[throwawaysea]

This post is correct about two things - first, that NPS does not give you direction on how to act. Second, that statistical realities make sampling approaches like NPS very noisy. I don’t think either discounts the concept of NPS itself. “Bad statistical work” is going to be bad regardless of what metric we’re talking about. And a metrics being too abstract to action doesn’t make it useless - at least not any more so than something like free cash flow. Yes, connecting the dots to some outcome is difficult. But it’s a pulse on the business, and if done right, that has some inherent value.