Feedback on Machine Utilization
How to cite
Schneider, Sebastian O.; Langer, Bernd; Sutter, Matthias; Alfitian, Jakob; Schaetze, Theresa. 2026. "Feedback on Machine Utilization." AEA RCT Registry. AEARCTR-0018407. https://doi.org/10.1257/rct.18407-1.0
Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.
* Primary investigator
Abstract
This study investigates how feedback on machine utilization affects the productivity of workers operating automated machinery in manufacturing. We conduct a randomized controlled trial across seven plants in six manufacturing firms, in which groups of workers within plants are randomly assigned to one of three experimental conditions. Workers in the control group continue their work as usual without any intervention. Workers in treatment arm 1 receive a weekly printout for each machine they operate, visualizing machine utilization throughout the workday based on data from the previous week, the entire experimental period to date, and a three-month baseline period prior to the intervention. Prior to the start of the intervention, workers in treatment arm 1 participate in a workshop in which they receive instructions on how to read and interpret the printout. Workers in treatment arm 2 receive the same weekly printout and participate in an extended kick-off workshop prior to the start of the intervention, which includes an in-depth analysis of potential problems and a structured discussion of possible solutions to improve machine utilization. Treatment arm 2 also includes weekly meetings between workers and their supervisors to review patterns in machine utilization and discuss possible solutions, documented in a meeting log. The intervention runs for 13 weeks.
Intervention
Control group:
Workers in the control group continue their work as usual. No intervention takes place.
Treatment arm 1:
Workers in the first treatment arm receive a weekly printout for each machine they operate. The printout is generated automatically from machine data using a standardized software routine designed by the research team, implemented by an external consultant, and run by the firms. At the beginning of each week, a designated staff member within each plant distributes the printouts to the relevant machines. The printouts are placed directly at the relevant machines, making them visible to the workers who operate them. Each printout contains three graphs, all displaying machine utilization over the course of a 24-hour workday. For every minute of the day, each graph shows the proportion of days on which the machine was running. The three graphs display this information for: (1) the previous week, (2) the entire experimental period to date, and (3) a three-month baseline period prior to the experiment. Prior to the start of the intervention, all workers in treatment arm 1 participate in a workshop in which they receive instructions on how to read and interpret the printout.
Treatment arm 2:
Workers in the second treatment arm receive the same weekly printout as treatment arm 1, distributed in the same way. In addition, prior to the start of the intervention, workers participate in an extended kick-off workshop led by a consultant with expertise in lean management. Compared to the workshop in treatment arm 1, this workshop goes beyond printout interpretation and includes an in-depth analysis of potential problems and a structured discussion of possible solutions to improve machine utilization. Moreover, throughout the 13-week intervention period, workers meet weekly with their direct supervisor to review the new printout, discuss patterns in machine utilization, and identify possible solutions. Following each weekly meeting, workers and supervisors are asked to record areas for improvement in a meeting log. The meeting log also contains a designated field in which supervisors record the duration of the meeting.
Experimental design
Experimental conditions are assigned at the block level using stratified randomization, where a block is the smallest organizational entity within a plant comprising a group of employees that has their own supervisor, corresponding to either a production department or a shift group. This level of assignment is deliberately chosen to minimize the risk of spillovers between experimental conditions within the same plant. Randomization is stratified at the plant level, ensuring that within each plant, blocks are randomly assigned to experimental conditions such that every plant contains at least one block in each of the three experimental conditions. The intervention runs for 13 weeks, from June 22 to September 18. Prior to the intervention, a baseline period of approximately three months is considered. At the end of the intervention, an endline survey is conducted among workers to capture their perceptions. The primary data source is machine data, which records for every machine and every minute whether the machine was running or experiencing downtime. In addition, a designated staff member within each plant provides weekly data on staffing and planned production times, which are used to construct the machine utilization measure.
Design details
Cited by
No citation data available for this registration yet.