We collaborate with a retailer operating 69 stores in Australia and New Zealand. The retailer employs salesworkers whose main task is to sell the firm's products in the stores. In an RCT, two (new) compensation systems are implemented in (about) half of the firm's stores, respectively, and the company has agreed to keep the treatments in the field for at least four months (i.e., until the end of May). The first system is solely based on workers' individual performance (i.e., their revenue). In the second system, revenues of different types of products are assigned different weights, and commission payments are based on weighted revenue. Both compensation schemes are piecewise linear; they have kinks at two pre-specified targets and become steeper once a target is reached. The commission systems are designed such that the average commission is about the same across systems (as calibrated with the 2019 revenue distribution).
Treatment A: Salesworkers' monthly compensation depends on their (unweighted) revenue in that month.
Treatment B: Salesworkers' monthly compensation depends on their weighted revenue in that month. Different weights are assigned to revenues of different types of products, with high-margin products receiving higher weights than low-margin products.
We study how the different compensation systems affect revenues from different types of product, turnover, conversion rates (i.e., the rate at which customers being served by salesworkers buy the firm's products), and granted discounts. We also study heterogeneous treatment effects with respect to workers’ pre-treatment performance (high- vs. low-performers) and tenure. We further conduct employee surveys to measure employee attitudes.