Sign up FAST! Login

Ford’s $100 million data machine!

Ford’s $100 million data machine!

model T

To manage its sprawling dealer supply chain, Ford Motor developed a system that uses data analytics to discover and predict buying trends. Eight years ago Alan Mulally arrived at Ford Motor with a plan to get the automaker back on track. Two of his initiatives were focused on tightening the automaker’s extensive supply chain for its dealership network. The first set out to optimize what Ford (No. 8 on the Fortune 500) offered its customers, meaning that it needed to avoid building cars configured in ways that customers didn’t want. The second initiative sought to radically simplify Ford’s product portfolio and sell the same vehicle platforms worldwide.

The initiatives intended to bring cost savings to both Ford  F 2.40%  and its dealerships, but the reality wasn’t that simple. Centralizing vehicle assembly meant that delivery times would increase, adversely impacting the amount and newness of inventory on a dealer’s lot. “It really matters what’s on the lot at that moment,” says Bryan Goodman, a research scientist at Ford. “Having the wrong colors or options could have a big effect on our sales.” And rashly upping the number of vehicles in stock would just multiply inventory holding costs — about $10 per vehicle per day, Goodman says. 

Which meant Ford needed to get smarter about its inventory. In 2007, Ford began in-house development of its Smart Inventory Management System, or SIMS, which sought to equip dealers with data they could use to better predict which vehicles people would want to buy well before they set foot on the lot, allowing assembly plants enough time to make and ship them. “We present a lot of the analytics to the dealers in case they want to make their own inventory decisions,” says Michael Cavaretta, data scientist and manager at Ford. “We were concerned about improving sales but also improving the supply chain.”

It’s a formidable task: For a complex vehicle like Ford’s Transit Connect van, there are about 60 choices that a customer can make, from exterior color to roof height to wheelbase length to door style — in all, more than 27 quadrillion combinations for a single vehicle. “As the downturn started and sales slowed down, it was critical to operate with lower inventory levels, which makes it harder to keep vehicles in stock with options that customers are looking for,” Goodman says. “If they come in looking for a vehicle that’s red and with a moon roof, we better have one that’s red with a moon roof.” After a U.S.-only pilot, Ford implemented SIMS in 2009, running it alongside crash tests and fluid dynamics on a 40,000-processor supercomputer. (Consider: Ford’s dealers in North America alone make about 50,000 vehicle orders per week.) The new system’s impact was significant: Its recommendations saved dealers $90 per vehicle and reduced vehicle trades between dealers from 40% to 30%. “We’ve seen reduced time on lots and quicker inventory turnover,” Goodman says. 

And dealers are embracing the system. “In the U.S., when we look at the match rate between the recommended orders and the overall orders that were entered,” Cavaretta says, “we’re at about a 98% match.” It’s led to some quirky discoveries too. A truck with gold and green features selling unusually well in South Bend, Ind., turned out to be the result of University of Notre Dame school spirit. And options that were a sales deterrent in isolation — such as a trailer-towing package or a load-leveling suspension — sold quite well together. “SIMS has been worth well over $100 million a year to Ford,” Goodman says, “if not higher to our dealers.”Continue reading at: Ford’s $100 million data machine!

Stashed in: Cars!, Big Data!, Ford

To save this post, select a stash from drop-down menu or type in a new one:

Is it safe to say that their attitude toward data is one of the reasons Ford does so well as a company?

You May Also Like: