The Measure That Becomes the Job

Measurements help organisations steer, but over time the indicator can quietly replace the purpose it was meant to represent.

Most organisations measure because they want to know whether something important is happening. Backlog stands in for maintenance control, on-time delivery for reliability, utilisation for productive use and customer scores for satisfaction. None of these measures is foolish. The difficulty begins when the relationship reverses and the organisation manages the number as though it were the thing itself, the pattern economists know as Goodhart’s law.

The reversal is usually gradual. A measure is introduced because the underlying result is hard to see directly. It proves useful, so it appears in reviews; because it appears in reviews, somebody becomes responsible for it; once responsibility exists, targets follow, and incentives, reputation and budgets begin to depend on them. At that point improving the measure is no longer a means of improving the result. It has become an objective in its own right.

No dishonesty is needed for this to cause harm, because much of the distortion occurs through defensible decisions. A maintenance team reduces backlog by redefining what belongs in it. A project improves schedule performance by changing the point at which an activity counts as complete. Inventory falls because stock moves to a contractor while the operational dependency remains. Each decision can be explained, and the measure genuinely improves while the underlying concern barely changes.

Asset availability shows how much context a single number carries. High availability sounds positive, but an asset may be available because maintenance has been deferred, because redundancy masks deterioration or because demand has fallen. Low availability may reflect a deliberate intervention that prevents a larger failure. Once the operating context is removed, the same percentage can support opposite conclusions.

The danger grows when the measure is easier to communicate than the result. A board understands a trend line in seconds, while the underlying condition may need context, judgement and an uncomfortable discussion of uncertainty. The cleaner the metric, the more authority it acquires, until the person explaining why the number misleads sounds defensive and the person presenting it sounds disciplined. The same drift occurs at strategic level, as growth becomes revenue growth, innovation becomes a project count and safety becomes incident frequency.

The answer is not to stop measuring, since without measurement organisations substitute anecdote and confidence for evidence. It is to keep the indicator connected to its purpose. One protection is an explicit expiry test: not an expiry date, but a recurring challenge. Does this number still represent what we think it represents? What behaviour has grown around it? What would improve the number without improving the result? What important change could occur while the number stayed the same? A second protection is to keep more than one perspective visible where the result is inherently multidimensional: cost with reliability, output with quality, schedule with risk. The aim is not to measure everything, but to stop one convenient number from becoming a complete description of something that was never one-dimensional.

Measures are valuable because they compress reality, and dangerous for the same reason. The detail removed at the start may later be the part that matters most. Once people protect the instrument more carefully than the result it was built to reveal, the measurement system has stopped helping the organisation see and started telling it where to look.