Welch's hired Oco, a software-as-a-service vendor, to provide analytics and data warehousing. Oco accesses Welch's ERP system through a secure Internet connection, skimming the data from order fulfillment and other modules and putting it in a data warehouse at Oco. The process takes about half an hour, Coyne says. Each morning, Welch's analysts tap into Oco to draw the fresh data to their own PCs. They can study it in pre-written reports or formulate their own queries.
The time saved on data collection and analysis enables Welch's to bid out all of its transportation routes and do it more than once per year, Coyne says. Welch's can also tweak those routes to save money. For example, the company discovered that if it relocated some of its distribution points, it could use rail instead of trucks, further reducing fuel use and rates as well as carbon emissions. All told, Welch's cut 12 to 15 percent from its $50 million in annual transportation costs, Kilcoyne says.
Conversations with customers
As Kilcoyne and Coyne learned, modern business intelligence and analytics tools can extract data from enterprise software, populate prebuilt statistical models and quickly produce insights that used to take weeks. "In the past, doing predictive analytics needed a Ph.D. in statistics to build a model and interpret results," says Aberdeen's White. But newer analytics tools "hide the underlying statistical nerd details," he says. "Business people don't have to worry about how the sausage gets made." Many tools, have models built in for such complex analyses as segmenting customers based not simply on demographics and the products and services they buy, but also on less cut-and-dried information, such as how they behave at a website or what comments they make during call-center interactions.
Key to game-changing decision making is the ability to detect and respond to market changes, taking into account historical knowledge. DirecTV uses analytics to save customers who want to cancel their television service. The company started the program two years ago when it sought to cut churn rates.
If a regular call-center agent can't persuade a customer to stay, the agent will let her go, promising to discontinue service in 24 to 48 hours. During this cooling-off period, a specially trained agent calls the customer, armed with the knowledge of why she wanted to cancel and a series of proposals designed to change her mind. The proposals are ranked according to the likelihood that they will work, says Jack Gustafson, director of business intelligence at DirecTV.
To someone who cited a competing offer from Verizon, DirecTV will offer a better deal. To someone who complained about technical issues, DirecTV offers free service, support, and perhaps upgraded hardware.
How hard agents press depends on how valuable the customer has been to DirecTV, Gustafson says. "There are some people we just do not want to lose." About 60 percent of customers who want to depart are deemed worth trying to save, he says. The company uses tools from Teradata and SAS to analyze past behavior, evaluating data such as the average annual revenue the customer represents, her payment history, and how many pay-per-view shows she buys.