Analytics Without Upside Theories Fizzle
Don’t use analytics in distribution to grind existing information finer and faster. You will get interesting, but non-actionable data. Start instead with an improvement theory. Then, build an analytical model to validate an improvement theory for your distribution. You will find insights to exploit and can track subsequent change experiments with new metrics.
Distributors, for example, have a mix of very profitable and unprofitable items, picks, orders, and customers hiding within averaged-out, aggregate financial numbers. Instead, create a cost to serve (CTS) model to expose the big profit cross-subsidies and then pursue a new metric like: make 100% of customers profitable.
Continue reading 50. Analytics in Distribution, Theories, Bias Inefficiencies and Courage →