What gets measured and watched gets better, mostly

Bob Parsons, who founded the domain registrar Go Daddy, kept a list of sixteen rules for business and life and published it on his own site. The ninth rule is headed “Measure everything of significance”, and the explanation under it is this one sentence, prefaced by his assurance that it is true. The full list pairs it with a rule that says the opposite of the same idea: whatever is left unmanaged declines.

The sentence has two verbs, and the second one does most of the work. Measuring alone produces a number in a spreadsheet that nobody opens. Watching means someone looks at the number on a regular schedule, compares it with last week, and asks why it moved. That loop turns a vague intention into a concrete question with a deadline. A support team that starts posting its average time to first reply on a wall every morning will usually see it fall within weeks, often without any new tools or staff, because people now know what is expected and can see whether they are delivering it.

Watching also needs an owner. A number that belongs to everyone tends to be looked at by nobody, while a number with one person’s name next to it gets checked, explained and defended. Small companies often find that assigning each important figure to a single person does more than any dashboard.

The same effect shows up in manufacturing defect rates and in website performance. Once page load time is charted daily, engineers notice when a release makes it worse, and it stops creeping up. Visibility also settles arguments. Two people who disagree about whether a change helped can look at the same line on the same chart.

The trouble is that a number can improve while the thing it was meant to represent gets worse. The British economist Charles Goodhart made the point about monetary targets: once a measure is used to control behavior, people learn to move the measure. Call centers that track average call length find agents hurrying customers off the line or transferring them, and the callbacks show up somewhere else. News sites that counted page views learned to split articles into slideshows of twenty pages each, which pleased the chart and irritated the reader. In both cases the metric was measured and watched, and it did improve.

A second failure is choosing what is easy to count over what matters. Sign-ups are simple to measure. Whether those people are still using the product three months later takes more work, and a team rewarded on the first number may quietly neglect the second.

A related line circulates as “Pearson’s law”, credited to the statistician Karl Pearson on thin evidence. Parsons’s version at least has a traceable author, and his list leaves no doubt about what he meant.

The practical reading keeps the rule and adds a check. Measure the thing that matters, watch it often enough to act, and pair every target with a second number that would reveal gaming: call length with repeat calls, sign-ups with retention. A metric watched alone improves. A metric watched alongside its counterweight is more likely to mean something.