To help determine whether someone embraces this approach, Fink advocates “a portfolio review model to assess people’s data analysis problem-solving skills.” By walking through a project in which the candidate changed course, for example, she can evaluate his approach in using technology to solve challenges.
“You can gauge their judgment” and fit for a specific position, Fink says. “For example, did they use statistical analysis to understand whether a pattern was forming that justified the change?” Further reviewing candidates’ work samples helps her “determine the quality of the code they are writing and the appropriateness of the statistical analysis that they run.”
Borne suggests looking for people who espouse a penchant for lifelong learning and have demonstrated an ability to accept change. These qualities tend to align well with data fueled projects.
At Booz-Allen Hamilton, he says, “We have internal training for employees who want to become data scientists or learn data analytics skills. The course runs an hour a week for nearly a year; to qualify, employees must pass a math and programming test.”
Building a Data-Driven Culture
A data-driven culture is one that rewards data collectors across the organization. It’s led by executives who want to know what the data suggest, who develop a decision-making structure that includes data analysis, and who base plans on that analysis.
It might sound like a big effort to get there, but to a large extent, it’s just a formalization of common behaviors.
For example, Chevron is now rolling out a training program that uses real data and actual problems facing line-of-business employees.
“Most people see a lot of data in their daily jobs,” she says. “That is the data we want them to work with” in the training program.
Chevron pairs training participants with analysts who can use algorithms to come up with answers. This practical approach to training sharpens employees’ ability to frame business challenges and identify how the data can help solve them.
Over time, a company can formalize this process further.
Among other data-focused policies, Chevron has implemented a company wide mandate that every project proposal above a certain dollar threshold includes specific types of analysis.
“You have to prove that you included ranges of uncertainties in the data or the economic assumptions,” says Amy Absher, general manager of strategy planning, service, and control at Chevron. “Funding is dependent on this insight.”
For large projects at Chevron, Connor says, “An independent group comprising representatives from across the company comes in and looks at the decisions we are making, the alternatives we considered, and the uncertainties to determine whether we did a broad assessment prior to moving forward.”
Chevron also performs a formal project review to compare predictions with results.
“The fact that senior management expects that analytical work to be done and asks these questions sets the expectation and the tone throughout Chevron,” Absher says.
Most businesses today are gradually building a data centric culture, Borne notes. To those just beginning, he advises simply finding an analytics project with a good chance of payoff.
A financial-services firm Borne consulted, for example, decided to search its Web analytics logs for an indication that a customer was thinking about defecting to a competitor.
“They found a signal in their data,” Borne recalls, involving “a lot of price comparisons. They applied a very soft ‘customer services’ response, inviting people to look at some of the company’s new products or offering to answer questions.” After a quarter, “they estimated they saved the company more than $100 million in potential lost customers.”
A smaller company can experience the same impact with smaller-scale returns. And as analytics projects deliver value, the culture and processes to support more data work can be built step by step.