How to be a great data storyteller?
Data storytelling seems to be the buzz words these days. Interestingly, even among the very experienced analysts I know, this is one area where people continue to show a lack of confidence. This is something I am also learning and discovering every day. However, let me share with you my approach, and hope that it gives you some ideas to apply to your work.
The essential components of a good story
To be a great storyteller, it is worthwhile to spend some time to understand the essential components of a good story: A setting, a twist, and a resolution. Pretty much every story you read follows this format. Let’s take Cinderella for example.
Cinderella lives a very happy life with her mom, dad and some animal friends. One day, Mom passed away. Dad re-married an evil stepmother with two evil stepsisters.
Unfortunately, Dad passed away too.
Met a prince. Married him and live happily ever after.
Let’s keep this in mind and park it temporarily for now. I want to briefly talk about the essential components of a good business recommendation.
The essential components of a good business recommendation
Unlike the essential components of a good story, which is quite universally agreed upon. Depending on who you talk to, you might get a slightly different version of what one thinks makes up a good business recommendation. In my view, there are only two things that matter: A clear problem statement and an actionable outcome. Again, let me use the same Cinderella story as an example. This time, I will treat it as a business problem that requires a recommendation.
A clear problem statement
How can Cinderella get out of her miserable life and live happily ever after again?
An actionable outcome
Cinderella should get her fairy godmother to help her to look pretty at the royal ball. Make the prince fall in love with her. Get the prince to propose so she can move to the castle and will not have to see the evil stepmother and sisters ever again.
Ok, I hear you “We’ve talked a lot about Cinderella, but I’m still not clear how to connect the dots”. One principle:
Use the setting and the twist to bring out the problem statement. The resolution is the actionable outcome.
Still feeling confused? Let’s do a deep dive on how you could write the story now. We will again, follow the exact three essential components of a good story, but I will point out some common errors to watch out for.
Use the setting and the twist to bring out the problem statement
The purpose of having the setting and the twist is to bring out the problem statement. Hence the most important thing to remember here is that you don’t want to over-complicate things. You just need to bring out enough information so that people agree with you that it is the right problem statement to solve.
If you recall the Cinderella movie, Disney probably only spends the first 5–10 minutes to build up the context. The twist? Maybe 1 min. The relative length of context and twist to the overall story is something you need to keep in mind.
The most common mistake I have seen is people spend way too much time on the context. Most analysts have a tendency to present EVERYTHING they have analyzed. Just because you have done the hard work, doesn’t mean they are worth presenting. This is my number one take-home message for you.
As an analyst, you should spend as much time (if not more) synthesizing your findings, and pick the part that is relevant and worth presenting as the actual analytical work. The rest? Keep it for yourself. How do you decide which part to present? If they help people to understand the problem statement then present it. If not, then kill it.
Let’s go back to our Cinderella story for a second here. The context we want to build is that Cinderella had a very happy life, even after her mom passed away, it is still fine as her dad has always been there. Hence anything that supports that, can go into your story. “Cinderella has a very loving family”, yes. “Cinderella has a lot of animal friends”, yes. “Cinderella is upset because her animal friends fight for food all the time”, no. I hope you get the idea.
Lastly, sometimes even the findings support the problem statement, but because you’ve already built up a very strong business statement, you can perhaps leave them in the appendix. Over-building your business case feels a bit like your girlfriend has already said yes to your proposal, and you are still telling her why she should marry you. It’s time to bring out the ring man. Once your stakeholders are already on board, move on.
Quite opposite to what most analysts do when it comes to context (they over-do it), most analysts miss out twist altogether. The twist is very important because it tells the audience “why this is a problem worth solving for”. A twist is like a trigger that calls for actions. If Cinderella’s dad didn’t pass away, then there wouldn’t be a story, because she would still be living a happy life. A twist highlights a situation that is different from the past, most of the time, an unwelcomed situation that requires attention.
As important as it is, twist by itself, did not serve too much of a purpose. Imagine, if I just tell you that Cinderella’s dad passed away. You would probably feel sorry for this girl, but you wouldn’t think that we need to do something. Because a lot of people have lost their parents, and they are still fine. It is the unique context, combined with the trigger that makes you feel “we need to help this girl”.
A good context and twist combo bring out the problem statement that gets people’s attention. This is how you keep your audience interested, and now it is time to feed them more with the actual solutions.
The resolution is the actionable outcome
In reality, an actionable outcome is never as simple as “Cinderella should pretty up and go to the ball”. The resolution is in fact, where most of the heavy-lifting analytics happens. To have the best resolution, I would recommend one to adopt the MECE approach. MECE stands for Mutually Exclusive, Collectively Exhaustive. You can google it up in more details if you want, it’s quite a useful concept. In short, it means you have to examine a set of possible solutions. Each solution is an independent idea, and together, they make up the entire possibility universe. i.e. you could not think of anything else you can do that is not already covered in what has already been proposed.
In the Cinderella example, I know the success would be Cinderella having enough money to move out of the house and be independent (from the evil stepfamily). To achieve that, I can take the MECE approach to all the possible actions Cinderella could consider:
- Can she get a job? Looking at her skillsets, singing and talking to animals, quite unlikely in the human world she’ll find a decent job.
- Can she get an inheritance? Given how naïve she is, even there’s something in the will for her, her stepmother will most likely grab her share already.
- Can she marry someone rich? Quite possible. She’s pretty with a godmother who has magical power.
As you could see in here, despite I’ve only shown you one possible outcome in my final recommendation — marrying someone rich. In my analysis, I have actually taken a closer look at all the possible actions you could take. I am only presenting to you what I think makes the most sense.
A good analyst would look into a set of actionable outcomes that are relevant to the business problem. A great analyst would take one step further, and tell you which one recommendation among all being investigated that you should take.
Again, keep in mind that your value add here is not to show the business how much time you have spent investigating each outcome, but to recommend the most relevant for them to take actions.
Wrapping it all up
Start with a business problem, and think about what sort of information you need to present to your audience so they get the overall context. Bring out the twist to get them more interested in the business problem. Finish off by presenting with an actionable outcome that will most effectively address the problem.
Note I did not mention data visualization in here at all. Personally, I do not believe data visualization being all that important for good storytelling. Data visualization is a bit like make-up. When you are already pretty, it helps you to look even prettier, but without it, you are still beautiful (the reverse is also true, when you are ugly, even with a lot of make-up, you are still ugly). Hence spend time on crafting up your story, visualization could be the final touch-up, not the focus.