Effective Communication

“What we have here is a failure to communicate.” – Captain (played by Strother Martin) in “Cool Hand Luke”

Clear, concise communication – verbal, written, nonverbal – or lack thereof represents a significant challenge in business in general upon which we all strive to improve. In the field of data science, the communication challenge is even more acute for several reasons, not the least of which is that business people and data scientists rarely speak the same language. Data scientists, who speak a language of mathematical models and symbols and “code,” must endeavor to understand the business domain and problem space that is defined by terminology and acronyms that are inherently foreign. Managers who speak a language of KPIs and business jargon and acronyms must try to understand how a complex, sophisticated mathematical model, automated on a computer, is going to solve their business problem. Data scientists must be intellectually curious and dig deep to understand the business problem and context. Managers, who are likely not mathematicians themselves, must “trust but verify” through rigorous experimentation, verification and validation that the model and solution (inside the “black box”) are functioning appropriately to solve the problem at hand.

It is human nature that people inherently abhor change and fear what they cannot understand. Famed operations researcher Gene Woolsey said, “A manager would rather live with a problem they cannot solve than accept a solution they cannot understand.” The communication challenge then requires the data scientist and manager to create a conceptual and practical intersection between their two worlds in which they can communicate and understand each other.

One of my favorites of Stephen Covey’s “Habits of Highly Effective People” in the communication realm is #5: First seek to understand, then be understood. This habit compels us to listen before we speak. The old adage that we have two ears and one mouth so that we listen twice as much as we speak provides an excellent heuristic for data scientists to govern their communication approach. Throughout a project, but especially in the early stages, data scientists should be listening two-thirds of the time and speaking one-third of the time, and when they are speaking, they should be asking exploratory and clarifying questions. This ratio will tend to change toward the end of the project to be more 50-50 listening and speaking as the data scientist explains how the model and system works; presents findings, conclusions and recommendations; and answers what will no doubt be a myriad of questions from the manager.

In every data science project, it is critical to consider the context and audience in each situation and adapt your communication approach and content accordingly. Are you, the data scientist, talking to another data scientist on your team or on the business side? Are you talking to the manager on the business side – or perhaps their up-line executive, such as a director, vice president or above – perhaps giving a demonstration or presentation of your model and solution? Maybe you are talking to a business analyst equipped with an engineering degree or quantitative MBA who has a much better understanding of data science models, computers and software applications than their manager. Know your audience, prepare and communicate accordingly.

Communication is crucial and should be engaged in before, during and after the project.

• Before the project to mutually set expectations on scope, timing, budget, critical success factors and criteria and to achieve a crystal-like mutual clarity of the problem at hand and the solution approach.
• During the project to ensure tight feedback loops because modeling is by nature and necessity iterative, and not necessarily strictly “linear,” and to provide updates on status and negotiate changes in direction or approach as new information and discoveries come to light.
• After the project to communicate and act on the findings, results, conclusions and recommendations and most importantly, to quantify the business value and economic impact of the model.

After the acceptance of the model and solution comes the substantive communication that must go into implementing the model as part of the business process (the next installment will cover this and change management).

One piece of advice on communication media: Avoid email if at all possible, especially on critical, sensitive topics. Email is a horrible communication vehicle for nuanced, complex information sharing. Data science project communication is the utmost in nuanced, complex information sharing, and email creates myriad opportunities for misinterpretation and misunderstanding. There is no substitute for face-to-face communication, whenever possible, even via video conference in the new post-COVID-19 age of remote work.

Stories as Communication

The most effective data scientists are storytellers. They tell a story of what life was like before the model was developed and implemented and how life will change (hopefully for the better) afterward. They start presentations by grabbing the attention of the audience – in particular, executives who are prone to reading the news, email or their calendar on their mobile device. The most effective data scientists ask provocative rhetorical questions such as, “What if I told you that we could increase sales (or decrease inventory costs) and make (or save) the company an extra $X gazillion using data and data science?” Now you have everyone’s attention! The key is to communicate in the language of your audience – i.e., managers, executives, domain experts – not data science!

Lastly, for any data science project to move forward, you will inevitably have to address and adequately answer the age-old question: “What’s in it for me, my team, my department, the company?” As the late NBC Sports television executive Don Ohlmeyer once said, “The answer to all of your questions is money.” The answer may be operating or capital expenditure cost savings or avoidance, increased revenue, increased customer satisfaction or increased resource utilization, all of which may lead to some economic improvement for the people involved (like a bonus, raise or promotion!) or the company at large (higher stock price, increased dividend, increased profit sharing, etc.). Everyone wants to understand how they, and their stakeholders and constituents, are going to benefit by undergoing this cataclysmic change in their business process.