Is it better to measure fifteen well-chosen indicators or to work with the three or four you already check \"by feel\"? There is no single answer: it depends on how many of the numbers currently in use are really tied to a decision, and how many are there out of habit.
\nIt's common to end up with a dashboard of metrics updated every Monday morning that hardly anyone looks at. It happens to the solo professional who measures revenue but not the real margin per customer. It happens to the business owner with a few dozen employees who gets a monthly report from the accountant and feels that \"something's missing,\" without knowing what. It even happens to the owner of a mid-sized company who has installed a sophisticated business intelligence tool and still ends up reading only the same four recurring numbers.
\nEuropean academic research on small and medium-sized companies confirms the pattern: the indicators actually in use tend to be mostly short-term financial ones, unbalanced and poorly integrated with process, customer and people indicators [3].
\nKPIs — Key Performance Indicators — are the few key indicators that really steer a company's decisions. Not every indicator is a KPI: the distinction matters more than people think, because it makes the difference between a dashboard that drives decisions and one that is only looked at [1].
\nThis article covers what sets a KPI apart from metrics and from other related indicators, how to choose the few that really matter for your business, proposes a selection of 15 essential indicators grouped by area, explains how to build and read them without weighing down the organization, and closes with the most common mistakes that drain meaning even from the best-built selections.
\nIn many companies, \"KPI\" has become a synonym for \"any number on an Excel sheet.\" It's a costly imprecision, because it leads to dashboards of thirty or forty rows from which you can no longer tell what really steers a decision. This section pins down what a KPI is in operational terms and — right after — contrasts it with the four terms it is most often confused with: metric, OKR, SLA, KRI. The distinction isn't academic: it changes how you select what to measure.
\nHow many of the numbers currently called \"KPIs\" in your company are really key indicators?\nYou find the answer by applying Parmenter's selectivity criterion, which puts a company's true KPIs at around ten [1]: all the other numbers are useful metrics, but they aren't \"key\" — and treating them on a par with the few that drive decisions only creates noise.
\nA KPI is an indicator that meets one precise condition: reading it steers a concrete, recurring decision. The word \"key\" isn't decorative — it signals selectivity [1]. Parmenter proposes an operational order of magnitude: a company should have at most about ten KRIs (key result indicators by area), around 80 PIs and RIs (performance indicators and operational result indicators), and no more than 10 true KPIs — the indicators that really drive action [1]. When this distinction isn't made, you end up calling any number that gets calculated a KPI.
\nKPI vs metric. Every KPI is a metric, but not every metric is a KPI. A metric is any quantitative measure; a KPI is the \"key\" metric that steers a decision. Confusing the two leads to calling indicators KPIs when they aren't, and to building encyclopedic dashboards that nobody uses.
\nKPI vs OKR (Objectives and Key Results). They get confused because both \"set measurable goals.\" The difference is sharp: a KPI is an indicator of an ongoing phenomenon (e.g., margin, conversion rate); an OKR is a pair of objective + expected results within a fixed time horizon (usually quarterly). They coexist: the same number can be a KPI for ongoing monitoring and a Key Result in a quarterly OKR.
\nKPI vs SLA (Service Level Agreement). In service companies both measure \"response time,\" \"error rate,\" \"availability.\" The difference: a KPI is internal (it steers management decisions); an SLA faces the customer (it's a contractual commitment, with contractual consequences). The same indicator can be a KPI for whoever monitors it and an SLA for whoever is bound to meet it.
\nKPI vs KRI (Key Risk Indicator). Confused because both are \"key indicators\" and are often measured on the same processes. The difference: a KPI signals how expected performance is trending; a KRI signals a rising risk (customer concentration, financial exposure, unavailability of key staff). KPI = \"are we heading where we want?\"; KRI = \"are we getting closer to a risk?\" Different logics, different dashboards.
\nThis article is part of the pillar on business management, which places KPIs in the broader context of company governance.
\nStudies on small and medium-sized European companies document a recurring imbalance: the KPIs actually in use concentrate on revenue and very short-term margins, while processes, customers and people stay off the dashboard [3]. It isn't a lack of technical skill: it's a problem of selection criteria. This section proposes four operational criteria for choosing the few indicators that really steer decisions in a small or mid-sized company, and for keeping the framework from sliding toward the only area where numbers are already at hand.
\nWhich of the four criteria is skipped most often in smaller companies?\nThe fourth: the explicit owner. Without a name next to it, a KPI becomes shared property — and in a company, whatever has no owner gets read less consistently.
\nCriterion 1 — An explicit link to a decision. If the number never changes a decision, it isn't a KPI [1]. The test is simple: before adding an indicator to the dashboard, answer the question \"if this number drops by 20%, what do we do differently?\" If the answer is \"we don't know,\" the indicator is a useful metric but not a KPI.
\nCriterion 2 — Balance across the four areas. Kaplan and Norton [2] showed that a well-run company measures across four perspectives at once: financial, internal processes, customer, learning and growth (people). Applied at the scale of a smaller company, this means a dashboard with only financial KPIs is a short-sighted dashboard: it spots problems when they have already settled into the economic results, not while they are forming in processes and in relationships with customers and people. The distribution across areas doesn't have to be rigid (4-4-4-3), but it does have to be balanced.
\nCriterion 3 — Data available at a sustainable cost. In a smaller company, a perfect indicator that is extremely expensive to build is as good as a nonexistent one. Real margin per customer is an excellent KPI, but in a company without customer analytics it's too expensive to build month after month: better to start from a proxy like revenue concentration on the top 5 customers, which can be built in twenty minutes. The criterion isn't \"which indicator would be ideal\" but \"which indicator can be built with the data and time available today.\"
\nCriterion 4 — A clear owner. Every KPI has a name next to it: who reads it, who updates it, who reacts to a deviation. Without an explicit owner, the KPI belongs to everyone and no one. Academic research on small and medium-sized European companies documents this as one of the most widespread failure patterns in managing indicators [3].
\nTo place KPIs inside the system that holds them together, it's useful to read the cluster on business performance measurement.
\nThe selection that follows isn't an exhaustive list: it's an operational proposal of 15 indicators — four or five per area — that, taken together, give a balanced picture of a company's health. The four areas (finance, operations, customer, people) follow Kaplan and Norton's matrix [2], but calibrated to the reality of a small or mid-sized company: indicators you can build without a dedicated controller, read in twenty minutes, and that can flag problems before they settle in.
\nHow many of the 15 KPIs does your company already measure consistently, and how many only \"by feel\"?\nThe review of performance measurement systems in small and medium-sized European companies reports that the indicators in use are mostly short-term financial ones and poorly integrated with process, customer and people indicators [3]: full coverage of all fifteen is therefore unlikely, and the people area is the best place to start the comparison.
\n1. Contribution margin (by main product / service line)\nWhat it measures: revenue minus variable costs, as a percentage and in absolute terms.\nHow it's calculated: (Revenue - Variable costs) / Revenue × 100.\nThe question it answers: \"where do we really make money?\" It identifies the business lines that bring in margin and those that erode it.\nWarning threshold: varies by industry; a contribution margin below 40-50% for a service company often points to variable costs out of control.
\n2. Monthly operating cash flow\nWhat it measures: cash generated by day-to-day operations, before investments and financing.\nHow it's calculated: cash collected in the month - payments made in the month (operating costs, suppliers, staff).\nThe question it answers: \"does revenue turn into cash, and how fast?\" A company can be profitable and in a cash crisis at the same time.\nWarning threshold: negative operating cash flow for three consecutive months is an alarm signal to treat as a priority.
\n3. Days sales outstanding (DSO)\nWhat it measures: the average time between issuing an invoice and actually collecting it.\nHow it's calculated: (Accounts receivable / Revenue) × number of days in the period.\nThe question it answers: \"how long does it take us to turn revenue into cash?\" A lengthening DSO erodes liquidity even when revenue is growing.\nWarning threshold: higher than the contractual payment terms (e.g., if terms are 30 days and DSO is 75, there's a systematic problem).
\n4. Revenue concentration on the top 5 customers\nWhat it measures: the percentage of total revenue generated by the five largest customers.\nHow it's calculated: (Revenue from top 5 customers / Total revenue) × 100.\nThe question it answers: \"how exposed are we to losing a single customer?\" A proxy for commercial risk.\nWarning threshold: above 50% signals significant dependence; above 70% is a risk to manage actively.
\n5. On-time delivery rate\nWhat it measures: the percentage of orders/jobs/services delivered within the agreed time.\nHow it's calculated: (On-time deliveries / Total deliveries in the period) × 100.\nThe question it answers: \"are we reliable?\" A proxy for internal reliability and customer perception.\nWarning threshold: below 85-90% signals a systemic planning or capacity problem.
\n6. Rework / nonconformity rate\nWhat it measures: the percentage of output that needs correcting before it's delivered or accepted.\nHow it's calculated: (Reworked output / Total output in the period) × 100.\nThe question it answers: \"how much hidden cost are we paying to fix what we should have done right the first time?\"\nWarning threshold: above 5% in a service or manufacturing company is a significant hidden cost.
\n7. Lead time of the core process\nWhat it measures: the average time between the start and the close of a job / production run / delivery of the core service.\nHow it's calculated: (Sum of throughput times) / Number of jobs in the period.\nThe question it answers: \"how long does it really take us to produce our output?\" A lead time that is high compared with competitors indicates internal inefficiency.\nWarning threshold: comparison with the industry benchmark and with your own previous periods.
\n8. Utilization of key resources\nWhat it measures: the percentage of productive time on bottleneck resources (critical people or equipment).\nHow it's calculated: (Productive hours of the resource / Available hours of the resource) × 100.\nThe question it answers: \"are our critical resources underused or overloaded?\"\nWarning threshold: utilization above 90% is a bottleneck risk; below 60% signals underuse or inefficient allocation.
\n9. New customer acquisition rate\nWhat it measures: the number of new customers acquired in the period, and as a percentage of the active portfolio.\nHow it's calculated: (New customers in the period / Total customers at the start of the period) × 100.\nThe question it answers: \"is our customer base growing?\"\nWarning threshold: an acquisition rate lower than the churn rate means the portfolio is shrinking.
\n10. Active customer retention rate (or repeat rate for transactional businesses)\nWhat it measures: the percentage of customers who renew or come back in the following period.\nHow it's calculated: (Customers at the end of the period who were there at the start / Customers at the start) × 100.\nThe question it answers: \"are customers staying or leaving?\" Until it's measured, churn stays invisible until it shows up in revenue; and the comparison between acquisition cost and retention cost, for your own company, can only be made after you've calculated both.\nWarning threshold: in recurring-revenue businesses, a retention rate below 80% is a signal worth investigating.
\n11. Net Promoter Score (NPS) or an equivalent satisfaction measure\nWhat it measures: the likelihood that a customer would recommend the company (0-10 scale).\nHow it's calculated: NPS = % promoters (9-10) - % detractors (0-6). Simplified alternative: the average rating on a 1-5 scale from a single monthly question.\nThe question it answers: \"how satisfied are customers, in a nutshell?\"\nWarning threshold: a negative NPS (more detractors than promoters) is an urgent signal; an NPS below 20 in industries where the average is 40-50 points to a problem with perceived quality.
\n12. Average response time to customer requests\nWhat it measures: the average time between receiving a request and the first substantive reply.\nHow it's calculated: the average of response times on a sample from the period (e.g., within 24 hours vs 48 vs longer).\nThe question it answers: \"how responsive are we?\" Particularly relevant for services and B2B.\nWarning threshold: varies by industry, but beyond 24 hours in B2B the risk of eroding satisfaction rises significantly.
\n13. Voluntary employee turnover rate (annualized)\nWhat it measures: the percentage of people who leave the company of their own accord in a year.\nHow it's calculated: (Voluntary departures in the period / Average headcount) × 100 × (12 / months in the period).\nThe question it answers: \"do people choose to stay?\" Unmeasured turnover isn't managed, but its cost (recruiting, training, loss of know-how) is always there.\nWarning threshold: above 15-20% a year in non-seasonal industries signals a problem with workplace climate or compensation.
\n14. Absenteeism rate\nWhat it measures: days of unplanned absence over available working days.\nHow it's calculated: (Days of unplanned absence / Total available working days) × 100.\nThe question it answers: \"are there signs of discomfort or excessive workload?\" Absenteeism is often an early indicator of a deteriorating work climate.\nWarning threshold: above 3-5% in non-seasonal industries is a signal worth investigating.
\n15. Engagement / climate indicator (even in its minimal form)\nWhat it measures: the subjective perception of how engaged and satisfied people are at work.\nHow it's calculated: a single monthly question on a 1-10 scale (\"How would you rate your level of job satisfaction this month?\"), averaged across the team. Enough to detect trends.\nThe question it answers: \"are people engaged, or are they keeping their heads down waiting for the right opportunity?\"\nWarning threshold: a drop of more than one point in the monthly average for two consecutive months is a signal to discuss with the team.
\nSummary table of the 15 KPIs:
\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n| # | KPI | Area | Decision question | Indicative threshold |
|---|---|---|---|---|
| 1 | Contribution margin | Finance | Where do we really make money? | > 40% services |
| 2 | Monthly operating cash flow | Finance | Does revenue turn into cash? | Positive 3+ months |
| 3 | DSO — Days sales outstanding | Finance | How long to collect? | ≤ contractual terms |
| 4 | Revenue concentration top 5 | Finance | Exposure to a single customer? | < 50% |
| 5 | On-time delivery | Operations | Are we reliable? | > 85-90% |
| 6 | Rework rate | Operations | How much hidden cost? | < 5% |
| 7 | Core process lead time | Operations | How long to produce? | Industry benchmark |
| 8 | Key resource utilization | Operations | Bottleneck or underuse? | 60-90% |
| 9 | New customer acquisition rate | Customer | Are we growing? | > churn rate |
| 10 | Customer retention rate | Customer | Are customers staying? | > 80% |
| 11 | NPS / satisfaction | Customer | Perceived quality? | NPS > 20 |
| 12 | Customer response time | Customer | Are we responsive? | < 24h B2B |
| 13 | Voluntary turnover | People | Do people choose to stay? | < 15-20% a year |
| 14 | Absenteeism | People | Signs of discomfort? | < 3-5% |
| 15 | Engagement / climate | People | Team engagement? | Stable or rising trend |
scroll the table →
\nNote: the thresholds are indicative. They vary by industry, size and stage of development of the company. They need to be calibrated to your specific business before they become operational thresholds.
\nThe same indicator — \"on-time delivery rate,\" for example — can be built in five different ways, each producing a different number. If the formula isn't written down in black and white, the KPI changes from month to month without the company noticing. This section defines the four elements every KPI must have before it goes on the dashboard: formula, data source, scope, frequency.
\nHow can you tell a KPI is \"well built\"?\nBy one thing only: two different people, starting from the same data, get the same number. If that doesn't happen, it isn't a calculation problem: it's a definition problem.
\nElement 1 — An explicit formula. Numerator, denominator, unit of measure: written once and frozen. Parmenter points out that most faulty KPIs aren't faulty because the wrong indicator was chosen, but because of how it was defined [1]. A practical example: the \"on-time delivery rate\" calculated in three different ways produces three different numbers —
\nWithout a fixed formula, month after month you're comparing apples and oranges.
\nElement 2 — Data source. Which system the data comes from, when it is extracted, from what starting data. If the accountant calculates the collection rate from the provisional financial statements and the business owner calculates it from the bank account, they'll get different numbers. The source must be single and declared.
\nElement 3 — Scope. Which customers, jobs and cost centers are included in the calculation and which aren't. A \"retention rate\" that includes occasional customers and one that counts only recurring customers produce very different values. The scope must be defined up front and stated every time the KPI is reported.
\nElement 4 — Frequency. Daily, weekly, monthly or quarterly: the frequency must match the speed of the decision the KPI steers. Cash flow is checked weekly (or even more often in critical periods); engagement is measured monthly; turnover is calculated quarterly or annually. Updating the formula without applying it retroactively to past data produces a trend that looks readable but compares non-homogeneous periods [1].
\nThe foundation of a systemized set of KPIs — the written formula, the declared source, the explicit scope — is an integral part of a broader approach to business systemization.
\nIn Italy, 48.8% of small and medium-sized companies use ERP software, but the share that has a structured monthly ritual for reading KPIs would need to be checked against a specific source [5]. The tool may be there, the KPI may exist, and still not steer decisions if there is no systematic reading, at the right frequency, in front of the right people. This section describes how to set up an essential reading ritual — even in a very small company — so that every KPI produces at least one operational decision per period.
\nHow can you tell whether a KPI is really active in the company?\nBy one thing only: in the last three review meetings, at least once a deviation in the KPI changed a decision that would otherwise have been made differently. If that doesn't happen, the KPI is decorative, not operational.
\nFrequency by area:
\nThe three roles for every KPI:
\nThe minimum ritual for a smaller company:\nOne review meeting a month, lasting no more than 30 minutes, focused exclusively on KPIs with a significant deviation from the previous period or from the warning threshold. For each deviation, at least one operational decision: \"DSO has risen to 85 days — who calls the customers more than 60 days late by the end of this week?\"
\nTools by category, never by specific vendor:
\nThe tool follows the maturity of the ritual, not the other way around. For the process that orchestrates the reading ritual within the budgeting-and-reporting cycle, the reference is the cluster on management control.
\nIn a smaller company, a KPI that isn't linked to an explicit operational decision gets read and then forgotten. The deviation is noticed, discussed in the meeting, but the next meeting finds the same deviation — because it wasn't clear who was supposed to do what. This section proposes a simple action logic for every KPI: decide in advance, even before you start measuring it, which decision will be made if the deviation crosses a certain threshold.
\nFor each of the 15 KPIs selected, is there already an explicit decision tied to a deviation?\nThe answer is rarely \"yes\" for all fifteen: the associated decision tends to exist for financial KPIs, because cash speaks for itself, and to be missing for customer and people KPIs — the same imbalance documented by the review of measurement systems in small and medium-sized European companies [3], and exactly the areas that, when ignored, generate the most expensive problems in the medium term.
\nEvery KPI is \"christened\" with three pieces of information before it goes on the dashboard:
\nOperational example for a solo professional:\n\"Revenue concentration on the top 5 customers > 60% → warning threshold: launch at least two business development actions in the quarter; > 75% → alarm threshold: the quarter's sole commercial priority becomes diversification.\"
\nOperational example for a small or mid-sized company:\n\"DSO > 75 days → warning threshold: review payment terms for new contracts; > 90 days → alarm threshold: extraordinary meeting with sales and accounting, with a collections plan within seven days.\"
\nThe useful distinction is between an automatic action (decided in advance, triggered by itself when the threshold is crossed — e.g., \"if the retention rate falls below 75%, the following week all non-renewing customers are contacted\") and a discretionary action (the deviation triggers an analysis meeting, not an immediate decision — e.g., \"if lead time rises by 20%, the cause is assessed together with the operations manager before acting\"). The distinction avoids both inertia and overreaction.
\nFor the connection between KPIs as a measurement tool and the ongoing process that turns them into decisions, the reference is the cluster on management control.
\nThe review by Bititci and colleagues of more than 6,000 academic articles identifies a set of failure patterns in measurement systems that recur with surprising regularity, regardless of industry and size [4]. Applied to the selection and management of KPIs in a smaller company, they become six concrete mistakes — at least two of them counterintuitive. Knowing them before you start lets you avoid the \"second attempt after the first one failed,\" which in a company costs twice.
\nWhich of the six mistakes is the most expensive when it shows up silently?\nIt isn't overload — that one is visible. It's the fifth: a definition that changes over time. Because it produces a trend that looks readable but is comparing apples and oranges without the change being made explicit.
\nMistake 1 — Overload. Choosing too many indicators, ignoring the selectivity rule [1]. The symptom is a dashboard with more than 15 KPIs that gets only partly read. The fix is to cut: keep only the indicators that meet the criterion of an explicit link to a decision.
\nMistake 2 — Imbalance. KPIs concentrated in a single area, almost always financial [3]. The symptom is the complete absence of people or customer indicators — and the problem arrives from those directions six to twelve months later. The fix is to check the distribution across the four areas before finalizing the selection.
\nMistake 3 — Output-only indicators, no process indicators. Measuring the final result without measuring the intermediate steps that produce it. The symptom is discovering the problem when it's too late to correct it in the current period. The fix is to pair the result indicators (revenue, margin) with at least one process indicator (lead time, on-time delivery) per area.
\nMistake 4 — The wrong frequency. (Counterintuitive mistake no. 1.) Cash KPIs read quarterly (too late to step in) or engagement KPIs read weekly (too early to get a reliable signal). The fix is to align the frequency with the speed of the decision the KPI steers.
\nMistake 5 — A definition that is inconsistent over time. (Counterintuitive mistake no. 2.) The KPI formula changes silently from month to month — because the person calculating it isn't always the same, or because it gets adjusted \"for convenience\" without the change being declared. The result is a history that looks readable but compares non-homogeneous periods [1]. The fix is to document the formula explicitly and change the version number every time the formula is modified.
\nMistake 6 — KPIs with no associated decision. Indicators measured out of habit, not to decide [4]. The symptom is a deviation that is commented on in the meeting and then forgotten. The fix is to assign, before launch, a warning threshold, an alarm threshold and a typical decision to every KPI on the dashboard.
\nThe selection of 15 KPIs and the criteria proposed in this article are calibrated for small and mid-sized companies in services, retail and light manufacturing. Some conditions limit their applicability:
\nHow many KPIs should a 10-person company have?\nA dashboard of 8-12 KPIs, spread across three or four areas (finance, operations, customer), is enough to steer the main decisions. The people area can be covered by 1-2 simple indicators. Beyond 15 indicators in total, the dashboard tends to become unmanageable.
\nShould KPIs be the same across all industries?\nNo. The proposed selection is an indicative baseline; calibrating it to the specifics of your industry is essential. A software company has different process KPIs from a manufacturing company. The four-area framework (finance, operations, customer, people) remains valid; the individual indicators need to be adapted.
\nDo you need specific software to manage KPIs?\nNot at the start. A structured spreadsheet, updated consistently, is enough for organizations of up to 15-20 people. The tool is chosen based on the maturity of the reading ritual, not the other way around.
\nWhat should you do if a KPI never moves?\nEither the threshold is poorly calibrated (too far from the current value), or the indicator isn't really critical for the company. In both cases it needs to be reviewed: by changing the threshold or replacing the KPI with one more sensitive to operational decisions.
\nHow often should you review your KPI selection?\nAn annual review of the selection is the cadence recommended here. KPIs aren't permanent: they change as the company's stage of development, strategic priorities and data availability change.
\nA company's KPIs work when they meet four conditions: they are selected with explicit criteria (a link to a decision, balance across four areas, data availability, a declared owner), they are built with a fixed formula, source, scope and frequency, they are read through a structured, periodic ritual, and they have an operational decision tied to every deviation threshold. The proposed selection of 15 indicators — four or five per area (finance, operations, customer, people) — is an operational baseline, not dogma: it has to be calibrated to your specific business and reviewed at least once a year.
\nA company's KPIs aren't chosen one at a time: they are chosen as a balanced selection across four areas — finance, operations, customer, people — and managed with four operational disciplines: define them with a written formula, read them at the right frequency, tie them to explicit decisions, and avoid the six recurring mistakes. This is the key point: a growing company doesn't need an encyclopedic dashboard, it needs around fifteen indicators that, taken together, give a reliable picture of the company's health — and, above all, that produce decisions.
\nTo place the 15 KPIs inside the system that holds them together — the map linking indicators and strategy — it's worth reading the cluster on business performance measurement. To understand the ongoing process that turns KPIs into decisions month after month — the budgeting, reporting and variance analysis cycle — the reference is the cluster on management control. To frame it all within the broader architecture of running a small or mid-sized company, the reference pillar is the one on business management.
\nA company that masters its KPIs stops discovering problems once they've already settled in: it catches them while they're forming, because it has decided in advance what to look at and what to do when the number moves. The business owner goes from a feeling that \"something doesn't add up\" to being able to point out, in a few minutes, which indicator has changed direction and which decision it calls for. At the national level, the Bank of Italy measured structured management practices in about 3,200 Italian companies with at least twenty employees — among the questions, how many performance indicators are monitored in the company — and found a positive association with productivity [6]: a finding that places KPI management within the set of things that distinguish the most productive companies, without proving that it causes their productivity.
\nFor managers and business owners, KPIs aren't a technical exercise: they are the cheapest way to stop deciding by feel.
\n[1] Parmenter, D., \"Key Performance Indicators: Developing, Implementing, and Using Winning KPIs\", Wiley, 4th edition, 2019-2020. Available at: https://www.wiley.com/en-us/Key+Performance+Indicators%3A+Developing%2C+Implementing%2C+and+Using+Winning+KPIs%2C+4th+Edition-p-9781119620778
\n[2] Kaplan, R. S., Norton, D. P., \"The Balanced Scorecard: Translating Strategy into Action\", Harvard Business School Press, 1996. Available at: https://hbswk.hbs.edu/item/the-balanced-scorecard-translating-strategy-into-action
\n[3] Garengo, P., Biazzo, S., Bititci, U. S., \"Performance Measurement Systems in SMEs: A Review for a Research Agenda\", International Journal of Management Reviews, vol. 7, 2005. Available at: https://onlinelibrary.wiley.com/doi/10.1111/j.1468-2370.2005.00105.x
\n[4] Bititci, U. S., Garengo, P., Dörfler, V., Nudurupati, S., \"Performance Measurement: Challenges for Tomorrow\", International Journal of Management Reviews, vol. 14, 2012. Available at: https://onlinelibrary.wiley.com/doi/10.1111/j.1468-2370.2012.00336.x
\n[5] ISTAT, \"Imprese e ICT, Anno 2025\", Istituto Nazionale di Statistica, 2025. Available at: https://www.istat.it/comunicato-stampa/imprese-e-ict-anno-2025/
\n[6] Baltrunaite, A., Formai, S., Linarello, A., Mocetti, S., \"Proprietà, governance, management e performance delle imprese: evidenze dalle imprese italiane\", Banca d'Italia, Questioni di Economia e Finanza n. 678, March 2022. Available at: https://www.bancaditalia.it/pubblicazioni/qef/2022-0678/QEF_678_22.pdf
","headings":[{"level":2,"text":"What a KPI really is (and why many indicators aren't)","id":"what-a-kpi-really-is-and-why-many-indicators-arent"},{"level":2,"text":"How to choose the right KPIs for your company","id":"how-to-choose-the-right-kpis-for-your-company"},{"level":2,"text":"The 15 essential business KPIs, grouped by area","id":"the-15-essential-business-kpis-grouped-by-area"},{"level":3,"text":"Financial indicators","id":"financial-indicators"},{"level":3,"text":"Operational indicators","id":"operational-indicators"},{"level":3,"text":"Customer indicators","id":"customer-indicators"},{"level":3,"text":"People indicators","id":"people-indicators"},{"level":2,"text":"Building each indicator correctly (formula, source, frequency)","id":"building-each-indicator-correctly-formula-source-frequency"},{"level":2,"text":"Reading KPIs at the right frequency, with the right owners","id":"reading-kpis-at-the-right-frequency-with-the-right-owners"},{"level":2,"text":"Linking every KPI to an operational decision","id":"linking-every-kpi-to-an-operational-decision"},{"level":2,"text":"Avoiding the most common mistakes in choosing and managing KPIs","id":"avoiding-the-most-common-mistakes-in-choosing-and-managing-kpis"},{"level":2,"text":"Limits and conditions of applicability","id":"limits-and-conditions-of-applicability"},{"level":2,"text":"FAQ","id":"faq"},{"level":2,"text":"Operational summary","id":"operational-summary"},{"level":2,"text":"Conclusion","id":"conclusion"},{"level":2,"text":"Sources and references","id":"sources-and-references"}],"tldr":"A KPI is an indicator whose reading steers a concrete, recurring decision: every KPI is a metric, but not every metric is a KPI [1].; KPIs are chosen with four criteria: a link to a decision, balance across finance, operations, customer and people [2], data available at a sustainable cost, and a declared owner.; The selection is 15 indicators across four areas, calibrated for small and mid-sized companies in services, retail and light manufacturing; for a solo professional it is oversized.; Every KPI goes on the dashboard with a fixed formula, source, scope and frequency: two people, starting from the same data, must get the same number.; Before you measure it, every KPI needs a warning threshold, an alarm threshold and the decision triggered when they are crossed.","tldrItems":["A KPI is an indicator whose reading steers a concrete, recurring decision: every KPI is a metric, but not every metric is a KPI [1].","KPIs are chosen with four criteria: a link to a decision, balance across finance, operations, customer and people [2], data available at a sustainable cost, and a declared owner.","The selection is 15 indicators across four areas, calibrated for small and mid-sized companies in services, retail and light manufacturing; for a solo professional it is oversized.","Every KPI goes on the dashboard with a fixed formula, source, scope and frequency: two people, starting from the same data, must get the same number.","Before you measure it, every KPI needs a warning threshold, an alarm threshold and the decision triggered when they are crossed."]}