Tesco. Poster-child of the data-driven retail experience, thanks to its loyalty card and early work with DunnHumby. But Tom Hebbert didn’t talk about that. Instead, he focused on their supply chain work.
Rough notes from the session follow.
Big Data projects deliver huge returns at Tesco; improving promotions to ensure 30% fewer gaps on shelves, predicting the weather and behaviour to deliver £6million less food wastage in the summer, £50million less stock in warehouses, optimising store operations to give £30million less wastage.
Tesco – 2,979 stores, 30,000+ products in a big store, 23 depots in the UK, 50million cases delivered into stores each week, 16million of which contain perishable foodstuffs.
A ten degree rise in temperature means 300% more barbeque meat. 45% more lettuce. 50% more coleslaw, 25% fewer sprouts.
But weather is not straightforward, and it is very localised. It also has different effects dependent upon the day of the week, and the location. A city-centre store will see an uplift in sandwiches (to eat outside) on a warm weekday… and almost no effect at all on a warm weekend.
And basic facts around temperature, sunshine etc don’t tell the whole story. A 16 degree sunny Saturday in late April will cause a spike. Exactly the same figures a couple of weeks later will not, as people have had their first BBQ of the season already… Complicated stuff.
Tesco gets weather forecasts three times per day, and updates the store forecast for 18million items three times per day. Data is freely and immediately stored with suppliers using a web-based system called TescoConnect.
Largest depots hold around 12,000 products from 700 suppliers. Regional depots hold over 5,000 products from over 200 suppliers. Different suppliers deliver stock with different lead times and at different rates and volumes.
Some suppliers can supply in hours, others in days… or weeks.
Economic order quantities… cost of ordering decreases as order volumes rise. Textbook way to manage and order stock. But… Tesco and its suppliers are really dealing with a single item. They order lots of different things from Nestle, with different variables at play. Understand the range of products you’re interested in, and group/order/hold accordingly.
Tesco runs a full simulation of all their stores and all their depots. They can play back exactly how stock moved historically, and optimise for future situations. They do it in Matlab, running on desktops ??? Takes 3 days to run the simulation. Largely responsible for freeing up £50million in working capital.
Reducing waste in stores
Tesco sells half a billion reduced-price products every year, as they approach sell-by dates. In fresh food, 3% of volume must be reduced as it reaches sell-by dates.
Some products are more price elastic than others, and Tesco models to understand exactly when and by how much they need to reduce in order to ensure stock shifts. Cream cakes show the biggest sales uplift with the smallest price drop, so psychology important too.
Project delivered £30million ‘in pure profit’ by discounting prices more intelligently – not going for a big price cut too early.
(Cross-posted @ The Cloud of Data)