Balanced Scorecard

Goal

The Balanced Scorecard is an instrument of strategic management: for each of the four perspectives it sets out what the strategy demands, how that is measured, what value is aimed at and which initiative gets there. Kaplan and Norton formulated it in 1992 in answer to a weakness of steering by the accounts: a financial measure reports on decisions already taken. A financial year can close on a margin held, with a customer base eroding, skills ageing and production equipment falling behind.

The technique supports allocation decisions: which objectives go into the annual plan, which initiatives to fund, which to stop. An initiative that moves no measure on the scorecard consumes a budget without serving the stated strategy. The Balanced Scorecard makes that trade-off possible by holding the four perspectives on one page, against one deadline.

The deliverable is the scorecard itself: per perspective, a set of rows with four columns. Alongside it come the strategy map, the assignment of every measure to a named owner and the review cadence. The measures on the scorecard are key performance indicators, each tied to a strategic objective.

Usage

When to use it

• A strategy stated as intentions: turn it into quantified objectives before committing the initiative portfolio.
• Steering run off the income statement alone: add the measures that announce the result before it lands.
• An initiative portfolio to arbitrate: tie every project to a measure and a target so the budget decision can be taken.
• A solution evaluation after deployment: judge the value delivered across all four perspectives together.
• Steering at several levels: break company objectives down into department objectives that stay consistent with one another.

When not to use it

• No settled strategy: frame the business model first with a business model canvas.
• A need for an external reference: benchmarking gives the gap to the market that the scorecard does not compute.
• A single team over one quarter: the cadence and the cascade cost more than they return, a set of operational indicators is enough.

Description

The four perspectives

The Financial perspective states what the strategy has to produce for the owners of the business: profitability, revenue growth, economic value added, return on capital employed. It is where the reasoning arrives in a commercial company. In a public administration or a non-profit it reads as budget discipline and the cost of the service delivered. The beneficiary perspective often rises into its place.

The Customer perspective asks how the organisation must appear to its customers if it is to reach its vision: satisfaction, retention, share of the targeted segments, perceived value of the offer. This is where the commercial promise is measured; next year's revenue depends on it.

The Internal Business Process perspective names the processes the organisation has to excel at to hold that promise and the financial result: production quality, cycle time, defect rate, time to market, adherence to service commitments.

The Learning and Growth perspective asks what has to be built to stay able to change and improve: employee skills, information systems, culture, capacity to innovate. Its measures are the easiest to sacrifice when times are tight, because their effect shows up with the longest delay.

The word perspective comes from Kaplan and Norton and practice has kept it; the BABOK Guide speaks of dimensions. The four groups are the same.

The row, the analyst's unit of work

A perspective with no row filled in stays a heading; the row carries four columns.

• Objective: the result the strategy demands in this perspective, written as a state reached, "deliver on time" or "master five-axis machining".
• Measure: the indicator that makes the objective observable, with its calculation formula, its data source and its reading frequency.
• Target: the value to reach, its deadline and the starting value it is read against.
• Initiative: the programme, project or action charged with moving the measure towards the target, with its budget and its owner.

An objective without a measure is a declaration; a measure without a target cannot be read; a target without an initiative waits for progress to arrive on its own. The initiative column ties the scorecard to the budget: this is where the strategy consumes its resources. Kaplan and Norton put the useful size of a scorecard at around twenty measures, twenty-five at most, all perspectives taken together.

The analyst's work lies mostly in turning an objective into a measure. A management team that wants to "become the most reliable supplier in its segment" has stated an intention; the analyst makes that intention observable by proposing the on-time delivery rate, writing its calculation definition, establishing that the data exists in the ERP and negotiating the target with the sponsor. That negotiation brings out the disagreements the statement of intent was hiding.

The cause-and-effect chain

Investment in learning and growth improves the performance of the internal processes; better processes produce the customer results, satisfaction and retention; those customer results produce the revenue and the margin of the Financial perspective. The BABOK Guide notes that each perspective acts on the others and is acted on by them; Kaplan and Norton name the mechanism and give it a direction of reading.

They formalised that reading under the name strategy map: the objectives of the four perspectives are laid out in stacked bands and joined by arrows saying which objective serves which. It makes the hypothesis arguable in committee, because an arrow invites challenge where a list of measures is accepted in silence. It also gives every objective an admission test: an objective whose path to a financial result nobody can trace leaves the scorecard.

Lagging and leading indicators

A lagging indicator reports on an action already completed: the quarter's revenue, the margin, the score of the annual satisfaction survey. A leading indicator gives early word of the performance to come: training hours delivered, drift in cycle time, the order book, the number of products in development. Objectives attract lagging indicators by default, since accounting and surveys already produce them.

Pairing the two is what puts the scorecard ahead of the accounts. When the leading measure moves and the lagging measure does not follow within the expected delay, the causal hypothesis is wrong or incomplete. That is the signal calling the strategy itself into question, where most scorecard variances call execution into question.

The review cadence

Four processes that Kaplan and Norton describe in 1996 turn the scorecard into a management system.

1. Translating the vision: convert the strategy into objectives per perspective, down to the level of precision where the management team agrees on what the words mean.
2. Communicating and linking: bring company objectives down to department objectives, then, once the scorecard has settled, to individual objectives.
3. Planning and allocating: set the targets, select the initiatives that serve them and assign them resources in the annual budget.
4. Feedback and learning: compare the results against the targets, then reopen the causal hypothesis when the numbers contradict it.

The review frequency is tiered: operational measures are read monthly or weekly, the strategic review is held quarterly and the map is revised yearly or as soon as a pair of indicators contradicts itself. A scorecard reviewed once a year works as a poster.

The pitfalls

The first is the accumulation of measures. A scorecard carrying thirty or forty of them scatters the attention of a committee that has to read it in an hour, and the causal chain disappears into the noise. The discipline comes down to one rule: remove a measure whenever you add one.

The second is the scorecard that displays four perspectives while the organisation goes on deciding off the income statement. The three non-financial perspectives then receive measures that are easy to collect and detached from any objective, and the budget trade-offs are made as before. The symptom is legible in the committee minutes, in the time spent on each perspective.

The third is the measure nobody owns. A measure with no named owner and no attached initiative appears in the quarterly report without anything following from it. The initiative column exists for that. Abridged scorecards drop it first.

The fourth is gaming the measure. A measure is an approximation of the objective; once it becomes the target of a bonus, it gets optimised for its own sake: a satisfaction score raised by surveying only the easiest customers, an on-time delivery rate held by redefining what on time means. The safeguard lies in the written definition of every measure and in pairing it with a second measure that the same manoeuvre degrades.

The fifth is cascading too early. Taking the objectives down to individual objectives and variable pay before the causal hypothesis has been checked over one or two cycles freezes badly chosen measures and makes them politically expensive to correct: an employee whose bonus depends on a measure defends that measure.

AI considerations

The first use that pays is generating candidate measures. From a stated objective and the industry, an assistant proposes a dozen plausible indicators per perspective, with their usual calculation formula, which the analyst then filters. The second is matching them against the available data: putting a list of measures next to the data dictionary of the ERP and the warehouse to say which already compute, which need development and which rest on data nobody collects. That is inventory work, tedious by hand. The third is testing the causal chain statistically against history: measuring the lagged correlation between a leading indicator and the lagging indicator it is supposed to announce, which turns a committee hypothesis into a verifiable statement. The fourth is drafting the quarterly review commentary from the variances between values and targets.

Three points escape the tool. The causal hypothesis first: a model will readily produce a plausible chain across four perspectives, and a correlation over eight quarters of data demonstrates no causality. Deciding that training on five-axis machines will bring defects down is a strategic argument the management owns. The weighting between perspectives next: choosing to hold the margin at the price of satisfaction for two quarters is a political trade-off with no analytical answer. The sensitivity of the data last: the cascade reaches down to individual measures and variable pay, so the scorecard's data sets contain personal data within the meaning of the revised Federal Act on Data Protection (nFADP); they are not deposited with an external service to have a trend analysed.

Examples

A Vaud manufacturer of precision components for the medtech sector, 185 employees and CHF 42'000'000 in revenue, settles one row per perspective for the financial year: four complete rows, each linking an objective to a dated target and to the initiative charged with reaching it.

The customer row shows what the translation costs. The sponsor asked for 98% on-time deliveries. The ERP extraction showed that the existing measure dated the delivery to the date the plant had confirmed, where the customer counts from the date it had requested: on that basis the starting point fell from 94 to 89%. The target settled at 96% on the corrected definition, and that definition went into the measures dictionary together with the query that produces it. It took two management meetings.

The four rows form a single chain, read from the bottom up: training in five-axis machining should bring the defect rate down, a production line reworking fewer parts should hold the delivery date and delivery reliability should sustain the price that carries the EBIT margin. The strategy map lays that reading out flat, in a form the committee either challenges or endorses.

Visualisations

The scorecard belongs in a live grid, whose rows reorder when a measure changes and whose target column stands out to the eye. A screenshot pasted into a presentation freezes numbers that the smallest revision of a target makes wrong.

Two conventions make the reading immediate. The target carries its starting value, without which a percentage says nothing about the effort being asked for. The initiative stays on the same row as the objective it serves: splitting them into two tables is the commonest way to lose the link between strategy and budget.

The structure of the technique, by contrast, does lend itself to a drawing, because it is made of positions and links. A "vision and strategy" core surrounded by the four perspectives, each carrying its guiding question and the strip of four columns, shows in one stroke that there are four of them and that they all have the same anatomy. The strategy map calls for the same treatment: four stacked bands and arrows climbing from one objective to the next carry the causal order, which a table cannot render.

Cost

Tooling

A spreadsheet is enough for the first two cycles: the rows get written there, the targets get compared against the readings and the structure gets reworked while the measures are still moving. Committing a tool before the scorecard is stable amounts to industrialising measures that are about to change.

Once the measures are settled, business intelligence tooling plugged into the ERP and the data warehouse automates the collection, the history and the reading frequency. Power BI, Tableau and Qlik cover the need, provided somebody keeps the calculation definitions alongside the dashboards produced. The risk at this stage is the proliferation of views: the management scorecard stays a single screen, whatever the number of reports the tool can produce.

Software dedicated to strategic management carries the objective-measure-target-initiative structure natively, together with the strategy map and the cascade by unit, a review cycle and named assignment of the measures. ClearPoint Strategy, Corporater and Spider Impact belong to this family. They earn their keep from the moment the cascade runs down several levels of the organisation, where tracking versions by hand becomes expensive.

OKR tools such as Workboard or Perdoo carry a neighbouring structure, an objective and measurable results, on a quarterly cadence and a logic of team commitment. They suit the operational breakdown; the strategy map and the pairing of leading with lagging indicators are absent from them, so they complement the Balanced Scorecard without replacing it. Whatever the tool, the measures dictionary lives in a shared and versioned document: it is what gets opened when two departments report two figures for the same measure.

Sources

• Kaplan, R. S. & Norton, D. P. (1992), The Balanced Scorecard, Measures That Drive Performance, Harvard Business Review 70(1): the original article, which introduces the four perspectives and their guiding questions.
• Kaplan, R. S. & Norton, D. P. (1996), The Balanced Scorecard: Translating Strategy into Action, Harvard Business School Press, ISBN 978-0-87584-651-4: the full treatment, where the four-column row is formalised and where the measurement framework becomes a management system.
• Kaplan, R. S. & Norton, D. P. (1996), Using the Balanced Scorecard as a Strategic Management System, Harvard Business Review 74(1): the four management processes (translating the vision, communicating and linking, planning and allocating, feedback and learning) as well as the review cadence.
• Kaplan, R. S. & Norton, D. P. (2004), Strategy Maps: Converting Intangible Assets into Tangible Outcomes, Harvard Business School Press, ISBN 978-1-59139-134-0: the strategy map and the explicit cause-and-effect chain between learning and growth, internal processes, customer and financial result.
• IIBA, A Guide to the Business Analysis Body of Knowledge (BABOK Guide) v3, §10.3 Balanced Scorecard: the framing of the technique for business analysis, its use at enterprise, department and project level, the vocabulary of leading and lagging indicators, as well as the recognised strengths and limitations.