Effectively Communicating with a Data Scientist

If you have the distinct pleasure of working with a data scientist, you may have noticed that communication with data scientists is a little more difficult than with others on your project teams. As a data scientist, I can tell you that is it often just a frustrating for both sides of the conversation. I have compiled some tips below on how to effectively talk to folks like me – analytical types, that spend their days mired in data and on a constant search for details and facts. We tend to think things over, comb through details, and apply critical and logical thinking before making a decision or forming an opinion.

I have first-hand experience from the data scientist side of the conversation, and have seen my fair share of people get frustrated, annoyed, and even angry when trying to speak with me. I can tell you that it is often just as frustrating and annoying for the data scientist.

The irony of this article is not lost on me. I realize that I am making general statements and providing an opinion based on a limited set of data. I have not polled a significant subset of the data scientist population. I am confident, however, that my extrapolation is accurate for a majority of that population. I am also sure there are exceptions. Try the following techniques during your next conversation, and it may be a little more comfortable for everyone involved.

1. Be Prepared. Take some time to prepare ahead for a conversation. Gather the relevant facts. You want to provide a full explanation, and present it as concisely as possible. Sometimes this may mean being prepared to follow up the higher-level conversation with documentation of meaningful data points. If you appear unable to support your position, you will lose credibility in the conversation. Don’t be afraid that you are going to provide too much information. That has never really happened.
2. Leave emotion out of the conversation. We analytical types prefer to see issued explained in a transparent, unbiased, and factual way. Anecdotes and topical examples are often lost on my data science peers. It is not that we do not appreciate personal stories or current cultural topics, but we compartmentalize naturally, and will file these things in the ‘not related to this issue’ file. Any analogous point you are trying to make, may not have its desired effect. We are passionate about data, but our critical thinking style is rooted in dispassionately examining all sides to connect relevant data points.
3. Do not be easily offended. As data scientists, we are natural skeptics. W. Edwards Deming is famously quoted as having said “In God we trust; all others bring data.” You may feel a little off-put when speaking to someone who insists on not forming an opinion on anything until all the facts are known. We are not meaning to offend, or imply that we mistrust you or do not believe that you are genuine. This also does not mean that you cannot introduce new ideas or offer opposing viewpoints during a conversation – quite the opposite, actually. Data scientists want to have all of the data, and sometimes a different perspective exposes a new way to approach a problem or a new data set that had not been previously considered.
4. Use appropriate communication channels. Generally speaking, we analytical types are not only skeptics, we tend to be introverted, have a great deal of focus, and might even be considered anti-social. This may mean that your data scientist is naturally uncomfortable having a real-time conversation whether it be face-to-face or via telephone. Your data scientist may far prefer an email or text message to a phone call. It is typically easier to deliver a clear, concise, unemotional, and unbiased message in written form than it would be orally. On top of that, your data scientist will appreciate the documentation and the ability to mull over details of your message.