At the ODSC East event, held this week in Boston and virtually, Scouts Consulting Group head of data science Ken Jee diagnosed a problem that plagues the data world: the culture gap between data scientists and business/technology decision makers.
Jee suggested that there are five factors that create suboptimal outcomes when data scientists and decision makers come together:
- Communication Style
- Capabilities Understanding
- Project Lifespan
Here’s what each of these five problems look like and how to overcome them, according to Jee.
Data scientists will tend to walk decision makers through the process and data before getting to the “bottom line.” Decision makers, however, typically prefer to know the upshot before they wade into the details.
Jee recommended that data scientists use business abstracts to reach decision makers. A business abstract might concisely cover three process points, then provide the key suggestion or finding. Providing a business abstract can make it easier for decision makers to grasp the bottom line.
Decision makers can get stuck trying to understand the difference between what is realistically possible and what is ideal or perfect, Jee said. When decision makers appreciate the degree to which limited resources, limited systems, and limited data restrict the set of possibilities, they will recognize that the outcomes are bounded by the inputs.
Jee recommended that data scientists focus their communications efforts on explaining constraints and how they limit potential outcomes.
To put it plainly, data science is complex and dependent on many systems, a lot of hardware, and a good amount of time. Data scientists are used to this level of complexity, whereas decision makers are used to thinking in a different vein.
In this area, data visualization can help decision makers see the complexities at hand.
Decision makers typically think of projects as being measurable and with clear ROIs. However, given the experimental and discovery-oriented nature of data science projects, calculable ROI is often impossible to determine pre-project.
Data scientists must help decision makers understand why that is.
Here, an irony prevails: Data science projects are best when they take the longest, ingest the most data, and are subject to the most analysis -- and this does not jibe with the quarterly-reporting timelines that govern the world of business decision makers.
Data scientists can overcome this by showing incremental gains and wins and always contextualizing their progress.