Pandata Blog

AI design and development for high risk industries

There is no doubt that data science is revolutionizing the business world and the companies that aren’t investing in it are being left behind.  With all the hype around the data scientist, with skills that span advanced mathematics, database manipulation, statistical prowess, and coding, it’s tempting to assume data science falls strictly under the domain of IT. This is one of the biggest missteps we see. Data science is, by definition, a business function – the discipline of transforming raw data into meaningful information that impacts the bottom line.

According to McKinsey, data science is about much more than crunching numbers. It requires a cross-functional team of professionals ranging from business experts to software engineers. More importantly, it depends on collaboration, trust, inclusion, and an emphasis on both short-term and long-term wins across an organization.

Successful companies tend to leverage two different configurations for data science teams:

  • Embedded – where data scientists are embedded as subject matter experts within specific business units. This tends to work for larger organizations that can dedicate one or more full-time data scientists to fully understand the business needs of each business unit. A common drawback, however, is lack of communication and non-standardization of toolsets used by data scientists.
  • Centralized – where data scientists function as an internal consulting team, serving the various needs across business units within an organization. This tends to work for smaller organizations with more limited resources. It also enforces more consistency on tools and processes used to deliver data solutions. This model can face objection from skeptical business units, and the data science team must work diligently to earn trust across the organization.

While it’s tempting, and admittedly fun, to pursue the latest in artificial intelligence and machine learning trends, organizations must constantly question the business utility. Organizations must keep in mind the business nature of data science product.  At Pandata, we use the SMARTEN process to ensure data science projects consistently deliver value. Regardless of configuration, any successful data science group must be empowered to address and act on the following questions:

  • Specific – Is the current task aligned with a business goal?
  • Measurable – Can validation criteria be established in advance? Not communicating what “success” looks like is a common mishap. If success criteria cannot be established, what is an appropriate intermediary step that delivers business results?
  • Actionable – Will developing the specific data solution result in actions that impact the bottom line? If not, what is the specific business utility of the task at hand?
  • Return – Is there a quantifiable return? Is it worth the effort to invest in the solution?
  • Timely – Will there be results in time for appropriate business action? If not, how can a task be scaled back?
  • Ethical – Today’s data capabilities evolve much quicker than the legislation that governs them. Organizations who have strong internal ethics and procedures for handling data were best prepared for regulations like GDPR.
  • Novel– Data science should be focused on creating new and unique knowledge. What can be learned from other data scientists and business units? How can these solutions and lessons be integrated and extended? What API’s can be leveraged?

Not sure where to begin?  Whether you have data scientists on staff or are looking to work with one, Pandata can help you create or refine a data strategy to align with specific business goals.  Get started today!

Cal Al Dhubaib is Chief Data Scientist & Partner at Pandata.