How food delivery platforms are speeding up deliveries in pursuit of unit profitability
Episode 25 Quick Summary November 2021
Upward pressure on food delivery courier costs, from regulation and market shortage, is heightened by delivery platforms competing to build rider and driver (courier) capacity
Technology and logistics innovation has yielded gains in unit courier productivity, but gains are now becoming more marginal
Broadening supply into lunch, grocery and convenience is increasing courier productivity, but only where significant market share can be held
Delivery platforms will need to integrate more logistics and in-restaurant technology to control time leakage and maintain customer satisfaction
Full speeding up delivery article available to paid Aggregators subscribers.
On a quiet Monday lunchtime on the edge of Cambridge city centre, a Just Eat rider bustled in and out of Subway, pausing just long enough to tell me he was expecting to do 2 deliveries per hour throughout the afternoon. A year ago I wouldn’t have thought this run rate possible outside of evening peak times. It seems that the delivery platforms are able to use their couriers more and more fully, through more and more portions of the day. And it’s not just in the sweet spots of university towns. I’ve seen the same levels of afternoon activity in McDonalds in the suburbs of places like Hemel Hempstead and Oxford.
However, the combination of regulatory and market pressure on courier costs (see Aggregators Episode 24, ‘Online takeaway sales growth decelerating, whilst costs are finely poised’) means that productivity is going to have to keep improving in order for courier costs per order not to rise as the food delivery platforms expand out further and further from their big city roots. Even after over six years of using data science and tracking technology to optimise their logistics - inspired by the groundbreaking routing and batching algorithms pioneered even earlier by grocery pickup and delivery service Instacart.
Wolt reduces delivery times by 26% with order stacking
Amongst the various optimisation methods, order ‘stacking’ (or batching or pooling) is key. Wolt, the Finnish delivery logistics player acquired by DoorDash, reported, as part of the acquisition presentation, September metrics of 35 minutes per delivered order (customer wait time), and 26 minutes per courier delivery…implying that 9 minutes, or 26% of delivery time, is ‘shared’ between different orders because of stacking - a significant efficiency benefit.
DoorDash have made stacking explicit for consumers with their (patented) ‘double dash’ offer for a grocery order to be added to the takeaway order. Instacart goes even further: its grocery customers can choose to accept their order being ‘batched’ with another customer’s. Building on this is the ‘food hall’ model where the consumer chooses across a range of clustered outlets. With such mingled orders, though, there are risks of late orders. Just Eat Takeaway has been working on stacking for two years without finding a method that doesn’t compromise the customer experience. Even without stacking, though, grocery and lunch orders fill out afternoon capacity, so couriers can operate at 2 drops per hour for longer.
There are other, more technological ways that delivery platforms are squeezing more out of their driver capacity. ‘Routing’ is key among these, with Instacart an early exponent in grocery. By continuously correcting actual against predicted distances travelled, Instacart optimises routes to reduce the distance travelled by its ‘shoppers’ in multi-order trips by 9%. Complementing routing are ‘delayed dispatch’, where orders are allocated to the courier at the last possible moment, allowing more time to choose the optimal courier; and ‘predictive driver repositioning’ instructing the courier where to go to await the next order.
These methods complement each other, but with diminishing returns as they are combined together. Open Door Logistics ran simulations in Paris during 2018 in which the combination of stacking, routing, predictive positioning and delayed dispatch took down order delivery time from 29 minutes to 21 minutes. Most of the time reduction was from stacking alone: a 21% cut - similar to what Wolt appear to achieve with stacking. Full analysis in the speeding up delivery article for paid Aggregators subscribers.
Underpinning these logistics efficiencies are some foundational technologies. Courier apps make the process of a courier accepting orders more seamless. Driver tracking underpins the effort to orchestrate where and when the couriers are deployed. Uber Eats pioneered the use of motion data as well as location data, to track what the courier is doing (waiting? riding?) as well as where they are. Other technological enablers are provided by third parties, for example routing depends on third-party API like Mapbox, and data lakes like Databricks enable superfast iterations. In the pursuit of lower unit costs, the big aggregator platforms need to invest to integrate with the burgeoning open system of third party ordering, fulfilment and analytics technologies (see Aggregators Episode 23, ‘Restaurant ordering technology is consolidating’)
DoorDash appears to have been holding unit courier cost pressures at bay for three quarters
How have the delivery platforms fared in deploying efficiencies to offset pressures on courier pay? For most, we await year-end figures, but DoorDash for one appears to have been holding unit courier cost pressures at bay for three quarters so far in 2021. Just Eat Takeaway have referred to falling order fulfilment costs in the second half of the year, but quantification has not yet been provided. Uber, by my estimates, had significantly higher unit courier costs in Q3 than H1, but much of that has come from the incentives required to reactivate Uber Rides drivers as economies reopened.
A key question that remains is whether routing and tracking technology, together with data science, can continue to enable delivery speed gains. Is there an equivalent of Moore’s Law whereby technology yields more and more increases in delivery speeds? If not, platforms and technologies could pursue other efficiencies by continuing to merge together. Or some competitors will evolve their business models to offer more profitable services to the consumers and restaurants that they have managed to hold on to. Before that, Aggregators will look at what 2021’s final round of company reporting tells us about unit courier costs.
Book a consultation with me: email thephilbranston@gmail.com, or schedule through these expert networks: