How improving staff scheduling optimises conversion for DFS

Furniture retailer DFS employs 5,500 workers across 230 stores. However, it previously didn’t have an accurate view of future demand in stores, says Chris Allen, senior technology pattern at the retailer. Consequently it had too few staff at peak times and too many at low-demand periods.

“A lot of responsibilities go into store [planning] and trying to accommodate manually and then to align the footfall is a task in itself that is unobtainable,” he says – speaking this week at technology conference CogX 2020.

By partnering with AI company Satalia, the company was able to use data to predict demand to an accuracy of around 80%, compared with actual footfall, and then feed that information into a workforce optimisation tool. That generates predictions and schedules six-weeks in advance and redeploys people during more demanding periods.

“We have seen like-for-like sales in stores increase, using less hours across the week but using people in a more targeted way… and an increases in conversion,” he says.

As well as increasing sales, employees benefit from fair schedule – where everyone gets a chance to maximise commission  – with the scheduling also taking into account shift preferences and allocating staff tasks they are supposed to do. “This is not an exercise in cutting heads,” he notes

“If someone can work 20 hours and earn the same as they did doing 30, that’s better for the business and better for the employee.” Having an empty store full of sales people can also have a negative impact on customer experience, he adds. “Customers spend less time in store when the store is not busy”

Data lake

DFS’ business operations are larger than just its stores. The firm has its own manufacturing facilities, and is the biggest manufacturer of sofas in Europe. It also delivers the majority of its own products and has its own supply chain logistics operations and warehousing  facilities.

Allen believes the data predictions can help improve the firm’s entire operations. “It’s about reusability, being able to take what you have found for a specific purpose and use that elsewhere is very important to us as a business,” he says. The demand prediction is being utilised in other parts of the business.

“We have a data lake at DFS,” he says. “And this can inform other predictions.” For example, using the AI tools it could test the impact marketing spend might have on footfall. “As a business it is very expensive to test marketing theories in the real world,” he says. “But the fact we are now predicting demand to a level of above 80% means we can change the marketing in the digital twin and predict what impact that will have on footfall.” And that in turn could help it gain a view of what it means for its manufacturing and delivery operations

“All that is digital and available in our data lake. Now it’s really about connecting the dots and understanding that journey.”