Spar owner to use AI to guess what customers want to buy

BWG, the company behind Spar in Ireland, is investing €6.5 million in new smart technology that anticipates customer behaviour based on factors such as local weather, sporting events and even social media activity. It then orders in stock automatically for stores.

The artificial intelligence-based predictive stock-ordering solution aims to reduce the amount of time shop owners spend compiling orders, while also allowing them to benefit from additional purchases on items they might not know would be in demand.

The convenience retail group, which also owns Mace, Londis, XL and the larger Eurospar format, serves more than a million shoppers every day in the Republic. It estimates an individual is never more than 10km away from one of its stores in Ireland.

The company says it expects both store owners and consumers will benefit from its new solution, which is being trialled in 22 stores currently, but is set to be rolled out across BWG’s 1,000-plus network by May.

Time to check your tech habits: What is working and what is hindering you?

---

LetsGetChecked to create 150 jobs as firm aims to expand global reach

---

Eir chief executive Lennon stepping down to be replaced by Oliver Loomes

---

Viatel snaps up Limerick-based ActionPoint

---

Traditional ordering platforms rely on past data, creating a replenishment-based order by analysing what has sold, taking into account safety stock in order to predict what retailers should purchase.

The new solution uses predictive analytics to make real-time forecasts for retailers based on what consumers will likely wish to purchase in the future. It is an add-on to the company’s existing ShopLink ordering system, which is fully integrated with retailers’ EPOS systems.

Sales

The predictions are based on 240 different data points, including local demographics. It could, for example, confidently predict additional demand for low or no alcohol during “Dry January”, an uplift in sales of convenience food when the World Cup is on, or the need for additional stocks of essentials such as bread and milk in times of bad weather.

Speaking to The Irish Times, Chris Donnelly, chief information officer at BWG, said the retailer's wide reach means that its system can even predict ultra-local activities – such as a local GAA derby – that could impact on buying behaviour.

“The delivery we will be sending out to a store is not a replenishment, but rather a forecast of what customers will want to see in the store next week. It can pick up on products that aren’t currently being sold or for those that are selling quickly but for which retailers haven’t anticipated they could sell a lot more of,” said Mr Donnelly.

He predicted that the solution would only improve over time based on the volumes of data it will have. Mr Donnelly said BWG is essentially bringing the types of technological advances that have made online shopping so easy to traditional bricks and mortar stores.

“, especially as they still get a chance to override it to include items in orders when they want to. They particularly love how time-consuming this is because rather than spend ages putting together an order they can literally press a button and it will be automatically compiled for them,” Mr Donnelly added.

Supply chain inaccuracies

It has been estimated that stores can lose up to 10 per cent of sales through supply chain inaccuracies a year. Mr Donnelly said in contrast, he expects retailers to see a minimum 2 to 3 per cent sales gain using the solution.

“We’ve being very conservative with such an estimate as it will likely be a 6 to 7 per cent gain,” he said.

Mr Donnelly said the solution, which has been developed with US consultancy R4, could be offered to other retailers through a licensing deal at some point in the future.

This is by no means the first time BWG has turned to technology to increase sales. Last year it trialled a solution that allows shoppers to pick up a product, scan the barcode with their smartphone to pay for it and then leave the store without queuing.