Share:

Campbell’s Law says that the more a metric is used, the more likely it is to corrupt the process it is intended to monitor. Is that true?

I once went for a motorbike ride with a new friend. She was experienced, and handled her very big motorcycle very capably. But on this ride, she crashed. The reason is commonly known as ‘target fixation’. She fixated on the ditch to the outside of a corner she was leaning into, and into that ditch is where the motorbike went.

Target fixation happens in performance measurement too (it’s related to what Jerry Muller describes as metric fixation). We focus too much on the metric or measure or KPI, without context. And that’s likely the problem that this law of measuring performance warns about:

The more a metric is used, the more likely it is to corrupt the process it is intended to monitor. — Campbell’s Law

Campbell’s Law will certainly hold true when a metric is used in a vacuum, used with a blinkered focus. It’s like those typical project management metrics that track on-time completion of milestones and on-budget completion of tasks. Just focus on those, and we end up corrupting the very purpose we designed the project to serve in the first place.

How much we use a metric is not exactly the problem…

But how much we use a metric is not the reason why metrics corrupt the process they exist to monitor. Metrics corrupt the process they monitor when one or more of three things happen:

1. They are measuring the wrong results about the process, and lead to actions that undermine the process performance.
2. They are measured too frequently, and encourage over-reaction to changes in the numbers.
3. They are measured in isolation, and are not balanced with other measures and information that collectively give the big picture.

Every measure needs to be designed or chosen as evidence of the results that matter most about the process we’re monitoring.

Convenient or traditional measures can corrupt process performance when they don’t match the process’s purpose. For example, if the purpose of a customer support process is to help solve the customers’ problems, then measuring call time will lead to actions that don’t fix the customers’ problems first time.

Every measure has a cadence that should be set to suit how quickly or slowly the performance result it measures can respond to change.

Some metrics do need to be used daily or weekly to keep track of fast-changing performance results. For example, water quality, as monitored by local municipalities, can change daily and needs to be measured at least that often to pick up signals of change. But measuring water consumption every day might drive us to unnecessarily react to fluctuations that just don’t matter in the bigger picture.

Every measure is part of a bigger picture of what’s really going on with a process.

No measure can provide enough information in isolation. It’s the relationships it has with other measures that give us a fuller understanding of how a process is performing. For example, in this PuMP Results Map, it’s easy to see these relationships for each process, like Design, or Sales, or Marketing. It helps us see the cluster of measures that need to be used in concert with each other, to understand the bigger picture of each process’s performance.

***

Rather than taking Campbell’s Law at face value, it’s possibly more useful to say:

The more a metric is used, when it’s used in a system of related metrics and monitored with a cadence that suits how quickly things can change, the faster and more accurately we can improve the process it monitors.

[By the way, my friend was not badly injured in the crash. She got the motorcycle repaired, and continues to ride. But with less risk of target fixation.]

Using a metric often is not the reason it corrupts a process. The reason a metric corrupts a process is when it is used in isolation of related measures. [tweet this]