Working with more than gut instinct
IBM Fellow Brenda Dietrich discusses how analytics helps make decisions throughout modern business
Analytics, once the savior of manufacturing and supply chains, is now enmeshed into almost every aspect of business. And Brenda Dietrich, an IBM Fellow and vice president, is among those leading the charge. The reason is simple: it works, in everything from assessing investment risk to improving communications.
“Analytics is the use of data and computation and mathematics to help make better decisions. And in business, a whole lot of decisions are made,” says Dietrich. She believes that humans, faced by an onslaught of information, desperately need analytic help.
“The world is changing, and the way things work today is not the way things worked five years ago. People have less time to build up experience that will remain relevant.”
Going beyond the gut
And that has diminished the value of gut instinct. “Today’s world is so complex, so fast moving and with such a range of competitors that, while gut instinct is still valuable, it really needs to be informed by actual information,” says Dietrich. “But, I don’t think analytics and gut instinct are exclusive of each other.”
Analytics does have its limits, however; it lacks creativity. “Analytics is never going to tell you that doing something completely different, something that you’ve never done before, is the right thing to do,” she says. “There are two ways to learn: exploitation–this worked last time so it should work again–and exploration, trying something new.” Analytics is useful primarily for exploitation. Sometimes, however, taking an action about which there is no data can be useful, just to gather new data.
IBM embraces analytics internally
The current limitations haven’t stopped Dietrich and her coworkers from introducing analytics into almost every area of IBM. “In IBM, we did a very good job at applying analytics in things that it’s not usually used in,” she says. “And that’s interesting because it’s not work that researchers in this field typically look at. The more common approach is ‘The formulation is known, so how do I solve it faster or more accurately or on a larger scale?’
“In most of the work we’re doing, with the exception of supply chain and quality control, the models aren’t known. We have to create models, and as a byproduct, we begin to understand what sort of data should be better managed and retained.”
Dietrich and two colleagues, Emily Plachy and Maureen Norton, have compiled their experience with analytics in IBM into a book, Analytics Across the Enterprise. Their central message is that analytics is not a technology, it is a way of doing business, and they reinforce that point with numerous case studies.
IBM, for example, has operations all around the world, on every continent except Antarctica. How does it determine if a country is too risky from a financial viewpoint because of factors like weather, crime, or unstable politics?
“We do a lot of work trying to understand risk and uncertainty,” says Dietrich. “Projects about risk are by their nature difficult. Most people have trouble thinking about uncertainty productively. They tend to jump to the worst case, or they just dismiss things by saying, ‘It’s a low probability, so I’m not going to think about it.’
“What we’ve tried to do at IBM is raise the dialogue internally about risk associated with decisions. It’s just not that we do a best case/worst case analysis, decide we can live with it, and go on. We try to understand risk as a] function of uncertainty in the world we operate in.”
That means not just identifying risk, but quantifying as best as possible how that risk can be met. Take living near an active volcano as a hypothetical example. That’s certainly a risk. But if there’s enough forewarning of an eruption, you can mitigate your risk by evacuating. The analytics team further looks at how much it would cost to recover from a risk and factors that into the final decision.
‘Listening’ for better communication
The team also helped measure how effective communications were within the company through a project called social media listening. The project aggregated internal communications—web site comments, blogs, etc.—and analyzed it. “Essentially we’re taking words that people might use, topics they might be talking about and turning those signals into a time series, a series of numbers with time stamps,” says Dietrich.
Of course, privacy protection was very important. “We don’t want to report, ‘Tuesday, Brenda said this, and Wednesday, Brenda said this.’ What we want to be able to discover is that a group of people, let’s say consultants in the eastern part of North America, is starting to talk about this subject. Or they seem to be expressing concern about this other area.”
Given a summary of such information, decision makers can addresses areas or audiences where they need to communicate better.
The next great challenge for business analytics, says Dietrich, is words. She recently joined the team that is developing business uses for Watson, the cognitive computing system.
“A lot of the information in the world is in written language. And most of analytics is about working with numbers. We’re looking at things we can do with natural language to essentially create new columns of data.”