Businesses use checklists to exercise managerial control and ensure quality, but excessive bureaucracy can frustrate skilled workers and reduce morale. This column presents a randomised control trial in a German bakery chain that removed two time-consuming, non-essential checklists in selected stores. After abolishing the checklists, sales rose by 2.7%, customer ratings improved, and qualified workers – especially store managers – were less likely to leave. The regional managers accurately predicted the outcomes, suggesting that empowering middle managers with discretion over control mechanisms could boost both performance and wellbeing.
Reducing bureaucracy is a perennial issue in public administrations around the world. Businesses are also not immune to implementing excess bureaucracy – and not all of it is owing to government regulations. Rather, it’s due to managers trying to exercise control over the workforce. In our new paper (Friebel et al. 2025), we look at what happens when a bakery chain reduces bureaucracy, or more concretely, control by checklists.
Checklists are used in many industries and administrations. Pilots use them to ensure that their airplane is fully functioning and safe; surgeons use them to help maintain hygiene in operating theatres. By ensuring that important work steps in a process are taken care of, checklists help in avoiding tail risks.
Checklists are, however, used not only for life-and-death matters but also for more trivial purposes, for instance, to maintain the same quality standards for burgers, sandwiches, or cakes. And the use of checklists is no free lunch. People do not necessarily like having to check too many boxes. They may find filling out checklists a time-consuming and wasteful activity. They may even perceive the use of checklists as annoying or insulting.
This is what the management of a high-quality bakery chain in Germany learned when it carried out an employee engagement survey. Employees were proud of the high-quality products and praised punctual wage payments, the camaraderie in sales teams, and good relationships with clients. Many, though, complained about having to fill out daily checklists covering many aspects of their work: to verify that the same number of strawberries are put on each cake, that fresh coffee is put into the coffee machine after each filtering process, or that doughnuts are handled in a way so as not to scratch sugar from their top. The employees explained that, after many years of training, they know quite well how to handle these processes.
The managers of the firm reached out to us because they were concerned about this feedback at a time when attrition of qualified workers had increased. They were aware of our previous research in bakeries and retail (Friebel et al. 2017, 2023). Could removing some checklists and the intensity of monitoring have positive effects? To answer this question, we designed an experiment to measure the effects of removing two of the checklists deemed by employees and managers to be relatively time-consuming and not crucial for the success of the firm.
The main idea: Workplace control, signals of (dis)trust and respect
Management science as proposed by Taylor and Fayol provides the theoretical base of workplace control and detailed training. These ideas were brought to practice by Ford and others, leading to enormous productivity gains. More recently, the literature on empirical management pioneered by Bloom et al. (2010) has shown that management practices such as clear targets and rewards have positive effects on productivity. This has been supported by a number of randomised controlled trials in different circumstances, industries, and countries.
But what if workplace control had negative effects on people’s behaviour? Behavioural game theory provides some theoretical answers to this. Ellingsen and Johannesson (2007, 2008) argue that workers care about what other people – say, their bosses – believe about them. This is reasonable; we all care about our reputation as good co-workers, as being professional or diligent.
Consider now that a boss can decide on how intensively they monitor a worker. In this context, a more controlling boss sends the signal that their belief in the worker being a ‘good’ worker is weak, while a less controlling boss signals a stronger belief in the worker being good. If the worker’s effort depends on receiving the signal that the boss believes them to be a good worker, then too much control might negatively affect the worker’s behaviour.
Putting the idea into practice: The randomised control trial
The theory of Ellingsen and Johannesson can be readily adapted to the work process of the bakery chain (and other industries, like restaurants, retail, or repair shops). Checklists provide the benefit of avoiding costly mistakes. They are useful when the workforce is not very qualified and hence more likely to make mistakes. They will, however, be less useful when the workforce is qualified, not only because the workers do not need the checklist to do a good job, but, more importantly, because they interpret the control by checklist as a signal of disrespect. This could lead to less effort or the decision to leave.
We designed a randomised controlled trial to check on checklists. First, members of the firm’s top management discussed the usefulness of more than 20 checklists that the firm used to steer its sales operations. We then carried out a survey among store managers and employees and chose two checklists to discontinue that were deemed relatively time-consuming and not very useful.
We randomly assigned all stores to treatment and control groups, in a way very similar to clinical studies in medicine and, increasingly, in economics (Duflo et al. 2017). The treatment group received a message from the chief operations officer that the two checklists would be removed. The message also expressed that the firm’s management trusted its workforce. We carried out surveys at different points in time and different levels in the firm hierarchy and monitored that, at any point in time, all managers in the firm were following the experiment protocol.
Findings
First, treatment stores had slightly larger sales (by around 2.7%), and in Google reviews, treatment stores received higher rankings. Second, and probably more interesting, is that the findings are in line with the theoretical idea that checklists may have quite different effects on different types of employees. In particular, in treatment stores, the attrition of qualified workers decreased significantly, especially among store managers. Workers without vocational training, though, had a larger (albeit statistically less significant) likelihood of leaving the firm. This is in line with the idea that control by checklists is perceived more negatively if people think they do not need them because of their qualifications, but it also shows that for less qualified workers, checklists may have the benefit of providing orientation in the workplace.
Finally, we found that regional managers were able to predict the treatment effects in the stores they supervised. In our research plan, we had asked these regional managers in which of the stores there might be benefits of removing checklists and why. These predictions turned out to be remarkably precise, even though at the time we asked, the stores had not yet been randomised into treatment and control groups and the managers had no incentive to influence the effects. The two panels in Figure 1 depict how precise managers were in their predictions.
Figure 1
Regional managers seem to be in a good position to decide where to control more and where to do less. But, as in many other retail firms, control systems have little flexibility.
Implications for public administrations and private firms
Profit-maximising firms are often more inclined to do a randomised control trial than public administrations. The experiment of Fornasari et al.’s (2025) with the Ghanaian civil service is a noteworthy exception. In the Ghanaian study, individual training led bureaucrats to share more ideas on how to render the organisations more effective, but trainings at the division level failed in doing so. Our findings complement theirs: in both cases, much of the dispersed information available in organisations is not used, in many ways reminiscent of Hayek’s insights. Upward communication is a difficult business in public and private organisations alike. Ideas may abound, but, for strategic or cultural reasons, the people who have them may not feel motivated or empowered to share them.
In our case, the middle managers appear to know best how employees will react to less workplace control. More discretion for middle managers in deciding how to lead their teams may improve profitability and worker well-being. Why companies and bureaucracies alike fail to do so and how this can be improved upon is a fascinating question for organisational research.
Source: VOXeu
