There is an old cartoon of a donkey pulling a cart, with a carrot dangling from a stick just in front of its nose. The donkey walks and walks, forever chasing a carrot it will never reach. It is a funny image, until you realise that it describes the working lives of many people who deliver our food, drive us home, and bring our groceries to the door.
Gig platforms love bonuses. Complete 60 trips this week and earn an extra amount. Hit a high rating and unlock priority access to the best orders. These rewards, set and tracked by algorithms, are meant to keep workers motivated and loyal. But a study presented recently shows that when the carrot is placed too far away, it does not just fail to motivate. It actively pushes workers out the door.
The revolving door of gig work
Gig work is enormous. The World Bank has identified more than 500 gig platforms across 186 countries, accounting for about 12 percent of the global labour market. Yet these platforms have a serious retention problem. On average, only about 30 percent of gig workers in China and OECD countries are still on a platform a year after joining. In the United States, the figure is closer to 10 percent.
The authors use a helpful idea called episodic loyalty. Gig workers rarely see a platform as a lifelong employer. Instead, they treat it as a short-term career partner, joining, leaving, and returning as their lives change. The goal for platforms is to make each episode last longer and to make workers want to come back.
To do this, platforms rely heavily on algorithmic rewards, which are bonuses tied to performance targets that the app sets and monitors automatically. Some are cash. Others are non-monetary perks, such as priority access to choose the most desirable orders, a feature offered by some grocery delivery platforms. Many workers appreciate these rewards. But there are also plenty of stories of workers chasing targets that move just out of reach, ending the week exhausted, frustrated, and no richer.
What makes people want to stay
To understand why, the authors turn to self-determination theory, one of psychology’s best-known explanations of motivation. It says that people thrive when three basic needs are met. This study focuses on two of them. Autonomy is the feeling that you are in control of your own choices. Competence is the feeling that you are capable and effective at what you do.
Crucially, the theory also recognises that these needs can be actively frustrated, not just left unsatisfied. Feeling that you are not in control, or that you are set up to fail, is not merely neutral. It is corrosive. A reward system can therefore have a double effect. Done well, it can make workers feel capable and in charge of their earnings. Done badly, it can make them feel helpless and incompetent.
The UrbanRun experiment
To test this, the researchers recruited 273 real gig workers in the United States through an online research platform. These participants worked across 23 real gig platforms in ridesharing, delivery, and home services, and they had substantial experience, working nearly four hours a day on average.
Each participant read a scenario about a fictional driving platform called UrbanRun, which was launching an “Elite Driver Rewards Program.” The researchers varied three elements. Some versions offered a cash bonus and others did not. Some offered a priority bonus giving access to better jobs and others did not. And most importantly, in some versions the target to qualify was reasonable, while in others it was very hard to achieve. Participants were randomly assigned to one of the eight combinations and then asked about their feelings and their intention to keep working on the platform.
What they found
The first finding confirms the importance of basic psychological needs. Workers who felt their autonomy and competence were satisfied were much more likely to say they would continue on the platform. Feeling incompetent, on the other hand, made them more likely to want to leave.
The second finding is the heart of the paper. Achievability turned out to be the single most powerful factor. When the target was reachable, cash and priority bonuses had positive effects on workers’ sense of autonomy and competence. For example, a cash bonus noticeably boosted workers’ feeling of autonomy, but only when the target was achievable. When the target was hard to reach, the bonuses lost their magic. Worse, an unattainable target actively frustrated workers’ sense of competence. It is one thing not to get a reward; it is another to feel that the system was designed to make you fail.
In other words, the same bonus that motivates when it is within reach becomes demoralising when it is not. The size of the carrot matters far less than whether the donkey has any real chance of eating it.
Lessons for platforms and managers
The authors argue that rewards should not be treated as a single, simple lever. Each component, whether the type of bonus or the difficulty of the target, has its own effect on how workers feel, and those effects interact. A generous bonus with an impossible target may be worse than a modest bonus with a fair one.
For gig platforms struggling with turnover, the message is clear. Algorithmic incentive systems should be designed with achievable targets that make workers feel capable and in control. Setting targets so high that only a tiny fraction of workers can reach them might look like a clever way to save money on bonuses. But it risks burning through the workforce, which is far more expensive in the long run.
The lesson also applies well beyond the gig economy. Sales teams with unrealistic quotas, students facing impossible grading curves, and employees chasing performance metrics that keep rising all experience the same dynamic. Motivation thrives on challenge, but it dies on futility.
The authors acknowledge some limitations. The study used a hypothetical scenario rather than tracking real workers on a real platform, and loyalty was measured through stated intentions. Future research using field data could capture the richer relational and cultural sides of loyalty.
Still, the core message is one that any manager, human or algorithmic, would do well to remember. If you want people to keep pulling the cart, put the carrot where they can reach it.
Reference
Huang, X., & Nguyen, L. T. (2025). Beyond reach: Algorithmic rewards’ effects on gig workers’ episodic loyalty. In ICIS 2025 Proceedings (Paper 3). Association for Information Systems.
