Menu Close

Using big data for smarter online supermarket shopping

What if you could save on your online shop by switching delivery times? Blue Square Thing

The online grocery market has been growing in the double digits in the UK and is on the rise all over the western world. Retailers are fighting to get their share of the business and spend big on launching delivery services. But making them run well is a serious challenge.

Profit margins are very tight in grocery delivery and the competition is fierce. Keeping costs down is an absolute must and failed deliveries are a big problem. But customers increasingly expect the convenience of narrow delivery time slots and can easily switch to buy their weekly shop from a rival if you don’t keep them happy.

A team I’m part of from the universities of Warwick, Lancaster and Southampton think that customers can be encouraged to choose specific delivery time slots if they are given access to real-time information about their order. That means retailers can make fewer trips to deliver their goods, saving them money and time and even helping the environment.

Choose your slot wisely

If, for example, two customers have booked a delivery within the same time slot but live at opposite sides of a town, the retailer would need to send out two vans. But if the retailer can convince one of those customers to pick a different time slot, both deliveries could be done with one vehicle.

One option might be to inform the customer that another delivery is being made in their area at a certain time and that they could get their shopping at the same time, which would save on emissions.

Perhaps even better, they could be offered discounts or rewards for changing delivery times. They could get cheaper delivery times by filling unpopular slots too.

Supermarkets are well known for gathering vast amounts of useful information about customer habits and their own processes. Every click on a website can be recorded and used to inform decisions. This research is based on extensive data on shopping and the fleet used by a major British retailer.

Such data allows us to look at time slot selection behaviour, to see which slots are most popular and how availability of alternative slots and incentives influences their popularity. It can also show us which slots are shown as available when the customer makes a decision, which means we can draw conclusions about the relative attractiveness of different slots. It might be obvious that a customer is likely to choose a 6pm slot to time a delivery with their arrival home from work but which slots do the consistently ignore? And which is their second choice if the coveted 6pm slot is taken?

The methodology also estimates the expected cost to the retailer making a requested delivery in a given time slot. If only one delivery is to be made that day, the cost of that delivery will be much higher than if several are due. This estimation in itself is a challenge since the actual delivery cost only becomes known once all orders have been collected for a specific delivery day and no further orders are accepted.

But if the retailer can better understand customer habits, it can make better decisions about cost estimation. Making even small savings can make a big difference in this cutthroat market.

Online delivery is a logistics monster that needs to be tamed. As more of our purchases are made via the Internet, more deliveries are going to be needed. We’ve already seen a rapid transformation in the way the system works, which shows us how serious retailers are about winning over customers. Where once you might have to pay a high price for an online delivery, wait days or even weeks for it to arrive and even sit at home all day in case it comes, you can now have your shopping exactly when you want it. Major players such as Amazon are heading towards same day delivery so the pressure is on for others to follow suit. They will find it a lot easier if they make use of data like this.

Want to write?

Write an article and join a growing community of more than 126,300 academics and researchers from 4,009 institutions.

Register now