Measures & Objectives

Organizations rarely suffer from a shortage of Measures.

They have scorecards, service levels, productivity indicators, risk Measures, financial Measures, customer Measures, project Measures, and increasingly AI-generated Measures.

The harder question is whether those Measures still support the results the organization intends to achieve.

A team can hit its target. A function can improve its score. Every dashboard can look green.

And the organization can still move away from the outcome that matters.

The management question is:

Do the Measures we use provide valid evidence about, and encourage behavior consistent with, the objectives or management purposes they are intended to support?

Start with the intended result

An Organizational Objective specifies a result the organization intends to achieve, maintain, prevent, protect, reduce, or change.

A Measure represents some characteristic of organizational reality.

A Performance Measure is deliberately used to assess a specified Measurement Object against an identified management purpose, criterion, or Organizational Objective.

The distinction matters.

If management allows the Measure to become the objective, it can end up improving what is easy to count rather than what actually matters.

Suppose the objective is:

Improve customer problem resolution.

Management selects average handling time.

Calls become shorter. The target is achieved. But customers call again because problems were not fully resolved.

The Measure improved.

The objective did not.

A Measure is not the objective

This boundary becomes harder to preserve once bonuses, performance ratings, funding, executive attention, escalation, or automated actions depend on the Measure.

People and systems adapt.

That adaptation may be desirable. A useful Performance Measure should influence behavior toward the intended result.

The risk begins when behavior shifts toward improving the Measure itself rather than the underlying condition.

A service organization optimizes handling time rather than resolution. Sales optimizes bookings while margin weakens. A delivery team optimizes milestone completion while unresolved quality problems accumulate.

The problem is not that Measures influence behavior.

It is misalignment between the behavior rewarded by the Measure and the result the Measure is supposed to support.

Validity and distortion are different questions

Two controls are needed.

First:

Does the Measure validly represent the characteristic management claims it represents?

That is Measurement Validity.

Second:

What behavior occurs because this Measure has become consequential?

That is the territory of Measurement Distortion.

A Measure can be valid and still create undesirable behavior when targets, incentives, or authority make it consequential.

The organization should therefore test both representation and behavioral effect.

Local objectives can legitimately conflict

Different functions often have different responsibilities.

Finance manages financial discipline. Operations manages throughput and reliability. Risk manages exposure. Sales manages growth. Customer teams manage service and retention.

These objectives can all be legitimate and still conflict because of shared resources, dependencies, Stakeholder effects, or different time horizons.

A procurement team can meet a legitimate cost objective while operations absorbs quality consequences. A risk function can pursue a legitimate control objective while execution slows materially.

That is not automatically a Measurement failure.

Sometimes management must reconcile legitimate objective conflict.

The aim is not to make every function use the same Measure.

It is to ensure that consequential local Measures do not systematically drive behavior away from broader organizational purpose.

Alignment does not mean identical Measures

Different parts of the organization should measure the work and outcomes they can meaningfully influence.

A customer outcome may depend on sales quality, order accuracy, provisioning speed, service reliability, billing accuracy, issue resolution, and retention.

Each function may need different Measures.

The organization should still be able to explain:

• What objective or management purpose does this Measure support?

• What part of that objective does it represent?

• What does it leave out?

• What behavior is likely when the Measure becomes consequential?

• Could improving it materially weaken another important objective?

That is stronger than simply cascading one KPI through every level.

Targets are useful, but they can displace purpose

Targets can create focus, make expected performance explicit, and support Evaluation and accountability.

The problem is not the target itself.

It is target displacement.

Once “95 percent” becomes the operational goal, the organization may stop asking what the 95 percent was intended to accomplish.

Keep three things distinct:

Objective — the result the organization intends.

Measure — evidence about the result or a relevant condition.

Target — a desired level, tolerance, state, or timing.

They should be connected without becoming interchangeable.

AI can optimize the wrong thing very effectively

AI increases the importance of this distinction because it can optimize a defined Measure continuously.

Give a system a response-time target and it may reduce response time. That does not mean customer value improved.

Give it a conversion target and it may increase conversion. That does not mean the resulting business is profitable, suitable, or consistent with risk appetite.

Give it a cost objective and it may identify substantial savings. That does not establish that the wider Management System can absorb the consequences.

The stronger the optimization capability, the more important objective clarity becomes.

AI capability does not independently determine what the organization ought to optimize.

Measures need continuing review

Even a well-designed Measure can lose alignment.

Objectives change. Processes change. Technology changes. Organizational boundaries move. A proxy weakens. People learn how to optimize the Measure.

Measurement Governance should therefore keep consequential Measures tied to an identifiable purpose and review them when use, context, definitions, consequences, or interpretation materially change.

A Measure should not remain on a scorecard simply because it has always been there.

What management should look for

Take a Measure that materially influences behavior.

Ask:

• What objective, criterion, or management purpose is it intended to support?

• Why should movement in this Measure provide useful evidence about that purpose?

• What important parts of the objective are not represented?

• What behavior does the Measure create?

• Could aggressive optimization weaken another objective?

• If we optimized this Measure perfectly, could we still fail to achieve the intended result?

If the answer to the last question is yes, the Measure may still be useful.

Management simply needs to remain conscious of what it does not represent.

Boundary

A Measure is not an Organizational Objective.

A target is not an Organizational Objective.

A Performance Measure does not become the performance condition it represents.

Improving a Measure does not automatically establish achievement of the objective.

Locally legitimate objectives do not automatically produce coherent Organizational Performance.

Alignment does not require identical Measures.

And AI's ability to optimize a Measure does not establish that the Measure is the right thing to optimize.

The governing rule is:

Define the result or management purpose first, then use Measures that provide valid evidence about it while remaining alert to what those Measures omit and what behavior they create.