How analytics helped Ford turn its fortunes

Ford is using big data to drive virtually every aspect of its global turnaround.

Big data and analytics permeate virtually every move Ford makes, from forecasting the worldwide price of commodities to figuring out what exactly consumers want, what it will build, where it should source parts and how to power its lineup of vehicles.

"Data will set you free."

That's how CEO Alan Mulally has launched many a meeting since his 2007 arrival at what was then the sinking ship called Ford Motor Co.

Six years later, Mulally's opening line has become the automaker's bona fide company mantra. Bearing good news or bad, Ford managers wouldn't dream of coming to a meeting without hard metrics in hand.

"Historically, you didn't want to share bad news at Ford," says John Ginder, manager of system analytics and environmental sciences and a 22-year Ford veteran. "What Alan brought to the company is a mindset that we would no longer operate on anecdotal evidence. I've seen that explode in the past five or six years. Our CFO and COO are huge proponents. Today, it's all about having a data-driven perspective."

Indeed, data and analytics permeate every business move that Ford makes, from forecasting the price of commodities to figuring out what consumers want, what the company will build, where it should source parts and how to power its lineup of cars and trucks.

Crunching data behind the scenes are some 200 big data and analytics experts from a broad spectrum of disciplines. They work in what Ford calls analytics centers of excellence, which are found in various units of the $134 billion company, including marketing, research, credit services and others.

Since 2007, these analytics experts have contributed mightily to urgent strategic and tactical turnaround decisions, working on projects that ultimately decided issues such as which brands and models to discontinue, where to procure parts and materials, and how to enable dealers to tweak their inventories to improve sales.

Quarterly financial figures, Ford's stock performance and the company's most recent annual report tell the rest of the story so far. In 2009 -- one year after reporting a record $4.6 billion loss -- Ford posted a profit for the first time in four years. That same year, it launched 25 new vehicle lines and sold 2.3 million cars and trucks in the U.S., becoming the first brand to top the 2 million mark in U.S. sales since 2007. In October, the automaker reported its 17th consecutive profitable quarter.

Data analytics is a key competitive tool for all carmakers, which are slicing and dicing data on customers, production, vehicles and more to predict demand and hone their product offerings. General Motors, for example, has been collecting vehicle diagnostic data and other information via its OnStar system for years. But Ford is heavily focusing its analytic efforts on customer preferences, an area where it appears to be ahead of its competitors, according to Thilo Koslowski, an automotive analyst at Gartner.

"Everyone is working on [analytics] and to varying degrees have [gained] insights with regard to product usage," he says. "I see Ford having explored this sooner than others, especially in terms of understanding customer preferences."

At Ford, "there was very much a sense of urgency, and that tends to drive people to think outside of the box," says Ginder. "There's nothing like a crisis to help get the type of acceptance that's important [with analytics]. The crisis of 2007 and 2008 was a major turning point for our company and the role that analytics can play. With that combination of crisis, new management and a new perspective, we were emboldened and empowered to look at things quite differently."

Ford has derived some of the greatest returns from analytics investments focused on three areas: ascertaining what customers want, managing vehicle complexity and delivering to individual dealerships the right cars with precisely the right features that customers in that particular geographic area want to buy.

The Right Car to the Right Dealer

For decades, Ford, like all automakers, has relied on extensive market research, surveys and focus groups to get a grip on the heart's desires of drivers.

"But that doesn't always give us a complete picture because that data needs to be standardized in order to do comparisons," says Mike Cavaretta, project leader for predictive analytics at Ford. One way the company is addressing the issue is by monitoring social media for more specific intelligence and customer feedback.

"The nice thing about social media is that people elaborate," Cavaretta says. "They talk about more things and go beyond whether something is simply cool or not."

For example, when Ford was monitoring social media streams to learn how people felt about its three-blink option for signaling a lane change, the company learned a lot more about turn signals than it set out to know. "We found that in some vehicle lines the turn signal wasn't high enough or [was] in the wrong place. The problems people talked about weren't with the three blinks, but other aspects," Cavaretta notes. Ford was able to incorporate such feedback into decisions about new products and features.

Building in the features that drivers want is one thing. But ensuring that dealers have those cars on hand to sell is absolutely critical to turning a profit. "Quite a few customers walk into a dealership and want to leave with a vehicle that day, so we're limited to vehicles on hand that day," explains Ford research scientist Bryan Goodman, who works on analytic systems that support sales and marketing and their intersection with materials planning and logistics. "We have to get the right vehicle with the right engine and right set of features and controls to the right dealerships."

Merging Data Lanes

To do that, Ford integrates and analyzes several data streams, including data on what it has already built and sold, data on what has sold in the context of what was available in inventory at the time of the sale, plus data on what customers are searching for and configuring on company websites. This data is then combined with economic data to predict vehicle sales relative to housing starts, employment rates and other information. The system is known as the Smart Inventory Management System, or SIMS.

Story continues on next page >

Data Analytics at Ford

Covering All Bets

Like all automakers, Ford wants to make the best choices when picking the next new vehicle and fuel technologies. Its future depends on it.

Data analytics is playing a central role in that process.

Ford has partnered with researchers at Chalmers University of Technology in Sweden to develop a global energy model to help decision-makers understand global energy supply and demand and how to meet its needs at a minimal cost and in a sustainable way.

"The model looks 100 years into the future and can be used to address what-if questions, such as, 'If we had a carbon dioxide emission target of x, what would that mean for autos, trains and planes?'" says senior technical leader Tim Wallington, who heads the Ford analytics team focused on sustainability issues.

"There is a wide range of vehicle fuel technologies in the future, and what we did in the model is include our best estimates of the current and likely efficiency of those technologies and a range of costs associated with them," Wallington explains.

For example, "when looking at electric vehicles, we know what they cost now, but how much they'll cost in the future we don't know," he says. "We feel confident that the cost of batteries will decline, but we don't know by how much, so we include in the model a number of different takes."

By manipulating various what-if factors, including emission targets, fuel types and costs, Ford researchers ultimately came to the conclusion that, for now, "there is no silver bullet," Wallington says. In other words, no single technology is the correct choice for the vehicle of the future, which is why Ford has adopted a portfolio strategy to developing sustainable technologies and fuel options.

"We did thousands of scenarios, and the bottom line is that given the uncertainties in future costs and efficiencies, it's not possible to pick a winner," Wallington says. "Customers can vote with their money as to which they want and which one wins the future. This is the high-level strategy."

The upshot is that Ford is making a range of vehicles with alternative fuel options. These include cars with advanced diesel engines, hybrids, plug-in hybrids, all-electric vehicles and alternative fuel vehicles. The company uses the catchphrase "the power of choice" to market that strategy.

Some of Ford's competitors, meanwhile, are focusing on battery-powered and electric vehicles. "Others have put more of their eggs in a single basket," says John Ginder, manager of system analytics and environmental science at Ford. "In the first decades of this century, [other automakers] spent a lot on fuel cell vehicles. They were very bullish on them. We are intrigued by them, but we also have a prudent risk management." That approach, says Ginder, is solidly based on analytics.

- Julia King

"We largely figured things out over various experiments and applications in this area over many years, but much of it was not enabled until the last few years because of the increase in computational power," Goodman says. "We have 20,000 compute cores in the building next door at our disposal. We have computers with 1.5TB of RAM. It's those sorts of resources that have enabled us to synthesize this kind of data."

All around, SIMS is acknowledged as a pivotal factor in Ford's turnaround and the overall success of Mulally's global "One Ford" strategy.

"As we globalize and leverage products from around the world, it means new complexity management challenges," says Goodman. Before, "if any feature was desired in any market, we'd engineer it and make it available. When you get into different roof heights, different interiors, different wheels and so on, we can offer an astronomically large number of combinations. Imagine 300 billion and then multiply it by itself again. Customers get overwhelmed by too many choices." Instead, Goodman says, "we want to match customers' wants with our global supply chain. It's a challenge throughout the industry, but there's huge value in getting it right." This requires knowing customers' key preferences, then tailoring the bulk of the company's production to those preferences, rather than building a very wide range of models with many combinations of features. It comes down to building cars that most customers want most of the time.

This is where SIMS has really paid off. Assembly plant schedules and parts forecasts have both significantly improved because "we can run those schedules with better algorithms," says Goodman. "In 2007, we could do it but it would take two to three weeks, and assemblers needed answers in 20 minutes or less. Today, we can do it in two minutes."

Ford dealerships, which are independent franchises, also have benefitted. Some 3,500 dealerships receive the weekly reports. Those that subscribe to SIMS recommendations say that they are selling cars at higher prices and more quickly, says Ginder.

Down the Road

Yet Ford's analytics experts say they have only just begun to scratch the surface with big data. The next big frontier is data that streams from vehicles themselves.

"The volume of data generated by vehicles is huge," says Ginder. Ford's Fusion Energi plug-in hybrid car gets the equivalent of 108 miles per gallon. It also generates 25GB of data per hour.

Story continues on next page >

Data Analytics at Ford

Translating Business Needs to Mathematical Language

As John Ginder sees it, the best data scientists possess a combination of mathematical expertise, familiarity with computer science and programming applications, and the ability to translate business needs into mathematical language. Usually, Ginder can find people skilled in one or two of those areas. "But it's hard to find all three traits in one person," he says.

Ginder, an early proponent of analytics at Ford, is a physicist by training. As it turns out, so are many of the automaker's other 200 or so data scientists.

"I personally look for people who can adapt and reinvent themselves on a regular basis," Ginder says. "We have physicists, chemists, applied mathematicians and research specialists. Physicists are a good source because they have a certain mindset on how to approach problems. But we also want people with MBAs and engineers."

Ford's data scientists are assigned to so-called centers of excellence associated with different departments, such as marketing and research. They work on both strategic and tactical issues, ranging from which models of cars to produce to where to source materials and where to build certain vehicles.

The key to a successful project is the stability of the data, Ginder says.

"I'm always looking for a stable source of data and looking to what degree we can expect to continue to have that data," he says. "My contention is if there's not a guarantee we'll have access to that data in an ongoing fashion, we don't want to develop an application around it."

Over the past decade and a half, Ford's centers of excellence have developed successful relationships with the company's IT department. But it wasn't always that way.

1 2 Page 1
Must read: 10 new UI features coming to Windows 10