How to Build a Big Data Team in Five Steps
By Anand S. Rao and Oliver Halter
Is the big data talent gap real or imagined? Some experts proclaim that talent shortages are blocking businesses from big data gains—and many companies seem to agree.
According to PwC's fifth annual Digital IQ survey of more than 1,100 business and technology executives, only 44 percent of respondents said they have an adequate supply of talent to capitalize on the promise of big data. Other onlookers contend that the cries of insufficient talent supplies are exaggerated and that companies aren’t looking hard enough and aren’t willing to pay enough.
We’re not going to agree or disagree with either group. We think we should shift the debate entirely.
Discussions about perceived talent deficits center on data scientists who bring deep statistical and analytical capabilities to the table. Big data is too big for one title to tackle. We need to build big data teams.
Forward-thinking companies are assembling multidisciplinary, cohesive groups from disparate departments and fanning them out on exploratory missions to rapidly experiment with new techniques and deploy new applications to demonstrate the value of the insights to the organization.
Here are five steps to building a big data exploratory team:
1. Break down big data talent needs.
The first thing you should do is break your big data talent needs into manageable and visible core competency components. Examine the situation thoughtfully and ask yourself, "What skill sets do we need to get started on big data?"
Typically, talent is needed in four areas:
· Business Analysis: Expansive knowledge about the business that spans marketing, sales, distribution, operations, pricing, products, finance, risk, etc.; the ability to ask the right business questions; and the skills to articulate how information, insights and analytics can determine the right course of action.
· Analytic Expertise: Solid understanding of statistical (e.g., regression analysis, cluster analysis, optimization techniques) and computational techniques (e.g., machine learning, natural language processing, graph or social network analysis, neural nets, simulation modelling); the skill to determine and apply the most appropriate techniques for different categories of problems; and the ability to communicate the business value of big data to business leaders.
· Data Technology Expertise: Keen understanding of external and internal data sources, how they are gathered, stored and retrieved; ability to extract, transform and load data stores; skills to retrieve data from external sources (through screen-scraping and data-transfer protocols); proficiency in using and manipulating large big data stores; and the ability to use disparate data sources to analyze the data and generate insights.
· Visualization Expertise: Good understanding of visual art and design; ability to turn statistical and computational analysis into graphs, charts and animations; ability to create new visualizations (e.g., motion charts, word maps) that draw insights from the data and the analytics; and ability to generate static and dynamic visualizations on a variety of visual media (e.g., reports, screens—from mobile screens to laptop/desktop screens to high-definition large visualization “walls,” interactive programs, to augmented reality glasses in the future).
2. Scan your internal landscape.
Scan your internal landscape for the aforementioned skills. You already have people who know the business, possess data-crunching capabilities and make data-driven decisions. They might not come from the department you suspected, and they might not be wearing a nametag that says “data scientist,” but they work in your organization and have the right skills.
3. Fill talent gaps.
You’re in a position to contemplate and fill talent gaps once you know exactly what you need and what you already have. If it’s true that math wizards are missing from the big data formula, consider statisticians, who are much easier to source than Renaissance data scientists. Augment teams by recruiting selected capabilities on a limited basis.
4. Cross-train to cultivate collaboration.
Once your team is in position, cross-training is critical. For example, we recently helped an insurance industry client pull together a big data team, helping them train the analytics members on insurance and the insurance business members on analytics. The goal was not to make everyone an expert on everything, but to establish a common language that the group could use to collaborate effectively.
Enabling team members to trade places also helped ensure that the group shared a vested interest in the success of the outcome. In addition, it exposed each of the team members to different work cultures. Don’t forget the importance of designing incentive structures that foster brainstorming and build a collective mindset.
5. Empower your explorers with freedom.
It’s also important to realize that you can’t throw all these people together and demand an instant return on investment. You should give them the freedom to explore big data without the pressure to produce a revenue-generating idea right off the bat. In the exploratory stage, their purpose is to investigate big data’s possibilities and to report their findings.
The thing about big data is that it constantly demands that business and technology executives sidestep their comfort zones. That includes how you think about sourcing and managing your talent.
The bottom line is that big data talent isn’t ready-made. Your big data team members will evolve over time as they learn from one another and are reshaped by the new world of big data.
Once the team is capable, it can be replicated and institutionalized in different parts of the organization. You are investing in the creation of a high-powered big data team that you can deploy across business units many times over.
Building a team that has the right skills to explore big data is part of the adventure, and it’s a process that will take time. Not because we are missing talent, but because we are seizing a new opportunity.