What happens to a company the day its best production manager quits?

For most SMEs, the honest answer is that nobody knows for certain, and that is precisely the problem. Not because the skills to replace him are missing, but because part of how the company actually functions was never written down anywhere. It lived in his head, in his ability to step in on any machine, to read a shop floor at a glance, to solve problems before they became visible to everyone else.

It’s worth asking this question not out of alarmism, but because it reveals something structural. In many SMEs, day-to-day performance depends more on the presence of a few excellent people than on the strength of the system around them. When an organization runs well mainly because a handful of people can continuously spot, interpret, and resolve exceptions, that individual competence has not yet turned into organizational capability. The company depends on whoever solves the problem, not on a system that enables the problem to be solved.

This is where the topic of KPIs becomes, in our view, far more relevant than simple reporting.

The thesis of this article is that, in SMEs, KPIs create value not when they increase control over people, but when they reduce dependence on individual intervention and increase the organization’s shared capacity to see, decide, and correct. When instead they are designed or used merely as control targets, they can degrade the quality of decisions, behaviors, and even the information management relies on.

Three arguments support this thesis.

1. The First Job of a KPI Is to Direct Managerial Attention, Not to Measure

A production manager can spend an entire day solving problems and be extremely productive. That, however, does not necessarily mean he is governing the department well. The difference is subtle but decisive, and it’s one of the distinctions we most often see slip past even highly capable, technically skilled managers.

In daily operations, attention naturally gets pulled toward whatever problem is most recent, most urgent, or loudest. The machine that stops, the client who calls, the late delivery, the material that’s missing.

Management needs to do something different: learn to tell the episode apart from the signal.

A machine stoppage can be an episode. A progressive increase in downtime hours on the same family of equipment is a signal. An urgent delivery handled manually can be an episode. A systematic deterioration in OTIF, concentrated on a specific product family or process, is a signal.

It is this shift, from event to managerial insight, that makes a KPI useful. Which is why the starting question shouldn’t be “what data can we measure?” but rather “which decisions do we need to be able to make better?”. It sounds like a semantic nuance, but it completely changes how a performance management system gets built. A metric describes. A KPI should direct attention, raise a question, and, when needed, trigger a decision.

This isn’t an isolated impression. Even APQC’s report on operational KPI priorities for 2026 doesn’t focus only on selecting the right indicators, but distinguishes three separate problems: choosing the correct measures, reporting them, and, above all, applying them. The issue, in other words, doesn’t end once the number shows up on a dashboard.

There’s also a striking figure, published in May 2026 by the World Economic Forum. António Costa, CEO of the Kaizen Institute, reports the results of a survey of more than 300 business leaders. 69% recognize the strategic potential of their own metrics, but 57% point to a lack of leadership involvement as the main threat to their effectiveness. The line he uses is a sharp one: “more dashboards do not solve a meaning problem.”

We now have a measurement capability far greater than we had ten or twenty years ago, yet the value we extract from it still depends on the quality of the questions management knows how to ask.

For an SME, this means a good KPI system should let a manager walk into a meeting and quickly understand where a meaningful deviation exists, which process is driving it, who can influence it, and what decision needs to be made. When that happens, the manager genuinely starts moving out of the primary-executor mindset and into the manager mindset. A KPI doesn’t eliminate hands-on experience. It raises it to a higher level.

2. A KPI Doesn’t Just Observe an Organization. It Changes It.

This is, in our view, the most delicate point in the entire discussion.

We tend to picture a measure as a mirror, something that objectively photographs what’s happening. But inside an organization, a measure is never fully neutral. The moment we declare that a number matters, we’re telling people that outcome deserves attention. When we then attach goals, rewards, or consequences to that number, its power grows further, and people naturally start adapting their behavior in response.

If we measure only units produced, we may unintentionally incentivize unnecessary output. If we evaluate a salesperson purely on revenue, we may generate sales growth without protecting margin or credit quality. If we reward on-time delivery in production without also tracking inventory and schedule changes together, we may improve one indicator simply by shifting the problem elsewhere in the company.

None of this is new. A review published in Management Accounting Research, drawing on 76 empirical studies, had already shown how performance measurement systems produce consequences across three dimensions: people’s behavior, organizational capabilities, and performance. But a 2026 systematic review in the Journal of Management Control, analyzing 95 studies published between 1956 and 2024, takes it further. It identifies recurring consequences of poorly designed or poorly used control systems, including gaming, information manipulation, ethical erosion, reduced motivation, and purely symbolic compliance.

We don’t think the right takeaway is that “KPIs are harmful.” That would be a superficial reading, and the research itself says as much. Outcomes depend heavily on how the system is designed and how management actually uses it. When very strong incentives are tied to overly narrow metrics, the temptation to optimize the number rather than the underlying result increases. When a system built for learning gets used mainly to punish, fear, information filtering, and defensive behavior tend to follow. When control is instead used interactively, through dialogue, interpretation, and adaptation, it can foster learning and the ability to catch problems earlier.

We can make this distinction concrete with an example that comes up often in production meetings. A line’s efficiency drops from 78% to 69%. The plant director can ask, “why are you at 69%?”. Or he can ask, “what’s driving the 69%, and what do we need to change in the process?”.

The number is the same. The information system is the same. But the managerial system is completely different. In the first case, the KPI risks becoming a grade handed out to a person. In the second, it becomes a shared object through which a group tries to understand a process.

None of this means eliminating accountability, quite the opposite. Accountability means knowing what result is expected, which levers can actually be pulled, and what actions need to be taken when the result deviates. But accountability and blame remain two different things. A good KPI shouldn’t reduce a person’s responsibility. It should make the boundaries of that responsibility clearer.

The 2026 review adds an interesting piece. Leadership style, culture, organizational structure, and the nature of the work itself all shape the effect of control systems. The same indicators can produce very different behaviors in different contexts, something we consistently see when working with different entrepreneurial environments, whether in Italy, in Türkiye, or in other markets. There is no “best practice” dashboard that can simply be copied. What exists is a performance management system that is coherent with a specific company’s strategy, processes, responsibilities, and culture.

3. More Technology Makes Good Management More Important, Not Less

A theme that comes up constantly in conversations with SME entrepreneurs is systems integration, getting tools, data, and processes that currently live in isolation to actually talk to each other. What’s telling is that, twenty years into technological transformation, the underlying problem these leaders perceive hasn’t really changed. The data exists, or should exist, but it doesn’t flow well. And when it doesn’t flow, day-to-day operations pay the price, with people forced to act as bridges, interpreting, chasing down information, patching things by hand.

ERP, MES, CRM, business intelligence, IoT, and now AI can dramatically reduce the work needed to collect and interpret information. That’s real progress. But it hides a risk: mistaking a greater capacity to measure for a greater capacity to manage.

A 2026 review in Frontiers, focused specifically on healthcare, offers an idea we find just as useful in an industrial setting. According to the authors, “technology acts less as a transformer than as an amplifier of KPI systems.” If the indicator is correct, responsibilities are clear, and management knows how to use it, technology amplifies the ability to see and to act. But if the indicator is wrong, or nobody really knows who’s supposed to act, technology simply lets the organization make the wrong move faster, with better-looking data.

The same review notes that greater availability, timeliness, and integration of data reduce many technical limitations, but they don’t automatically eliminate phenomena like metric dominance, gaming, and the performance paradox, situations in which people gradually learn to improve the indicator without actually improving the underlying performance.

This becomes even more important with AI. Harvard Business Review put it in a simple question: how can leaders ensure that speed and efficiency don’t come at the expense of judgment and accountability? The question is about AI, but it’s really about management.

A study in the Journal of Management Control, based on responses from 273 middle managers, found a positive association between the use of Big Data Analytics in performance management and the effectiveness of performance measurement systems. But it also found that top management support and employee empowerment are positively associated with the very use of these technologies. In other words, even with highly sophisticated tools, organizational variables remain central. A second 2026 study on the evolution of performance measurement and management systems reaches a similar conclusion by a different route. The concept of a KPI is progressively shifting away from being purely an information-systems matter and toward a broader dimension that includes organization, management, and the human factor.

This brings us back to a typical SME problem. When systems don’t talk to each other, people become the integration layer. They copy data, call a colleague, interpret, reconcile spreadsheets, correct things manually. And when information isn’t reliable, delegation itself becomes fragile, because the entrepreneur feels the need to step back into decisions personally. Integrating systems, though, is only half the job. The other half is deciding which information needs to reach whom, at what frequency, to trigger what action. Technical integration improves the flow of data. Performance management needs to improve the flow of decisions.

From a Reporting System to a Management System

For an entrepreneur or manager running an SME, all of this translates into five very concrete principles.

1. Start from decisions, not from data. Before creating a KPI, clarify which decision it should improve and which behavior it should encourage.
2. Assign ownership, not just targets. Every meaningful indicator needs someone who can interpret it, knows the available levers, and can trigger action.
3. Balance outcomes and drivers. Revenue, EBITDA, and cash flow are essential, but they mostly describe what has already happened. Management also needs to watch the processes and behaviors that will generate those results.
4. Use a KPI to open a conversation, not to close one. A deviation is not an explanation. It’s the starting point for investigating causes, responsibilities, and actions.
5. Review indicators periodically. A useful KPI today can become irrelevant tomorrow, or, worse, become so internalized by the organization that it no longer accurately represents the performance it was meant to capture.

There’s no need to measure more. Often, the better move is to measure less, but better.

A Company Grows When a Number Becomes Organizational Capability

There’s a simple way to tell whether a KPI system is actually working: don’t just look at the dashboard, look at what happens to the people around it.

Do managers spot problems earlier? Do discussions become more specific? Are decisions made closer to where the problem originates? Do people surface bad news faster, or do they try to hide it? Is the entrepreneur increasingly able to delegate without losing visibility? Do managers spend less time putting out fires and more time understanding why those fires keep starting?

These, in the end, are the questions that determine the quality of a performance management system. Because the real goal isn’t building an organization where the entrepreneur can control every number. It’s building one where the right people see the right numbers, know how to interpret them, and have both the responsibility and the tools to act.

An SME doesn’t become managerially mature by installing a dashboard. It becomes managerially mature when it manages to turn its people’s experience into a method the organization can rely on, even without depending continuously on those same people.

Because a strong organization isn’t one that no longer needs excellent people. It’s one that manages to turn their excellence from an individual act into a collective capability.

---

References

• APQC, 2026 KPI Priorities Report
• A. Costa (Kaizen Institute), World Economic Forum, May 2026 — “More dashboards do not solve a meaning problem”
• Review in Management Accounting Research — analysis of 76 empirical studies on the consequences of performance measurement systems
• Systematic review, Journal of Management Control, 2026 — analysis of 95 studies (1956–2024) on control systems
• Review, Frontiers, 2026 — “technology acts less as a transformer than as an amplifier of KPI systems” (healthcare sector)
• Harvard Business Review — on balancing speed/efficiency with judgment/accountability in the AI era
• Study in Journal of Management Control — survey of 273 middle managers on Big Data Analytics and performance management
• 2026 study on the evolution of performance measurement and management systems