Discover PerformanceHP Software's community for IT leaders // June 2014
How to connect data science and IT
Some say that data science and traditional IT are like "oil and water." But with the right approach, you can find a synergy that delivers business insights.
As far back as 2012, the Harvard Business Review declared data scientist the sexiest job of the 21st century. It’s certainly one that’s got CIOs talking—about how to hire and retain them, whether they’re needed, and where in the organization to place them. Though thousands of data scientists are already employed in startups and established businesses, many companies, prompted by the need to harness Big Data fast, are just now beginning to look seriously at the role—and its challenges.
Macbeath says that traditional IT organizations, having focused historically on highly structured, system-of-record, rows-and-columns data, are uncomfortable with all this unstructured data. "What we have today is a really different domain," he says, "one that IT doesn’t control. Data analysis in itself has not been IT’s strength."
Look before you leap
When you start to look at how data science might be deployed within your organization, think first about how IT is organized within the business.
"Your data science strategy has to respect the structure you have in place," Macbeath says. "One of the biggest challenges in hiring a data scientist is to find a home where they feel they can be productive."
Macbeath says to figure out how and where a data scientist fits in, you should first consider which of the three basic IT structures you have:
- The application development team resides within the business, with common shared services across the enterprise, and the CIO leads.
- IT is entirely centralized, has relations back to the biz, people related close to the business, and related knowledge.
- IT is completely decentralized, and each business unit has its own IT department, without much commonality.
Get the environment right, then go for a quick win
Creating the right environment for data science to become successful is key. According to Macbeath, "Your best approach may be to say, 'We’ll analyze unstructured data that has been proven useful elsewhere, and prove it here for a quick win.' Find a sponsor on the business side to let you apply analytics to something the business really cares about."
Make sure you put the data scientist to immediate use, Macbeath cautions. "What you don’t want is your data scientist engaging in science projects—you want business projects. If you hire them into IT, you should initially have done your thinking about where you intend to deploy them. Don’t have a floating data scientist. You want to hire them with specific problems in mind to sink their teeth into. Floating, they’re in danger of spinning their wheels as they get up to speed, or getting bored and over-optimizing."
Select a great test case
One way to start is by looking at all your possibilities and choosing the optimal test case. Another is to find one viable test case and start running. The latter, of course, is a lot more popular (in terms of justifying dollars invested) than a "boil the ocean" approach.
Macbeath recalls, "In some companies, I’ve seen IT identify a person who would work with the CIO to define and set up the data science function—sort of to set up the sandbox—and do initial work on a probable transition from sandbox to ops, clearing the runway before the data scientist comes in."
You might decide it’s best to first engage a consulting firm to run one or two analytics projects. To succeed, make sure to define and prepare for the project, Macbeath advises: "Don’t just ask a consultant for a 'data science project.' They’ll fail."
A winning environment
The most important challenges, Macbeath says, are to have an environment that lets you comfortably prove data science results, and to give the environment the flexibility and budget to identify useful patterns: "Basically, it’s all about being able to say, 'I think there’s something there to drive sales targeting or risk pricing or customer satisfaction—but I don’t know until I try.'"
Keith Macbeath, senior principal consultant with HP Professional Services, has more than 20 years of experience in helping clients use analytics to identify value, set performance agendas, and track results. Learn more about HP's approach to mastering, and profiting from, Big Data at hp.com/HAVEn.
HP Software’s Paul Muller hosts a weekly video digging into the hottest IT issues. Check out the latest episodes.
Introduction to Enterprise 20/20
What will a successful enterprise look like in the future?
Challenges and opportunities for the CIO of the future.
What the workforce of 2020 can expect from IT, and what IT can expect from the workforce.
Data Center 20/20
The innovation and revenue engine of the enterprise.
Dev Center 20/20
How will we organize development centers for the apps that will power our enterprises?
Welcome to a new reality of split-second decisions and marketing by the numbers.
IT Operations 20/20
How can you achieve the data center of the future?
Preparing today for tomorrow’s threats.
Looking toward the era when everyone — and everything — is connected.