Many organizations think that data science is solely about crunching numbers. Put a bunch of analysts in room, give them access to the data, and within a reasonable period of time, they’ll report back with their numbers and graphs revealing deep business insights. These organizations often assume that the numbers tell the true story. After all, “numbers don’t lie,” right?
Unfortunately, numbers and other factual information are often used to tell lies and spin stories. All you need to do is turn your attention to Washington, D.C. to witness the two competing political parties using the same data in an attempt to support conflicting narratives. For example, one party cites the Mueller report stating that it contains no evidence of collusion, while the other party cites the exact same report and claims that it does not exonerate the president. Yes, data can be used to bend the truth. Studies may arrive at false conclusions. People can misinterpret data. Sometimes, the data itself is misleading or incomplete.
Data science involves more than just crunching numbers or conducting detailed analysis of an issue. It also involves stepping back to look at the big picture, looking at issues from different perspectives, and using human intuition to make big leaps in thinking.
Combining Analytical and Conceptual Thinking
Mining data effectively for knowledge and insight requires that the data science team, along with others in the organization, engage in two types of thinking:
- Analytical thinking: A methodical, logical method of breaking down complex issues, gathering relevant data, examining the data for patterns or relationships, and drawing conclusions to solve problems or identify and capitalize on opportunities.
- Conceptual thinking: An intuitive, inductive method of applying one’s overall understanding to see solutions or possibilities that more analytical thinkers would be likely to overlook.
The process of assembling a jigsaw puzzle engages both types of thinking. You look at the picture of the assembled puzzle on the box to transfer the concept to your brain. This conceptual understanding gives you the big picture view that enables you to figure out the general position of each piece. You then analyze each piece based on its color combination and shape to figure out more precise placement.
Data science teams work in a similar manner, alternating between conceptual and analytical thinking. The research lead on the team, who generally has a broad knowledge of the organization and its business intelligence needs, tends to engage more in conceptual thinking. This person asks compelling questions intuitively. The data analysts on the team then collect and analyze the organization’s data to answer the questions. These answers often lead to follow-up questions, and the process continues until the team discovers actionable knowledge or insight.
Nurturing Conceptual Thinking
Data science teams tend to struggle more with conceptual thinking than with analytical thinking. When asked a question or presented with a problem, data analysts have numerous tools and techniques to mine the data for answers and solutions. However, they often have no idea what questions to ask or what problems need to be solved. It takes someone with a curious mind, a knowledge of the organization, and a broad knowledge of the world in general to ask compelling and relevant questions.
An interesting book on this topic of conceptual thinking is A Whole New Mind: Why Right-Brainers Will Rule the Future,by Daniel Pink (Riverhead Books, 2005). In the book, Pink argues that we’re near the end of the information age — that focusing solely on numbers and reports isn’t that valuable. The real value will come from the knowledge that we create. He calls this “the conceptual age.”
To prepare for the conceptual age, those engaged in data science will need to develop a new set of conceptual skills — what Pink refers to as the “senses” of the conceptual age. I’ve adapted these senses into three team values that all data science teams should embrace to encourage them to think more conceptually.
- Storytelling over reporting: Your data science team should strive to deliver interesting stories or compelling narratives about the data. Think about the data as characters in a play. Ask why they are doing one thing instead of the other, and then ask questions about their behavior.
- Generalists over specialists: Organizations often populate their data science teams with specialists, who are great when it comes to math, statistics, programming, and database management but have limited knowledge of the organization, the industry in which it operates, or the world in general. They’re great at analysis, but often poor at thinking conceptually. I recommend having at least one person on the team who’s a generalist and involving others in the organization, on a temporary basis, to bring in different perspectives.
- Empathy over certainty: Knowing what motivates people is one of the best ways to come up with questions about your data. Your data science team will want to understand what your customers think and what is important to them. Remember that data science can analyze the actions of millions of people. If your team can understand what motivates them, they can ask more interesting questions.
Remember that your data science team will have to use an entirely new set of skills to succeed. To ask good questions, you have to think conceptually. Try to use these team values as a way to remind yourself that data science is not solely about analysis and reporting. All members on your data science team need to use their conceptual thinking skills to ask good questions and create organizational knowledge.