Productivity is unmeasurable in white-collar workplaces

The speaker's central claim: modern corporate "productivity" for knowledge work cannot be directly measured, so organizations use proxies (activity, hours, quotas) that distort incentives and behaviour.

He contrasts manufacturing—where output-per-hour has been measured for "150 years" and cites "Ford Motorcycle Company"—with finance, product management, software engineering and executive roles, where output is qualitative and delayed. The company-level objective is "profitability," and productivity metrics are only proxies for profitability.

Three concrete examples illustrate the mismatch. Sales: quotas are used as the metric; hitting 103%–105% often maximizes individual pay, while sustained overperformance (speaker cites Google policy: exceeding "107%") triggers quota increases that remove upside. Finance: two employees, "Sally" (11-hour days, visibly busy) and "Barbara" (automates work with "Power BI", automates "95% of her job" and completes tasks in 15 minutes) show how activity-based monitoring rewards visible busyness; Barbara's automation threatens headcount and may lead to her being re-tasked or let go despite higher true productivity. Software engineering: "Bob" (fast, buggy code) versus "Frank" (slower, maintainable code) shows short-term activity metrics favour Bob while long-term company health favours Frank.

The practical takeaways the speaker offers: recognize the game being played; check which proxies your employer values and either "fill the proxies" to survive/payraise, or accept the risks of optimizing true productivity (automation, cleaner engineering) because organizations often cannot measure or reward it. He notes compensation design is hard and typically rewards the proxy managers can observe. The speaker personally prefers building and fixing things over gaming metrics but warns that career outcomes depend on company culture and managerial intelligence.

Choose to play the proxy game or not—with full awareness that companies prioritize profitability and observable activity over invisible, hard-to-measure output.