Effective data analytics can give companies a huge competitive edge, because business managers can gain new insights into trends and customer behaviors that might not otherwise be possible.
To get the most out of their information resources, enterprises need to have a strong analytics team in place. What does it take to assemble and maintain a top-notch team, and what should these teams be doing to make themselves successful?
These are not trivial questions. In this heavily data-driven environment, how companies go about building and operating a team of analytics experts could have a big impact on the business for years to come.
But before you put together your data analytics team, you need to formulate the mission and charter of the team, says Jeffry Nimeroff, CIO at Zeta Global, a customer lifecycle management marketing company.
“In too many organizations, data analytics is embedded in the more traditional and bland notion of ‘reporting and analytics,’” Nimeroff says. “In these configurations, it is often the case that reactive reporting takes precedence. As there is always another way to formulate more meaningful reports, this can become a never-ending cycle where the true power of data analytics is never fully realized.”
Data success begins with diversity
When building a team, don’t limit the focus to just finding analytics professionals. Diversity is critical for success, experts say.
“It’s very important to include not only people with analytical skills, but also those with business and relationship skills who can help frame the question in the first place and then communicate the results effectively at the end of the analysis,” says Tom Davenport, a senior advisor at Deloitte Analytics and author of the book Competing on Analytics: The New Science of Winning.
Multinational conglomerate GE values a diversity of capabilities for its analytics teams. “Data and analytics are most effective when world-class technology skills are paired with strong functional domain knowledge,” says Christina Clark, chief data officer at the company.
This can be achieved by having a team with a variety of business backgrounds; a mix of both IT and functional skills, Clark says. “We are making terrific progress in developing innovative solutions to support our finance function,” she says. “The data team supporting this effort is comprised of long-time IT professionals but also financial analysts, former auditors, and finance managers.”
Strong knowledge of data science is of course critical to any analytics team, and there should be statisticians, mathematicians, and machine learning experts on the team who understand algorithms and how they can be applied on data, adds TP Miglani, CEO at Incedo, a technology services firm.
“You [also] need technologists — data engineers who can build the pipelines to get the data in place for completing all the analysis,” Miglani says. “And you also need business experts who understand the complexities of the domain you are solving the problem for. For example, if the problem at hand is building data-driven drugs, then you need quantitative pharmacologists and biologists.”
Technically, a data scientist is supposed to be a “unicorn” that can do all of this simultaneously, Miglani says. “But unicorns don’t exist,” he says. “Successful data science teams are diverse, where individuals bring in these competencies that need to come together.”
Change management and the value of IT
If an analytics project involves prescriptive or operational analytics (for example, if the results will be tied into a business process or a set of jobs), there is also a need for someone to manage the change process, Davenport says. “The ORION project at UPS, which led to dramatic changes in driver routing, devoted a massive amount of time and energy to change management,” he notes.
Given that the team will be leaning heavily on technology infrastructure such as big data tools, having the IT department represented on the analytics team in some capacity is also important. “Even if the analytics group doesn’t report to IT, it’s usually a good idea to have some representation of the IT function on the team,” Davenport says.
Whoever’s on the analytics team should have lots of experience in their role, Nimeroff says.
“Data analytics is both an art and a science, and more experienced individuals are better able to leverage tools in a creative and effective way than novices,” he says. “I have also found that novices rely on tools to do heavy lifting that they may or may not be fully comfortable in doing themselves. On the flip side, I have met great data scientists who do everything by hand. They don’t scale or help a team accelerate. Finding individuals who can execute without tools but understand and embrace the value of modern tools is what I focus on.”
Outside expertise and embedded teams
Many companies turn to outside expertise for help with analytics projects. That’s fine, but it’s important to ensure that the efforts of the project are actually meeting organizational needs.
“If there are some members of the team who are outsourced workers, try to make sure there is at least one [internal] employee on every project, who can help to ensure that the results of the analytics are adopted,” Davenport says.
And whenever possible, the analytics team should either be a formal part of the business that’s doing the analysis, or at least embedded within it for the period of the project. Consumer goods company Procter & Gamble used to do this through “embedded” analysts, Davenport says, but now has them report to the head of the relevant business function or unit.
Once your team is in place, finding an operational model in which everyone can work is next, Nimeroff says.
“Companies are becoming more agile and, like with software development, finding an approach for prioritizing work, decomposing the execution into digestible chunks, developing specific success criteria for each work effort, and providing a framework for ongoing communication is often the difference between success and failure,” he says.
Also, the team will more likely to succeed if it’s able to demonstrate the business value of what it does, Miglani says.
“Engaging with stakeholders and consumers of data science recommendations helps them showcase this value, and also get a deeper understanding of the key pain points which they should focus on,” Miglani says. “Sharing results sooner than later, and building organizational structures where data science goals are aligned to the [business units] they are tagged to is a great way to create value.”
GE practices “ruthless prioritization” with its analytics efforts. “A commitment to clearly defined business priorities will enable the data and analytics team to be most successful,” Clark says. “When teams can demonstrate an impact in targeted areas, they are more likely to stay motivated and inspire business partner engagement.”
The company has seen significant productivity results from the “Digital League” in its aviation business, where a cross-functional team has come together to define priorities and then deliver insights in two-week sprints.
Emphasize experimentation and innovation
It’s also important to keep an experimental mindset on the team.
“The business case for these projects is not easy; you have to take a step into the unknown,” Miglani says. “Unlike technology projects that begin with a definite scope in mind, data science projects begin with a problem and a set of hypothesis which needs to be tested. There is no clear map of before-and-after processes for these projects, and teams which are new to data science need to understand and get comfortable with that.”
Along those lines, there should be outlets for innovation, Clark says. “There is an enormous amount of emerging technologies in this field,” she says. “Employees will want to know they have time and funding to continue to grow their own skills and try new approaches. We leverage our Global Digital Hubs as a place to incubate new technologies and pilot work in self-organizing teams; an atmosphere of innovation keeps teams motivated.”
As with science and learning in general, curiosity is a key element of analytics. “Curious people have a desire to follow up on their own analysis, whether or not our clients ask for it,” says Stuart Wilson, data scientist and analytics team lead at Paytronix Systems, a provider of reward program services to restaurants and retailers.
“One of our analysts decided to check up on a marketing campaign run six months prior,” Wilson says. “Because of this, we were able to discover an unanticipated result of this campaign, that would have otherwise been inconclusive.”
Another good practice is learning to ask the right questions and solve the right business problems.
“Every data science project should start off as a consulting exercise — understanding the ‘what’ and ‘why,’” Miglani says. “Also, the objective on any analytics exercise cannot be to implement a tool or platform. The objective should always be designed towards the right business outcomes, and you can get there by asking the right questions.”
Data: The foundation for success
The data analytics team is more likely to succeed if the organization creates a “data foundation,” Clark says. “Technical experts in the field of data will want to see real commitment from the organization to a data foundation,” she says. “In our treasury division we ran a coordinated program to streamline and improve data quality and accessibility over a two-year period. We have seen improved employee productivity, lower technology costs, and a broader community of digitally savvy employees.”
Ensuring high-quality data should be a cornerstone of any data foundation.
“Knowing and managing your data is critical to success,” Wilson says. “Your analysis will only be as accurate as your data. When we see success with our own analysis, we are often asked to leverage this analysis into reports or dashboards, so business users can leverage these findings from day to day. If your data process is unreliable or your data incomplete, your results will be flawed, and any actions taken on them will be erroneous.”
To keep on top of fast-changing developments in analytics, ongoing education and personal development are important to maintaining a vibrant and successful team, Nimeroff says. “Data analytics is amongst the set of fastest-growing fields out there, and even though the leading-edge techniques aren’t applicable in every situation or organization, it doesn’t mean that being able to stay current isn’t important,” he says.
Paytronix emphasizes ongoing training of the analytics staff, as well as the ability to communicate results.
“Your team needs to understand the finer points of your data, know how analysis may go wrong with statistical biases, and understand how to effectively distill and then communicate actionable results,” Wilson says.
“I often tell my team that the best analysis in the world will be a wasted effort if it is not clearly understood and acted upon,” Wilson says. “To this end, keep the end in mind when working through a problem: How will business users change behaviors based on this information? This should help you tailor your approach and curate the conclusions you provide.”