10 Tips For Data Scientists To Ace The Discovery Process Before Starting A New Project

The bad news is that Data Science projects tend to have problems.

Photo by charlesdeluvio on Unsplash

The good news is that most problems can be traced back to a very specific point in time: the design part of the project.

At this point, I have led hundreds of data science projects across many industries and countries. The consistency is remarkable: if something isn’t clear from the beginning, the chances that the project ends up having trouble are enormous.

One way to avoid this is to have a thorough interview process with your client. The three most important takeaways from that process are:

• Understanding what the customer really needs (not what they say they need).
• Understanding how a solution might integrate with their processes.
• What does success look like for them.

Here are 10 tips that will help you get the most out of the customer interview process and not fall into traps:

1. Ask Why. Never take the client’s request at face value. Ask why they need what they are asking for, ask how are they going to use it, understand the request from the inside out.
2. Encourage stories. Never ask binary questions. Let the customer tell you the story of the problem they want solved.
3. Never let the customer average out information. Don’t ask how they “usually” do things, ask for specific instances (e.g. “tell me about the last product you launched”)
4. Look for inconsistencies. They often reveal interesting things.
5. Pay attention to nonverbal cues.
6. Use silence. leave space for the customer to tell you what they are really thinking. Don’t feel the need to fill every second of the interview with questions.
7. Never lead the customer. “How do you plan to lower consumer churn?” is a better question than “So you want a predictive model to anticipate consumer churn, right?
8. No more than ten words per question. This shouldn’t be a heavy restriction if you prepare properly
9. One question and one person each time.
10. Have someone or something to capture the interview. It’s hard to ask questions, pay attention to nonverbal cues and also take notes. It’s better if you have someone who is focused on the taking notes aaspects of the interview.