Article Link

Summary of the Article

While data science is growing in industry, individuals in companies are not getting adequate credit where earned due to insufficient communication. In particular, a lack of clear communication between the (data) scientists and the executives or clients. Where data scientists have content to be shared but are unable to effectively communicate or sell their points - especially to nontechnical stakeholders. Likewise, executives with nontechnical backgrounds are unable to see nor comprehend the value being presented as they complain about the waste of funds as results fail to meet their ‘magical’ expectations. This feeds into an endless loop of miscommunication as executives continue to hire those with high technical backgrounds regardless of their ability to communicate in a nontechnical language, whereas the scientists believe that by presenting their works (visualizing) in ways the executives desire, the results lose their value and are oftentimes mistakenly interpreted. Berinato proposes the solution of hiring a mediator to handle the communication/visualization aspects, but to his disappointment, companies seem to lack the same understanding.

Common Cases of Miscommunication

Berinato depicts several of what he believes to be the most general cases of insufficient communication between company executives and data scientists. Where the higher-ups understand the value that analytical results can deliver, with little attention towards the process or delivery. In contrast, data scientists complain that their bosses are unable (or unwilling) to understand what they do and end up underutilizing their skills/abilities.

Proposed Solutions

In response to the issue described in the article, Berinato proposes the following tips.

  1. Define talents, not team members.
    • Identify the skills that team members need/have to fulfill the requirements of an ideal data scientist team. Where, collectively, the team members and their contributive skills will allow for a strong, well-rounded team.
  2. Hire to create a portfolio of necessary talents.
    • Instead of recruiting for the degree of experience or to fulfill gaps in the positions, have a high-level approach as some of the sub-skill topics can be collectively observed in one applicant. Just ensure that all necessary talent requirements for an ideal team are met rather than looking to fill the vacant positions.
    • Ideally, this would free applicants from the stress of having to be both a good communicator and an expert data scientist. Berinato claims that this division of expectations will allow for strong applicants of each, which will ultimately improve communication for all.
  3. Expose team members to talents they don’t have.
    • Diversify the settings and projects that workers are exposed to via collaborative actions.
    • Promote classes or activities that allow for introductory or basic level learning of other principles. Not to the point of experts but enough to recognize the efforts of other practices.
    • Conduct group sessions, teamwork, or mentorship programs.
  4. Structure projects around talents.

Conclusion

For a successful team:

  • Assign a single, empowered stakeholder.
    • A capable individual/group with the proper expertise and responsibility of sharing goals and communicating with project teams. Where the teams hold as much decision-making power as possible.
  • Assign leading talent and support talent.
    • Have a clear division between those who lead and those who support depending on the context of the project or task. Know when and how certain expertise is necessary for the optimal outcomes.
  • Colocate.
    • Ensure that all active contributors are physically present within the same space - especially on collaborative projects. Include a virtual component for inclusive collaboration.
  • Make it a real team.
    • Ensure that all necessary skills are represented or covered by the team members - including mediators between expert professionals.
  • Reuse and template.
    • Consider reusable templates (base, guidelines) that could be applicable to a wide range of projects. This will save time (and costs) as well as improve efficiency.

Author Information

Scott Berinato is a senior editor at Harvard Business Review and the author of several data analysis-related books. He actively works for HBR and HBR.org, with a focus on topics related to data, science, and technology. Berinato graduated with a MD from the Medill School of Journalism at Northwestern University and a BS from the University of Wisconsin-Madison.

CO2 Uptake in Grass Plants Plots

Plot of CO2 Uptake vs Concentration

Plot of the Range of CO2 Uptake per Plant

CO2 Data Table