Most companies have KPIs. Few have KPIs that change behaviour. The typical mid-market organisation tracks dozens of indicators, yet when pressed, neither the finance team nor the leadership can explain which three numbers they would check first if revenue dropped 15% tomorrow. The problem is not a lack of measurement. It is a lack of design discipline. KPIs are chosen based on what data happens to be available, not what decisions need support. This article sets out the principles that distinguish indicators which sit in a spreadsheet from indicators which prompt action.
Why KPI design matters more than KPI quantity
The instinct to measure more is understandable. More data feels like more control. But research from BCG and MIT found that 60% of managers believe they need better KPIs — not more KPIs, better ones. The word “better” points to a design gap, not a data gap.
Poorly designed KPIs create three specific costs:
Attention waste. Every indicator on a dashboard competes for executive time. When fifty metrics sit side by side with no hierarchy, none receives the scrutiny it needs. The result is a monthly review where participants skim everything and interrogate nothing.
False confidence. A green traffic light next to a lagging indicator can mask a deteriorating pipeline. If the only KPIs tracked are financial outcomes — revenue, margin, cash — the organisation sees problems only after they have already hit the P&L. Data from the Institute of Global Controlling in the Czech Republic shows that only 11% of companies manage controlling via process KPIs. The remaining 89% rely exclusively on financial lagging indicators and miss the early-warning signals that enable course correction.
Eroded trust in reporting. When KPIs fail to predict or explain, decision-makers stop trusting the numbers. They revert to gut instinct and ad hoc requests, which in turn overloads the finance team with one-off analyses. The 52 reports per week reaching executive inboxes (Verret/LinkedIn) is a symptom of this cycle.
The KPI Institute reports that 68% of organisations see positive performance improvements after introducing structured KPI tracking. The operative word is “structured” — not “extensive.”
What makes a KPI effective
A KPI framework is only as good as the individual indicators within it. An effective KPI satisfies five conditions:
| Condition | Test question |
|---|---|
| Decision-linked | Which specific decision does this KPI inform? |
| Precisely defined | Can two people calculate it independently and get the same number? |
| Owned | Who is accountable for performance against this KPI? |
| Target-bound | What is the threshold that triggers action? |
| Time-cadenced | How often is it reviewed, and by whom? |
If a metric fails any of these tests, it remains a useful measure but does not earn KPI status. The distinction matters: metrics monitor, KPIs manage.
Start with the decision, not the data
The most common KPI design failure in mid-market organisations is working backwards from available data. The finance team exports what the ERP provides, builds a dashboard around those fields, and calls the result “KPI reporting.” The indicators are real, the numbers are accurate, but nobody changes their behaviour based on what they see.
Effective design reverses the sequence. It begins with a business question — “Are we acquiring customers profitably?” — and works forward through the chain: strategy, then performance drivers , then metrics, then thresholds, then response actions, then review cadence. Skipping to metrics without first identifying the driver produces vanity indicators that describe but do not direct.
Define the calculation with zero ambiguity
Consider “revenue per employee.” Simple enough — until you ask whether it includes contractors, whether revenue means booked or recognised, and whether part-time employees count as full-time equivalents. Every KPI needs a written definition that specifies numerator, denominator, data source, inclusion and exclusion rules, and currency treatment. Without this, the same KPI will produce different numbers from different teams, and the monthly review becomes a debate about methodology rather than performance.
Assign clear ownership
A KPI without an owner is a number without consequence. Ownership means a named individual (not a department) who is accountable for understanding why the KPI moved, proposing corrective action when it breaches a threshold, and reporting on progress at the agreed cadence. Ownership does not mean the person controls every variable — it means they are responsible for interpreting the signal and coordinating the response.
Set meaningful targets — not arbitrary ones
Targets based on “10% improvement over last year” are common and usually meaningless. They assume last year was a reasonable baseline, that external conditions are similar, and that 10% is both achievable and sufficient. Meaningful targets derive from one of three sources: strategic requirements (the growth plan demands a specific gross margin), benchmarks (industry data on comparable companies), or capacity analysis (the operations team can physically handle a defined throughput). If none of these sources supports the target, it is a guess dressed as a goal.
Balance leading and lagging indicators
Lagging indicators confirm what has already happened: revenue, profit, cash position. Leading indicators predict what is about to happen: pipeline value, order backlog, customer churn signals, production reject rates.
Most organisations default to lagging indicators because they are easier to define and harder to argue with — the revenue number is the revenue number. But a dashboard composed entirely of lagging indicators is a rear-view mirror. By the time a lagging indicator deteriorates, the window for corrective action has already narrowed.
The IGC finding that only 11% of companies use process KPIs illustrates the scale of this imbalance. For every lagging indicator in a KPI set, there should be at least one leading indicator that provides an early signal. A revenue KPI paired with a pipeline coverage ratio. A margin KPI paired with a procurement cost trend. A cash flow KPI paired with a days-sales-outstanding trajectory.
The “revenue per X” test
One useful design heuristic is the “revenue per X” pattern: revenue per employee, revenue per customer, revenue per square metre, revenue per product line. These ratios are simple enough to explain in one sentence, directly tied to a lever the business can pull, and immediately comparable across periods.
The pattern illustrates three properties of well-designed KPIs:
- Simplicity. If a KPI cannot be explained to a non-financial manager in fifteen seconds, it is too complex for executive use.
- Lever connection. Revenue per employee changes when you hire, when you lose customers, or when you raise prices. Each scenario implies a different action. The KPI itself prompts the question “why did it move?” — and the answer points to a specific lever.
- Comparability. The ratio works across business units, across time periods, and against external benchmarks. It does not require context-dependent interpretation.
Not every KPI needs to follow this pattern, but every KPI should pass the same tests: can you explain it simply, does movement imply a specific action, and can you compare it meaningfully?
Common design mistakes
Designing around available data rather than decisions
This is the dominant mid-market pattern. The ERP exports certain fields, the BI layer visualises them, and the result gets labelled “KPI dashboard.” The indicators are real but often irrelevant to the decisions the leadership team actually faces. Design must start with the question “what do we need to decide?” — not “what data do we have?”
Creating too many KPIs and losing focus
The instinct to measure everything is understandable but counterproductive. Organisations that track fifty indicators with no hierarchy cannot distinguish signal from noise. Effective KPI design means ruthless prioritisation: seven to twelve KPIs at executive level, with supporting metrics available for drill-down but not competing for attention on the primary view.
Setting targets without baseline understanding
A target set without understanding historical performance, seasonality, and external drivers is an aspiration without a foundation. Before setting any target, establish at least twelve months of baseline data and identify the factors that caused variation. The target should reflect what is achievable given those factors, not what the board wishes were true.
Measuring activity instead of outcomes
“Number of reports produced” is an activity metric. “Percentage of decisions made within SLA” is an outcome metric. The distinction matters because activity metrics reward effort regardless of result, while outcome metrics reward impact. KPIs should measure what changed in the business, not how busy the team was.
Ignoring leading indicators entirely
As noted above, the 11% figure from IGC confirms that this is not an edge case — it is the norm. Organisations that track only financial lagging indicators are structurally unable to anticipate problems. Adding even two or three leading indicators to an existing KPI set can materially improve the organisation’s ability to respond before results deteriorate.
How to approach the design process
A practical sequence for designing KPIs:
Identify three to five strategic priorities the organisation must advance in the coming period. These are not financial targets — they are strategic choices (e.g. “grow recurring revenue share,” “reduce customer acquisition cost,” “improve operational throughput”).
For each priority, name the performance driver that determines success or failure. What must change for this priority to advance?
For each driver, define the metric that measures movement. Specify the calculation, data source, and inclusion rules.
Set thresholds — not just targets. A target is the desired level. Thresholds define the bands: green (on track), amber (attention required), red (action required). Without thresholds, there is no trigger for response.
Assign ownership and review cadence. Who reviews this KPI, how often, and what is the expected response protocol when a threshold is breached?
Test the set as a whole. Does the full KPI set cover both leading and lagging signals? Are there gaps where a strategic priority has no corresponding indicator? Are there redundancies where multiple KPIs measure the same driver?
This sequence — strategy, drivers, metrics, thresholds, actions, review cadence — is the chain that connects measurement to management. Breaking any link in the chain produces indicators that describe but do not direct.
Sector considerations
KPI design principles are universal, but the specific indicators vary by sector:
- Manufacturing: Production yield, machine utilisation, quality reject rates, and order-to-delivery lead time are common operational leading indicators. Financial lagging indicators alone miss the shop-floor signals that predict margin erosion.
- Professional services: Utilisation rate, revenue per fee-earner, project margin, and client retention rate. The leading indicator gap is typically in pipeline quality — most services firms track pipeline volume but not pipeline probability or average deal cycle length.
- Retail and distribution: Conversion rate, average basket value, stock turn, and shrinkage rate. Leading indicators often sit in supply chain metrics (stock availability, supplier lead time) that are not surfaced to executive dashboards.
Frequently asked questions
How many KPIs should an organisation track? At executive level, seven to twelve. Below that, each business unit or function may track additional KPIs relevant to their scope, but these should cascade from the strategic set rather than existing independently. If the executive dashboard has more than fifteen indicators, it is not a dashboard — it is a data dump.
What is the difference between a KPI and a metric? A metric is any measurable value. A KPI is a metric that has been tied to a specific objective, given a target, assigned an owner, and placed on a review cadence. All KPIs are metrics; most metrics are not KPIs. For a detailed treatment, see Metrics vs KPIs .
Should KPIs change over time? Yes. KPIs should be reviewed at least annually and adjusted when strategic priorities shift. A KPI that was critical during a growth phase (e.g. customer acquisition cost) may become less relevant during a consolidation phase. The danger is inertia — continuing to track indicators that no longer connect to current strategy.
How do leading and lagging indicators relate? Lagging indicators confirm outcomes (revenue, profit, cash). Leading indicators predict future outcomes (pipeline, backlog, churn signals). An effective KPI set pairs each lagging indicator with at least one leading counterpart. For further context, see KPI Framework for Financial Reporting .
Who should own a KPI? A named individual, not a team or department. Ownership means accountability for understanding movement, proposing corrective action, and reporting at the agreed cadence. The owner does not need to control every variable — they need to interpret the signal and coordinate the response.
Related Reading
- KPI Framework for Financial Reporting
- Metrics vs KPIs — Understanding the Difference
- KPI Hierarchies and Cascading
- Management Dashboard Design
- Variance Analysis Guide
Sources
- BCG/MIT — “60% of managers believe they need better KPIs.” BCG–MIT research on performance management effectiveness.
- Institute of Global Controlling (CZ) — Only 11% of companies manage controlling via process KPIs.
- KPI Institute — 68% of organisations report positive performance improvements after structured KPI tracking.
- Verret/LinkedIn — Executive reporting overload: 52 reports per week reaching senior decision-makers.
Martin Duben is the founder of Onetribe, advising mid-market companies across Central Europe on financial reporting, data governance, and performance management. He works with CFOs to build reporting structures that connect measurement to decision-making.