Some companies focus on quarter-to-quarter profits while others come up with five-year plans. At Dow Chemical, strategic thinkers are looking as far ahead as decades, considering population, energy, and environmental trends.
The company is evolving from a supplier of basic chemicals to a value-added manufacturer of high-margin specialty products. Advanced analytics has been a catalyst for that change, helping the company better anticipate demand even as markets have become more volatile.
"Market changes are happening more rapidly than we've ever seen in our history," says Paula Tolliver, VP of information systems. "Having early insight allows us to respond and react instead of finding out after the fact."
Dow generated $53 billion in revenue in 2011 selling to industries as diverse as automotive, agriculture, construction, electronics, healthcare, mining, oil and gas, and transportation. It has used predictive analytics and mathematical modeling techniques for decades within its engineering, production, and R&D areas. But in 2005, Dave Kepler, CIO and executive VP of business services, kicked off an experiment whereby the company's analytics expertise is shared with service departments and business units.
The effort began in the supply chain, purchasing, and health and environmental sciences areas. Half a dozen internal Six Sigma (process optimization) experts as well as outside consultants supported two analytics professionals from R&D.
One of the earliest successes was the development of freight and logistics cost models analyzing about $2.8 billion in annual truck, rail, ship, and air freight costs worldwide. Models were also developed to analyze $4 billion in annual raw materials spending. The models helped Dow save big bucks, says Tim Rey, Dow's competency director for advanced analytics. "If we know what our costs are going to be, our procurement people can make decisions about renegotiating contracts, buying early, or waiting to buy in order to reduce costs," Rey explains.
Success bred demand, leading to a corporate-wide initiative in 2010. Whereas the earliest efforts focused on service departments, 85% of the analytics projects are now tied to business units, which want better statistical forecasting.
Using a mix of internal data, economic data, and third-party industry information, the analytics team has developed a battery of leading indicators to help business units anticipate demand six to nine months in advance. Dow has also stepped up access to real-time information on orders, shipments, and sales trends by industry, market segment, and region.
The real-time business intelligence reporting, analytic forecasts, and third-party industry stats all come together in a series of dashboard-style reports for business decision-makers. These layered dashboards let business leaders roll up or drill down on the latest data and predictions for the specific business units, segments, and geographies they manage.
Dow has developed thousands of predictive models. Enhanced sales forecasts backed by advanced analytics have reduced forecasting errors. Business units now know by midmonth whether they'll meet monthly performance targets so they can adjust their strategies accordingly. Exchange-rate and margin analyses have helped Dow make decisions about where to buy raw materials and how to determine pricing of finished products. And data-driven staffing models have helped the company hire the right talent at the right time.
The statistical work is done mainly by 10 Ph.D.-level pros who lead the advanced analytics team. That group also draws on up to 15 analytics consultants hired on a project-by-project basis.
Those pros work with business analysts and subject-matter experts from each business who help identify the metrics for each model. Dow's BI team includes another 40 to 50 employees, and projects might also draw on a 30-employee business information services group that develops and manages data sources.
All of these groups, as well as the 741-employee IT department, report up through Tolliver, who says Dow is now looking for ways to keep up with still-growing demand for advanced analytics. "We're just beginning to make this a part of our standard processes and practices," she says, "and we need to get better at it for sure."