Ashley Stirrup

People with smaller palms live longer. That's a real correlation. Statistically significant. Backed by data. The good news is, you don't need to bandage your hands! The real variable is gender. Women have smaller palms on average and live about five years longer in the US. Palm size is just a proxy. It's riding along with the thing that actually matters. Ron Kohavi uses this example to set up something more uncomfortable: your product dashboards are full of palm-size correlations. And your team is presenting them at all-hands. This happened at Microsoft. "New users who use our new feature are half as likely to churn compared to new users that do not use the feature in the next 30 days." Of course. Heavy users also see more errors. They also stay longer, because they're power users who rely on the product. Error rate is like palm size. The actual variable is engagement. The trap isn't bad data. It's the assumption that correlation tells you what to do. Every dashboard metric is a proxy for something else. The question isn't whether the correlation exists. It is whether you're measuring the real variable or the palm size. The only way to answer that: run a controlled experiment. Change one thing. Measure the difference. Don't let a surrogate make your roadmap decisions. What's the most confidently presented correlation you've seen at your company that probably wasn't causal?