It can be grueling work to discover exactly what people want, particularly in the data field. People often don’t know what they need, or what analytics can do for them.

Questions Are Critical

As a data scientist, this is where you come in—and your task can be Herculean. You have to find out what your audience actually needs. The best way to do this is by asking the right questions and listening carefully to the answers.

Seems cliché, right? But I’ve learned from experience, it’s true. Part of the problem is that when people in the business world tell you to “ask the right questions,” they often stop short of telling you what those questions are. Or sometimes they’ll just throw out a few generic, hastily thought through questions, which still don’t help.

The Six Questions

In this blog post, I’m giving you six key questions that I wish I’d had in checklist form from the start. I’ve done the research for you, sifted through dozens and dozens of questions, and combined them with my own experience as a data scientist to come up with a set of six finely tuned questions, as well as a description of why they’re important. You can use them right now, today, to help you understand your audience’s needs and to produce data products that make a difference. And you can modify them according to your circumstances, if needed. (Download a cheat sheet version of these questions here.)

  1. What are you responsible for?
    You’ll find that your audience’s needs flow from their goals and responsibilities. The fundamental purpose of this question is for you to learn what these individuals are working on, what matters to them as a result of those projects, and what’s important to their higher ups, if they have higher ups. You will begin to see what stresses them out, how they see their roles within their organizations, and what specific tasks they are trying to accomplish. Ask for specific details so you can identify the drivers of their needs.
     
  2. How is your performance measured?
    Sometimes the people you speak with work for organizations that have very specific metrics they use to measure their employees’ performance. Sometimes those metrics are loose, like an employee’s ability to lead within the organization; to guide a team to complete a large, important project; or even to come up with new, meaningful strategic initiatives.Whatever these metrics are, you need to know the details of those measurements—specific, loose, or even how the people you’re speaking with think they should be measured. Knowing these measures helps quantify the importance of their various responsibilities, helping you prioritize your work toward the needs that will have the most meaning and impact.
     
  3. If you meet or exceed your performance goal, what happens?
    This one is a little more psychological, but because in data science we’re ultimately trying to help people make better decisions, it’s really, really important that we understand their psychology. What happens to your audience personally if they achieve this goal? What happens to the organization? What happens to the people they serve? Here you gain insight into why the goal matters to them, what their visions are, and why they want to achieve these goals in the first place.
     
    Bonus: Pay attention to the kinds of words they use while describing their visions of success—when you present the data to them, use those same terms and words, and they will feel like you are speaking directly to them. Using the same vocabulary in conversations with your audience that they use with you about the data product is a powerful tool. It will help you communicate more clearly to them because you are using terms they are very familiar with, and this helps eliminate confusion and bring more powerful understanding of the positive impact of the data product.
     
  4. What happens if you don’t meet your performance goal?
    This question—the exact opposite of the question above—helps you see where their greatest risks are, and it taps into what your audience is afraid of. When you design your data product, you should specifically address these risks and concerns, again, using their own language and words, so they clearly see how they can put their fears aside by using your product or your visual.
     
  5. What information obstacles stand in the way of completing this goal?
    The beauty of this question, besides clarifying your audience’s information needs, is that it makes your audience step back and reconsider the core problem. This is very, very important. Why? Because often people assume they understand the core problem, so they start a few steps ahead of the actual starting line, conceptually designing a data product before fully understanding their core information gaps.
     
    But this is a mistake. This is like a car novice telling a mechanic what to fix on a car all the while misdiagnosing the issue—it won’t fix the problem, and it will waste everyone’s time.Most of these people are very accomplished and capable in their respective fields, but they aren’t data experts. Without a full understanding of how data works, it’s unlikely their conceptual model is the best approach, which unfortunately can lead to products that are inefficient, not actionable, or impossible to produce.Asking this question in this way gives you a look at the core problems carte blanche so that you have full latitude to offer the best analytics solution for the job—allowing both of you to bring your expertise together to tackle the problem without assuming each other’s roles.
     
  6. What are you currently doing or what have you done to fill this information gap? How well did it work?
    You won’t want to finish the conversation without asking this question. It tells you several important things.First, what have they already done? If it hasn’t worked, maybe you’ll know not to go down that path, or you can offer ideas on how they could have done it differently. If they’ve already done some leg work that is valuable, there is no need for you to repeat the work.Second, it gives you an idea of how their organization or division operates and what they are comfortable with: What tools are they familiar with? How are they used to seeing data readouts (Excel, Tableau, etc.)? What are their processes? You want to change as little as possible about their workflow, because change is hard. The better you can fit your data product into what they already know and how they already work, the more successful you will be.
     

Download Cheat Sheets!
To help you remember these questions and use them when it counts, download the quick reference PDF version you want here:

You can use it as a reference, or bring it to a discovery meeting and fill it out as you talk with them. Your audience will appreciate that you prepared questions and are writing down their responses.

A Quick Request 

What is your biggest data storytelling challenge? Let us know in the comments!