Apps pressure delivery riders into courting danger – here’s what needs to change
- Written by The Conversation
Picture this: you’re competing in a time-trial cycling race along a route that’s not known in advance. Instead of following a marked course, you receive instructions via notifications on your mobile phone.
Looking at your phone while cycling is extremely dangerous. But to stay on track, you must consult it nearly continuously.
If such a race took place on the streets of a busy, car-oriented city like Sydney, you would likely opt out. Yet food-delivery riders face precisely this situation every day: they receive order notifications while riding, and if they don’t check them, they lose the order and their hourly earnings suffer.
This is just one example of the dangerous incentive structure under which riders operate. These incentives are a central focus of our study just published in Transportation Research Interdisciplinary Perspectives, based on in-depth interviews with ten food-delivery riders in Sydney.
Dying while delivering
Delivery rider safety is an urgent concern. In New South Wales, serious injuries involving food-delivery riders increased from just 2 in 2017 to 75 in 2020. In Victoria, data from the past five years has shown that one in 20 people who went to hospital with e-bike accident injuries said they were working at the time.
According to the Transport Workers Union, by 2024 at least 18 riders had tragically lost their lives while working in Australia.
Previous studies have mostly addressed individual factors that make some riders more prone to risky behaviour. In contrast, our study examines platform decisions and operations. This includes how orders are assigned, what information is transmitted and when, and how payment is structured.
What did our study find?
Our findings can be summarised in six platform-related issues that systematically generate time pressure and encourage risk-taking.
Several orders at once: the handling of multiple orders, often with opaque delivery sequences, creates conflicts with customers and forces riders to rush. As one rider explained:
When handling multiple orders, I’m often directed to deliver the second order first. This leaves the first customer waiting for a long time, causing a big delay and making me take risks because I’m getting messages.
No control over restaurant timing: unpredictable food preparation times reduce riders’ hourly earnings, incentivising them to “make up” lost time on the road.
Fierce competition: intense competition among riders requires immediate acceptance of orders, even while cycling. One rider said:
I have to compete with other riders for food orders. Because of this, I often engage in unsafe behaviours, such as checking my phone while riding my bike.
Gamified risks: incentive systems structured around tight time windows turn work into a target-driven “game”, where completing more deliveries unlocks bonuses. This encourages riders to push beyond safe limits in order to secure the reward.
Star ratings: customer review systems tie riders’ future access to jobs to punctuality and perceived service quality, amplifying anxiety about delays.
Ordering ahead: Pre-orders (scheduled deliveries) create stricter expectations on timing without reducing uncertainty in preparation. When delays occur upstream – especially at restaurants – the responsibility is effectively shifted onto riders. As one rider said:
Waiting time at restaurants reduces my hourly rate. To make up for it, I tend to rush more.
Taken together, these mechanisms actively structure the conditions under which risky cycling becomes an economically rational response for delivery riders.
It doesn’t have to be this way
Regulating these platforms has proven difficult worldwide. Our findings identify some obvious and less obvious measures to consider:
Prohibit sending new orders to riders on the road
Platforms know which riders are currently serving an order. Sending a new one prompts them to check their phone while cycling. On the other hand, preventing this would likely reduce order bundling – one of the main things that makes deliveries so efficient. But the apps could build a feature where, based on settings the rider selects beforehand, orders are assigned automatically.
Provide more transparent information to customers
If customers could see when a rider is handling multiple orders and how long preparation times are expected to be, they’d be less likely to criticise the riders or penalise them with low ratings.
Redesign incentive and rating systems
Incentive schemes and customer ratings should be redesigned so that riders are not penalised for delays beyond their control, such as restaurant preparation times. Similarly, bonuses should not reward completing a high number of deliveries within unrealistically short periods.
Introduce an effective maximum speed
Riders often run red lights or use footpaths to move faster and complete more deliveries. Platforms could calculate a reasonable return time to a “hotspot” with many restaurants in one place, and refrain from assigning new orders before that time. This would remove incentives for riders to rush.
Dangerous deliveries are cheaper
Implementing many of these measures is far from straightforward. For example, coming up with settings to automatically assign and accept orders would require negotiations between platforms and riders.
An effective maximum speed could be resisted by both riders and the platforms: by delivering fewer orders per hour, the total revenue would decrease. This would reduce both the platforms’ profit and the riders’ earnings.
This last aspect points to one of the crucial tensions in the meal-delivery industry: the lack of safety is actually profitable. The faster the riders go, the better the business performs and the cheaper the meals become.
However, improving safety means slowing deliveries down and thus reducing the income. Compensating for those earnings losses in a job that’s already precarious would likely require higher delivery prices.
In other words, the low price of a delivered meal partly depends on riders taking risks. Ultimately, we all have to ask ourselves: are we willing to pay the true cost of safe meal delivery?
Acknowledgements: this study was led by Minjun Song, School of Civil Engineering, University of Sydney.
