satalia-video-placeholder
satalia-logo

Route optimisation for DFS

We adapted our delivery product, and used machine learning to optimise their home delivery schedules - reducing costs, and improving the customer and employee experience.

Challenge

As a leader in their industry, DFS recognised the need to offer more convenient time-slots to their customers. And in a home delivery market where driver churn is high, wanted to ensure their driver experience was the best it possibly could be. 

Outcomes

8%

increase in NPS

19%

reduction in unforeseen driver overtime

18%

decrease in fuel consumption

Outcomes

8%

increase in NPS

19%

reduction in unforeseen driver overtime

18%

decrease in fuel consumption

Solution

We fine-tuned our last-mile delivery technology, and developed a series of custom machine learning modules to transform their home delivery offering — improving their customer and employee experience, and reducing their operational costs.

The solution is dynamic, re-optimising the schedule every time a new delivery is confirmed. The newly generated schedule informs the operator which slots could be offered to the next customer, and so on. This meant the slots could be offered with certainty, and that deliveries could be organised in accordance with the real-time capacity of the schedule, maximising the utility of the fleet whilst not putting drivers under undue pressure. The system has since been used to offer online, customer-facing slots - further improving the customer experience.

We used machine learning to squeeze as much efficiency from the schedules as possible. In one example, we used product dimensions data, mathematics, and parking data to predict the service time — the actual time spent delivering and installing the sofa. Given that a huge proportion of a driver’s time is spent making the delivery, and that these times vary drastically depending on the size of the item, or how close parking is, knowing these times accurately meant our slots were more reliable, and our schedules more efficient.