit and to argue about how they believe it works, which is often more valuable than the measures themselves.

For Appliances, the map hypothesizes that certified technicians and stronger engineering skills lead to higher first-pass yield, fewer warranty claims and faster product development; that these produce a reputation for reliability and efficiency among dealers and homeowners, shown in net promoter score and dealer retention; and that loyal dealers and a strong reputation lead to heat pump market share, sales growth and, eventually, higher ROI. Each link is a hypothesis that can be tested as data accumulate.

Scoring a scorecard

Many companies summarize scorecard performance in a single score, comparing each measure with its target, capping the credit for overachievement, and weighting the perspectives. The program below does this for Appliances' 2026 results, with equal perspective weights and a cap of 120 percent. For measures where lower is better, such as warranty claims, the ratio is inverted.

# Appliances 2026 balanced scorecard: achievement against target (illustrative)
measures = [  # perspective, measure, actual, target, higher_is_better
    ("Financial", "Return on investment (%)", 15.0, 14.0, True),
    ("Financial", "Sales growth (%)", 6.0, 8.0, True),
    ("Financial", "Heat pump share of sales (%)", 30.0, 35.0, True),
    ("Customer", "Net promoter score", 42, 45, True),
    ("Customer", "Dealer retention (%)", 92.0, 90.0, True),
    ("Customer", "Heat pump market share (%)", 11.0, 12.0, True),
    ("Internal process", "First-pass yield (%)", 96.5, 97.0, True),
    ("Internal process", "Warranty claims (% of units)", 2.1, 1.8, False),
    ("Internal process", "Time to market (months)", 14, 12, False),
    ("Learning and growth", "Training hours per employee", 32, 30, True),
    ("Learning and growth", "Engineer turnover (%)", 9.0, 8.0, False),
    ("Learning and growth", "Heat pump certified technicians (%)", 60.0, 70.0, True),
]
weights = {"Financial": 0.25, "Customer": 0.25, "Internal process": 0.25, "Learning and growth": 0.25}
cap = 1.20
scores = {}
for p, m, a, t, hib in measures:
    s = min(cap, a / t if hib else t / a)
    scores.setdefault(p, []).append(s)
    print(f"{p:<20} {m:<36} actual {a:>5}  target {t:>5}  score {s:6.1%}")
total = 0
for p, lst in scores.items():
    avg = sum(lst) / len(lst)
    total += weights[p] * avg
    print(f"{p:<20} average {avg:6.1%}")
print(f"Weighted scorecard result: {total:.1%}")

Financial            Return on investment (%)             actual  15.0  target  14.0  score 107.1%
Financial            Sales growth (%)                     actual   6.0  target   8.0  score  75.0%
Financial            Heat pump share of sales (%)         actual  30.0  target  35.0  score  85.7%
Customer             Net promoter score                   actual    42  target    45  score  93.3%
Customer             Dealer retention (%)                 actual  92.0  target  90.0  score 102.2%
Customer             Heat pump market share (%)           actual  11.0  target  12.0  score  91.7%
Internal process     First-pass yield (%)                 actual  96.5  target  97.0  score  99.5%
Internal process     Warranty claims (% of units)         actual   2.1  target   1.8  score  85.7%
Internal process     Time to market (months)              actual    14  target    12  score  85.7%
Learning and growth  Training hours per employee          actual    32  target    30  score 106.7%
Learning and growth  Engineer turnover (%)                actual   9.0  target   8.0  score  88.9%
Learning and growth  Heat pump certified technicians (%)  actual  60.0  target  70.0  score  85.7%
Financial            average  89.3%
Customer             average  95.7%
Internal process     average  90.3%
Learning and growth  average  93.8%
Weighted scorecard result: 92.3%

Appliances beat its ROI target, but its scorecard result is 92.3 percent, because it fell short on sales growth, heat pump share, time to market and the share of certified technicians. The pattern matters more than the overall number. Strong current profit combined with weak leading indicators in the heat pump strategy suggests the division may be harvesting its gas water heater business rather than building the new one. A purely financial evaluation would have praised the division's 15 percent ROI and missed the warning.

Weighted scores have drawbacks. They let strong performance on one measure offset weak performance on another, which may not reflect the strategy: no amount of training hours makes up for missing the heat pump share target. Weights are subjective, and managers quickly learn which measures carry the most weight and focus on them. Some companies therefore use thresholds, requiring a minimum on critical measures, or present the scorecard without a summary score and leave the overall judgment to senior management.

Testing the cause-and-effect links

A scorecard is only as good as the links it assumes. If dealer net promoter score really leads heat pump sales, then improving it is a sensible objective; if not, the division may be measuring the wrong thing. Testing the links with data is harder than it looks, because many business series trend together over time.

# Does dealer net promoter score lead heat pump sales? Quarterly, 2024 Q1 to 2026 Q4 (illustrative)
from statistics import correlation
nps   = [30, 31, 34, 33, 36, 38, 37, 40, 41, 40, 43, 42]
sales = [2.0, 2.1, 2.1, 2.3, 2.3, 2.4, 2.6, 2.6, 2.8, 3.0, 3.0, 3.2]   # heat pump sales, $ millions
print(f"Heat pump sales in 2026: {sum(sales[-4:]):.1f} million")
d_nps = [b - a for a, b in zip(nps, nps[1:])]
d_sales = [b - a for a, b in zip(sales, sales[1:])]
for lag in range(0, 3):
    lv = correlation(nps[:len(nps) - lag], sales[lag:])
    ch = correlation(d_nps[:len(d_nps) - lag], d_sales[lag:])
    print(f"Lag {lag}: levels {lv:5.2f}   quarter-to-quarter changes {ch:5.2f}")

Heat pump sales in 2026: 12.0 million
Lag 0: levels  0.92   quarter-to-quarter changes -0.94
Lag 1: levels  0.98   quarter-to-quarter changes  0.80
Lag 2: levels  0.94   quarter-to-quarter changes -0.23

Measured in levels, net promoter score and heat pump sales are highly correlated at every lag, because both rose steadily from 2024 to 2026. That correlation says almost nothing about cause. Measured as quarter-to-quarter changes, the picture is clearer: changes in net promoter score are strongly positively correlated with changes in sales one quarter later (0.80), but not in the same quarter or two quarters later. With only twelve quarters, the evidence is suggestive rather than conclusive, but it is consistent with the strategy map's hypothesis that dealer loyalty leads sales by about a quarter. Companies with more data use regression, control for seasonality and other drivers, and look across divisions or regions to test links more rigorously.

Cascading and alignment

Scorecards can be cascaded from the corporate level to divisions, departments and teams, so that each unit's measures support the objectives above it. Cormorant's corporate objective of leadership in heat pumps becomes, for Components, a time-to-market objective for variable-speed motors and a defect-rate objective for control boards; for Services, objectives for the number of certified technicians and the renewal rate on maintenance contracts. Cascading helps align a decentralized organization, but it must leave room for local judgment: a scorecard imposed rigidly from above undermines the autonomy that lesson 2 argued decentralization requires.

Kaplan and Norton later described the strategy-focused organization, in which the scorecard becomes a management system rather than a measurement tool: it is used to communicate strategy, link budgets and initiatives to strategic objectives, align incentives, and hold regular strategy reviews separate from operational reviews. Few companies implement all of this. Many use the scorecard as an expanded set of key performance indicators, which is useful but misses the point of tying measures to strategy.

Criticisms and evidence

The scorecard has attracted substantial criticism. Hanne Nørreklit argued in 2000 that the relationships between perspectives are often logical or definitional rather than causal, and that the time lags between them are rarely specified, so the cause-and-effect claims are hard to test. Others note that four perspectives may omit important stakeholders such as suppliers, regulators, communities and the environment, a gap lesson 10 addresses. Scorecards can grow to dozens of measures, diluting attention. And managers may game non-financial measures, which are often easier to manipulate than audited financial results.

Evidence on the scorecard's effects is mixed but informative. Research by Christopher Ittner and David Larcker found that companies that tested and validated the links between non-financial measures and financial results earned higher returns than those that did not, while many companies did not attempt such validation at all. Experimental studies have found that evaluators of scorecards tend to rely on measures common to all units and underweight unique measures tied to each unit's strategy, the common-measures bias, which undermines the scorecard's purpose. And the scorecard's use in compensation, discussed in lesson 11, raises questions about subjectivity and weighting.