Fans of fantasy and science fiction often dream about unicorns, but they are not the only ones: When I speak with IT and business leaders who are building analytics and data science departments, I hear fantasies about identifying a candidate who’s a unicorn: great at data, math, computer coding, communicating with other departments, and expert in the company’s industry, too. But since these are hard – if not impossible – to find, other strategies are needed to build a leading analytics team.
In April, I joined a panel moderated by Harvard Business Review Editor at Large Julia Kirby to talk through the problems of succeeding in analytics team building, especially at a time when analytics experts and data scientists are in demand. On the panel, held at the 2015 INFORMS Conference on Business Analytics and Operations Research, I was joined by Brenda Dietrich, an IBM Fellow and Vice President; Bill Klimack, Manager of Decision Analysis Consulting with Chevron’s Project Resources Company; Michelle Davis, Director of Analytics at FICO; and Zahir Balaporia, Director, Process and Technology at Schneider.
Given that analytics is such a hot profession, job candidates often interview those doing the hiring as much as management interviews them. This is where the job description requires finesse: Michelle Davis of FICO explained that the job description in our field shouldn’t just enumerate the quantitative requirements and soft skills; it also should be an advertisement that sells the job, excites the candidates, and draws them to the company. This can be especially important if the company isn’t located in a tech hub, as Zahir Balaporia noted about attracting people to Schneider, a transportation company located in Green Bay, Wisconsin.
For those in a position to hire, Brenda Dietrich remarked on one problem she finds among many young analytics folks and operations researchers: impatience. These bright people, she has seen, don’t understand why, once a team has conceived a new product, the company can’t rush to market within, say, a few weeks. Alas, corporations don’t move that quickly. They need to test a product, evaluate its profit potential, and engage everyone from manufacturing to marketing and PR. So hiring can require subtle, probing questions to determine if candidates comprehend business basics, and if they have the patience to stay with a project even when there are delays.
For companies taking their first steps in analytics, I remind them it’s possible to get started with software and services from providers, generate some successes and learnings, then a year or two later reassess and determine if they want to build their own data science or analytics team.
Brenda Dietrich observed that for those who do choose to start an analytics department, first hires are critical in setting the tone at the department. Look for a core of people with a breadth of skills. Bill Klimack echoes this, saying that people with a broad range of expertise come together nicely in a company like his, Chevron, which prides itself on collaboration among professionals from varied fields.
Many organizations ask where analytics teams should be placed and should they be centralized or disbursed within a company. I was glad to hear the panel’s response: Blend the two models. There are advantages to centralizing – especially as teams grow – to share skills, tools, methodology, and scale. However, with advanced analytics, it is critical to be close to the business and understand the decisions that are being made.
Gartner research shows that analytics groups are situated in many different corporate departments. We recently surveyed 600 advanced analytics practitioners and asked where their teams reported. About a quarter report to a CIO or to the IT department. That means that the remaining 75 percent report to various lines of business – the CFO, chief marketing officer, chief data officer, operations, strategy and planning, the chief risk officer, the head of engineering, and more.
If you choose to decentralize, said Michelle Davis, keep everyone in touch. Often, analytics groups within the same company lose contact. They gain a great deal by comparing projects and sharing ideas.
An important issue for analytics teams is handing off a finished project – how do you make sure your new concept works in practice? Sometimes, remarked Bill Klimack, you embed your analytics in the everyday software used throughout the company. Sometimes, though, you have to ensure buy-in. Zahir Balaporia remembered creating an optimization project for dispatching drivers, a project intended to make major improvements in the way that trucks were routed. When the Vice President of Operations determined that adoption was spotty, he turned to Balaporia and charged him with ensuring company-wide implementation.
Julia Kirby asked how companies retain analytics talent. Given that top analytics people relish a challenge, panelists recommended that analytics leaders identify hard company problems whose solution will make a visible improvement. Spending too much time developing the analytics and failing to persuade partners and managers to adopt a project may result in analytics projects that get vetoed or ignored, and that’s one result that definitely harms morale. But individuals are different, and you have to figure out what drives the people on your analytics team: Is it solving challenging problems? Working with cutting edge technology or a smart team? Or seeing the impact of their work?
Julia Kirby and this wonderful panel of analytics leaders did a great job of explaining what it takes to have a top analytics team and collaborate with others. You can view this panel discussion on the INFORMS website by clicking here.
Lisa Kart is a Research Director at Gartner covering advanced analytics. She leads research on predictive and prescriptive analytics and big data uses across multiple industries. She has more than 20 years of experience as an analytics practitioner.
Subscribe to Data Informed for the latest information and news on big data and analytics for the enterprise.