The Impact of Supply Learning on Customer Demand: Model and Estimation Methodology

by Nathan Craig, Nicole DeHoratius & Ananth Raman

Overview — "Supply learning" is the process by which customers predict a company's ability to fulfill product orders in the future using information about how well the company fulfilled orders in the past. A new paper investigates how and whether a customer's assumptions about future supplier performance will affect the likelihood that the customer will order from that supplier in the future. Research, based on data from apparel manufacturer Hugo Boss, was conducted by Nathan Craig and Ananth Raman of Harvard Business School, and Nicole DeHoratius of the University of Portland. Key concepts include:

Two key measures of supplier performance include "consistency", which is the likelihood that a company will continue to keep items in stock and meet demand, and "recovery", which is the likelihood that a company will deliver on time in spite of past stock-outs.
Improvements in consistency and recovery are associated with increases in orders from retail customers.
Increasing the level of service may lead to an increase in orders, even when the service level is already nearly perfect.

Author Abstract

To set service levels, firms must understand how changes in service affect customer demand. Supply learning is a process whereby customers use past supplier performance to build beliefs about supplier capabilities and hence about future supplier performance. This paper presents a multi-period model of service level competition among suppliers selling substitutable products to a customer that engages in supply learning. We observe how a supplier's service level performance molds a customer's beliefs as well as how a customer's beliefs affect its order quantities. We identify two dimensions of supplier performance: consistency, the probability that a supplier delivers in the current period conditional on availability in the prior period, and recovery, the probability that a supplier delivers in the current period conditional on a stockout in the prior period. We also provide a method for estimating the impact of changes in supplier performance along these two dimensions on customer demand. Using data from Hugo Boss, a manufacturer of branded apparel, we find increases in consistency and recovery to be associated with increases in orders from Hugo Boss's retailer customers.