Risk and measurement

Risk and measurement must be read together because every metric highlights something and leaves something else outside the frame. Measurement helps organisations decide, but it also shapes behaviour.

In brief

A good metric reduces ambiguity. A weak metric can create a false sense of control.

The issue is not whether to measure. The issue is what the measure represents, what it hides and what people will optimise once the measure becomes visible.

Operational definition

A measurement system has three layers: the condition the company wants to understand, the proxy used to represent it, and the behaviour created by that proxy.

Problems appear when the proxy is treated as the condition itself. For example, quote volume is not commercial quality, utilisation is not flow health, and speed is not customer value.

Why it matters for SMEs

SMEs often introduce KPIs during growth, professionalisation or investor preparation. This is useful, but it can distort priorities if indicators are chosen too quickly.

Measuring output without quality can increase rework. Measuring utilisation without WIP can increase lead time. Measuring sales without margin can reward poor growth.

The point is to design metrics around decisions, not around what is easiest to report.

Observable signals

Look for numbers improving while the operating problem remains.

Look for KPIs that teams do not trust.

Look for measures that are reported but do not change decisions.

Look for people changing behaviour to satisfy the indicator while the real condition weakens.

Common mistakes

Measuring what is available instead of what matters.

Adding indicators without deciding which decision they support.

Using one metric where a paired measure is needed.

Measuring individuals when the condition is systemic.

Keeping a metric after the process has changed.

Operational example

A company measures sales only by number of quotes sent. The number rises, but margins fall and corrections increase. The metric is not wrong, but it is incomplete.

The company keeps quote volume visible but pairs it with margin band, first pass yield and quote lead time. The discussion changes because the measure now represents the condition more accurately.

Diagnostic questions

What decision does this metric support?

What condition is it trying to represent?

What does it hide?

What behaviour does it encourage?

Can the metric improve while the real condition worsens?

Which paired measure is needed?

Practical implications

Before adding a metric, write the decision it will support. Define the condition, the proxy and the expected behaviour. Add a paired measure when the first metric creates an obvious blind spot.

Review metrics when processes, roles or decision rights change.

MARTRO reading

In MARTRO’s reading, measurement is part of the operating architecture. It does not only observe the organisation; it changes what people pay attention to.

This is why MARTRO treats measurement risk as a diagnostic theme. Weak metrics can make a company look controlled while hiding the structural condition that matters.

Frequently asked questions

Can a metric create risk? Yes, when it drives attention toward the proxy and away from the real condition.

Should every KPI support a decision? Yes. If no decision changes because of the metric, the metric is probably reporting noise.

What is a paired measure? A second indicator that catches what the first one might hide.

Why combine qualitative and quantitative measures? Because some structural conditions appear before hard numbers are stable.

When should metrics be revised? When the process, role structure or decision model changes.