Every business feels the pressure to adopt business intelligence, but many are struggling with how and where to best incorporate data scientists and analysts into their teams. No one company has an advantage so far. But the success of a few data-driven enterprises — the Amazons, the Alibabas — suggest a path forward for those who manage to navigate these waters.
Related Article: 3 Tips for Hiring Data Scientists
Separate Data Scientists From Product Delivery
One of the key issues is determining where data scientists fit within an organization. Managers often default to viewing data scientist responsibilities as just another set of developer tasks. Developing software code is familiar ground for most, so the comparison is understandable. But treating all technologists as interchangeable does a disservice to both the employees and to the business.
While they may be called on at times to create lines of code or similar tasks, programming only makes up a minor part of data scientists’ work. Data science works with statistical models to formulate solutions for well-defined problems. Data scientists must then interpret those results into language that speaks to key stakeholders and business objectives.
Managers must therefore provide data scientists with more of a research and development setting to collaborate in, rather than throwing them in with software engineering. Data science is an investment in validating data, its sources and statistic principles while allowing time and resources for exploratory studies on those data sources and statistic principles.
Related Article: 5 Communication Skills Every Data Scientist Needs
Inspiration From Outside
To better picture how this might work, look to other industries that produce engineered products for inspiration. Many of these companies have a dedicated advanced research organization, which reviews technology before helping disseminate that tech in the product development.
The automotive industry is one example of this in work. When I was an automotive engineer at Ford, I worked in the advanced engineering center to develop vehicle technology independent of the vehicle development teams, to get the basic functionality of the technology right. The vehicle development teams would then refine the tech to fit the vehicle and meet cost/performance targets. The separation of responsibilities allowed getting the core technology right without the weight of organizational complexity.
Having a dedicated data science team that’s separate from the customer-facing business, creates the right collaboration environment where professionals can work through the problems at hand without the distractions of other lines of business. The team can then relay these answers to the customer-facing parts of an organization.
Room to Explore
Data science is effective when the team can perform a dedicated exploration before applying data in production. Separating the exploration from production allows the teams to test the data and hypothesis before setting it loose in the wild, a step any business exploring machine learning initiatives will recognize the value of.
Data science can be marvelous in what it can reveal, but to reach those insights, data scientists need room to explore. Looking to how industries with engineered products manage their technology initiatives can offer innovative ways to manage a data science initiative. Giving your organization the opportunity to explore the nebulous aspect of data science can strengthen how it delivers products, services and customer experience with data.