How to use AI to manage inventory in a time of stock shortages

The festive season won’t be very cheerful for retailers if their supply problems persist, but warehouse AI will give them a better chance of keeping their customers happy

A winter of retail discontent is looming for consumers in the UK. Post-Brexit border delays and HGV driver shortages are already leading to empty shelves in supermarkets around the country. There could even be no turkey for dinner this Christmas. 

The Food and Drink Federation has warned that shoppers can expect not to find all the goods they want whenever they want them for the foreseeable future. Its CEO, Ian Wright, told an Institute for Government event in September: “The just-in-time system is no longer working. I don’t think it’ll work again.”

Retailers and their fulfilment partners are seeking ways to navigate this sticky situation. AI could offer them at least some of the answers.

When people think about AI applications in retail, they might immediately envisage robots picking and packing products at speed in a fulfilment centre, with drones hovering above them on the lookout for potential problems. These devices will indeed be a common feature in the warehouse of the future. But the AI-based technologies of most benefit right now are those that can offer businesses a better handle on their inventories and supply lines. 

As shoppers become more accustomed to seeing out-of-stock notices, both online and offline, their purchasing decisions are likely to change – as will demand signals. So says James Hyde, founder and CEO of James and James Fulfilment, which handles warehousing and order fulfilment for hundreds of small etail brands. 

Given the ongoing delays to inbound stock, Hyde notes that there is “a great opportunity to revisit the slow-moving lines that are clogging up expensive warehouse space”.

When integrated with enterprise resource planning software and internet-of-things technology, AI systems can constantly monitor stock levels and adjust their demand forecasts as stock flows in and out of a warehouse. 

Although fulfilment companies can’t control delays upstream of them, they must do their utmost to minimise any disruption during a product’s time in the warehouse that might impede its progress to the customer, says Daniel Hulme, chief AI officer at marketing giant WPP.

“The ineffectiveness of fulfilment is a primary cause of customer dissatisfaction. AI can help firms to achieve exponential efficiencies in warehousing, distribution centres and the last mile,” says Hulme, who sold Satalia, the AI consultancy he founded in 2007, to WPP in August. It has previously worked with Tesco to optimise its last-mile delivery processes.

He suggests that companies could use digital-twin technology to improve their control. In essence, a digital twin is a virtual model of a physical asset – in this case, a warehouse – that updates itself in real time. “This living simulation can connect and orchestrate components of the supply chain,” Hulme says. 

Companies can use the data generated by the digital twin to anticipate problems well before they strike. They can even apply it to inform their marketing campaigns, according to Hyde. 

“Retailers need to be proactive about using promotions to control the sales process, both to maximise sales and to minimise the risk of fulfilment delays caused by excessive demand,” he says. 

For instance, promotions on slow-moving product lines should be prioritised over popular lines to prevent overstocking and understocking. At the same time, “promoting products that aren’t yet in the warehouse, or even in the country, is a fast track to customer disappointment,” Hyde warns.

While businesses are generally aware of AI’s potential benefits, some are struggling to use it optimally. That’s the key finding of a survey of 350 warehouse managers in the UK and the US by Lucas Systems, a specialist in warehousing technology. The consensus among the respondents was that AI could give them a return on investment of 60% within five years of its adoption. Yet 99% admitted that they were finding it hard to make the best use of the technology, while 90% said that they needed more support with implementing it. 

“There is a huge misunderstanding about the capability of AI, which leads to so many companies getting this wrong,” Hulme says.

One common misconception that companies have about AI is that it’s costly and involves a highly technical set-up, he adds. Another is that it’s only truly beneficial to companies handling a high volume of products. 

Hyde believes there’s never been a more appropriate time for retailers and fulfilment companies to start using AI. Preparing for the festive period is always a challenge but, given the well-reported lack of HGV drivers on top of a general staff shortage this year, businesses will find it harder than usual to cover any spike in warehouse demand. 

“Vague promises aren’t good enough,” Hyde stresses. Customers expect immediate service, but this won’t be achievable with the current bottlenecks. By using AI-powered tools to keep on top of inventory, retailers can make realistic delivery pledges, he notes. Customers would rather receive a delivery within an agreed time slot than be offered a next-day delivery service that’s not feasible. 

Retailers should be asking themselves how quickly warehouses can respond to customer demand and get products dispatched, according to Hyde. And key to this is knowing exactly what’s in stock at any given point and which product lines need to be promoted.

“Careful planning is vital if retailers are to get the most from peak trading season and, crucially, keep their promises,” he concludes. “Retailers need to know at all times what they have and when they can deliver it.”