Research 2026-09-18 Decision-making Geneva

Numerical targets: what research allows us to expect

Few managerial practices rest on a literature as extensive as goal setting. Few, too, are transposed so casually. The useful question is not 'should there be targets?' but 'under what conditions does a numerical target improve anything other than the number itself?'.

What the theory establishes

Edwin Locke and Gary Latham (2002, American Psychologist) synthesised three decades of work: a specific, demanding goal is associated with higher performance than a vague 'do your best' instruction. The finding is robust but conditional: it assumes commitment to the goal, regular feedback on progress, and the ability to reach it. Removing one of these conditions does not merely weaken the effect — it makes it unpredictable.

The published critique, often ignored

Maurice Schweitzer, Lisa Ordóñez, Adam Galinsky and Max Bazerman (2009, Academy of Management Perspectives) documented the side effects of poorly calibrated goals: narrowing onto what is measured, increased risk-taking, eroded cooperation and, in some experimental settings, dishonest behaviour. Locke and Latham replied in the same journal that year. The debate is unsettled, and instructive precisely because it concerns conditions, not the principle.

Numerical targets: what research allows us to expect — SHR, Geneva — Genève

The confusion to clear up

A goal is not an indicator. Donald Campbell (1979) had already stated the mechanism: the more a quantitative indicator is used to decide, the more it is exposed to distortion, and the less it informs. An organisation that turns every measure into a target gradually loses its reading instruments — with nothing in the dashboard to signal it. (our executive and employee training programmes)

What remains defensible

Current knowledge supports a narrow version: a numerical target is useful when it bears on an activity whose outcome the person genuinely controls, when it comes with an intermediate checkpoint, and when it coexists with at least one qualitative criterion that resists conversion into a figure. None of this has been demonstrated specifically in a Swiss context, and none of it replaces a decision about resources.

One practical implication, testable internally

Before redesigning a target system, one measurement suffices: over the last quarter, how many individual goals gave rise to a genuine checkpoint distinct from the annual review. The count can be done from calendars. It is the first thing we examine with Geneva-based executive teams, ahead of any work on decision-making. To go further: explore the Decision-making training in Geneva, or browse our executive and employee training programmes in Switzerland.