The most underrated skill in Data Science: Communication

In June, I had the pleasure of giving a keynote at Women in Data Science (WiDS) Dublin. This blog post is based on that talk.

Some people refer to communication as a ‘soft skill’. I really dislike the phrase ‘soft skill’. It’s confusing, inaccurate and implies that there’s little rigour in learning, improving and applying what folks consider “soft skills”.

I believe that communication is one of the most critical skills in data science. Many people in our field focus on skills like programming, machine learning, or statistics. Yet, the ability to communicate effectively is just as important, if not more so, to driving real impact in data science and having a successful career in this field.

Here are 5 reasons why, along with some tips for improving.

1 Communication helps you understand and translate business requirements

Impactful data science projects are aligned to broader business objectives and stakeholder needs. Communication enables data scientists to gather all the necessary information, clarify needs and expectations of stakeholders, and align their work with broader business goals. It helps you understand and translate high-level business requirements into specific data problems and research questions. Doing this requires engaging with and actively listening to your stakeholders so that you can deeply understand their goals, priorities, and challenges. Only when you do this can you truly understand what your organisation and business needs from a data point of view.

Tips for improving:

• Listen actively and ask lots of questions of your stakeholders. Dig into what they want, what they need, and why they need it.
• Make sure you invest time and energy in understanding the unique dynamics of the business in which you are operating.

2 Communication helps you frame or reframe problems

Data science is used to solve tangible business and/or customers problems. To solve tangible problems, Data scientists must also be able to frame problems in a way that is meaningful. A well-framed problem sets the team and project up for success. An ill-framed problem has the opposite effect.

Framing a problem is all about expressing a problem on behalf of those who are experiencing it, in a way that a whole group or team can get a shared understanding of it, rally around it, and take action on it.

Doing this requires effective communication with stakeholders. In particular listening carefully to what they ask and probing to understand why. It involves digging in to clarify the nature and scope of the issue at hand, the causes and consequences, the possible actions and outcomes. This helps data scientists focus their efforts and ensures that any analysis is targeting the right problems.

Tips for improving:

• Never take an ask or problem at face value. Data science teams are often inundated with asks and there is rarely a shortage of problems to tackle. The trick is to ensure you are focused on the right problems — problems that are well framed, clear and will solve a tangible business or customer challenge. Always seek to unpack and understand the problem at hand.
• Apply a framework to help you unpack the problem space. One is the 5 whys which encourages you to go broad and explore the underlying cause-and-effect of a particular problem to get to the root case. Another is the 40–20–10–5 rule where you explain your problem in up to 40 words. Then cut it down to 20 words; then to 10, then finally to only 5 words. These 5 words are the root of your problem.

3 Communication helps you collaborate better

Data science projects often require working alongside other data scientists as part of a team as well as collaborating with cross-functional teams of software engineers, designers, product managers, researchers, or executives. Communication plays a vital role in facilitating effective collaboration.

• It enables team members understand each other.
• It helps folks coordinate efforts so that everyone understands who’s contributing what.
• It encourages folks to share information and ideas, and discuss issues openly which helps teams arrive at innovative solutions more efficiently.
• It provides an opportunity for team members to give and receive feedback which promotes a culture of continuous learning and growth.
• It can help address conflicts in a constructive manner by encouraging open dialogue.
• It helps team members work together to achieve common goals.

Tips for improving:

• Establish or co-create guidelines, conventions, principles for your team and/or stakeholders on how to communicate and work together. This could include things like when and how folks will host meetings, what channels you’ll use for communication, etc. This is sometimes called “contracting as a team”. It helps folks understand what’s expected of them and helps avoid making assumptions when working with others.

4 Communication helps you present results and insights effectively

Data and insights can be powerful, but only if they are presented in a compelling way. Effective communication skills can help you tell stories with your data, making it more meaningful and impactful for your stakeholders. Data storytelling is a structured approach for communicating insights which typically involve data, visuals and a narrative. Narratives help explain what’s happening in the data and why a particular insight is important. Visualisations help highlight key insights.

Data science projects also involve complex data sets, analytical models and technical concepts. Often data scientists are communicating these technical concepts and complex models to non-technical audiences. Effective communication can help you avoid misunderstandings and misinterpretations that can arise when technical concepts are not explained in a clear and understandable way. Great data scientists can bridge the gap between technical and non-technical stakeholders, ensuring that everyone is on the same page, communicating methods and results in a transparent and understandable way which helps ensure the work is actioned and drives tangible impact.

Tips for improving:

• Take time to understand your audience and tailor your message and zoom level accordingly. For example, how you present insights about a predictive model of churn is likely going to be very different when sharing with a sales team vs. your executive team.
• Always include a TLDR; which stands for “to long don’t read” or executive summary that emphasises the ‘so-what’ to help readers understand what key messages and findings they need to take away.
• Incorporate relatable examples, anecdotes or customer stories to bring your quantitative insights to life. Data scientists often forget this part as we tend to be laser focused on the large-scale, the numbers, the metrics. A great way to tell a more compelling and holistic story is to bring your quantitative insights to life with qualitative nuggets.
• Choose the right visualisations to support your narrative. Most data scientists immediately think of charts or graphs but sometimes simple text suffices in visualisation insights.

5 Communication helps you influence and persuade

Data science is used to drive effective decision making. It’s used to de-risk decisions. Great data science teams are those who drive actions and decisions through their insights. The harsh reality is that insights without actions or without decisions are essentially worthless. And inspiring action involves influencing.

Here’s a great reference by Deakinco which is a workplace education company on the relationship between communication and influencing.

You can communicate without influencing but you cannot influence without communicating.

To influence, you need to be able to get your audience to understand the benefits of a particular change or idea and get them to take action in some way. Effective communication helps you influence and persuade others to take action or drive to a decision.

Tips for improving:

• Build credibility and trust. This doesn’t come overnight. It takes time. With each project you’ll build a little more credibility and a reputation for being someone who can be trusted to deliver great, high impact work.
• Anchor your work and your interactions with stakeholders around decisions, in particular their default decisions. Let’s say your stakeholder has a problem, and they want your help solving that problem by answering a specific question to help them make a decision. Before embarking on any research, ask them what their default decision is. That is, what would they do right now, knowing what they know now, without any net new information? This helps give you a sense of how they think right now as well as what type of information or insight is needed to change their default decision. Anchoring on decisions, talking about decisions, encouraging folks to think about their default decisions will help you better influence those decisions.
• Be as clear, concise and confident as possible. The confident part can be especially hard but it’s hard to be persuaded by someone who lacks clarity, is overly verbose and who comes across as not being confident in themselves or their insights. While easier said than done, I find that being prepared helps me feel more confident, which helps me communicate more confidently.

I’m not suggesting that communication is the only thing that matters in data science. Or that technical skills don’t matter. But in my experience, communication is the competency that is often the difference between a truly high performing data scientist and a data scientist who is simply good. The difference in a truly high performing data science team and a data science team that is simply good.

By communicating effectively, data scientists can ensure that their work aligns with business goals, that they are working on the most impactful problems, that they collaborate successfully with others, that they present results and insights effectively, and that they ultimately influence decisions and actions.

To wrap up here’s a great quote from Warren Buffett:

“The one easy way to become worth 50% more than you are now is to hone your communication skills”

Karen Church is a Research & Data Science leader who loves building highly engaged, high performing teams who use data and insights to drive effective decision making. She’s currently VP of Research, Analytics and Data Science (a.k.a. RAD) at Intercom. Intercom is a B2B SaaS company that offers a complete Customer Service Solution. Karen’s team partners with leaders and functions across the company. They combines quantitative and qualitative approaches to extract insights that shape strategy and drive product and business decisions. She’s been based at Intercom for over 7.5 years, helping scale the team from a couple of folks to over 20. Prior to Intercom she spent 8 years in industrial research labs (Yahoo Labs & Telefonia Research) leading teams of scientists and conducting research to drive mobile product information. She’s hired, managed, mentored and helped develop many data scientists, researchers and insights leaders over the years. And she’s seen first hand how effective communication can accelerate a data team’s performance and in turn drive more impact for a company.