The productivity metric that became a target in public

In 2023 McKinsey published Yes, you can measure software developer productivity, and the practitioner community produced its most unified response in years, led by Kent Beck and Gergely Orosz's rebuttal, which walked through what happens to each proposed metric the day it becomes a target. Three years later the exchange reads as the clearest public case study we have of the industry's standing argument with itself.

Strip the heat away and each side holds a true thing. McKinsey's true thing: leadership cannot steer what it cannot see, and "trust us, it is unmeasurable" has been used to shield genuinely dysfunctional organizations. The rebuttal's true thing: every individual-level activity metric on the list had a known failure mode, most of them documented for decades, and the piece proposed them anyway. Measurement is possible. The proposed measurements were the problem.

The synthesis we use in engagements is boring and holds up: measure systems at the team level and above, use the measures to find friction rather than to rank humans, and treat any metric that has begun rewarding behavior as expired. Expiry is normal. A metric is a flashlight, and flashlights get pointed somewhere new once a corner is lit.

If you are relitigating this argument inside your organization this quarter, our advice is to make both camps read both pieces, then ask one question: what decision would this metric change, and who is allowed to see it. The answers sort most disputes faster than the philosophy does.