Asking the Right Question?

Within Data Science, we have all herd the phrase “it’s about asking the right question”, but the more I work with teams the more I have realized that most people have simply latched onto this phrase without truly understanding it. The right question is not “the one that came from a Sr. Leader”, there is more to it than that, and people really struggle with this concept.

Data Science and leveraging big data needs to be done with precision focus and a clear cut goal in order to be successful. Without this focus and identified goal, it’s easy to get lost in the data with so many possibilities, ideas, and shiny things that sparkle….but do not accomplish your goal, what was our goal again? When looking at data in general, even a small set it’s so easy to get off track, and even experienced analysts can sometimes get caught up on the politics, or pressure to perform and deliver insights, any insights to prove that there work has merit. The trap here is wondering if you are presenting the right insights, and here lies the point of the article here I am trying to write.

When you are starting an analytics project, you really need to focus on the overall purpose, and tear it apart to its very need and how it will impact the organization. This can be a difficult task when it was a Sr. leader who had asked for the analytics project, but this is your role in data science, and that it to challenge the purpose of the question being asked. This will help you as the analytics to understand why it’s important to the organization in case you fine small wins during the experiment, but it also throws the ball back into the Sr. leader’s court to justify the use of resources. Most of the time when you do this, the Sr. leaders appreciate the challenge, and get involved to become more a part of the experiment (but not always). In some cases as we break it down, I have had people have second thoughts about the benefit to be realized, and realize it was just some nice to know information.

Another angle you can take is “how do we measure the success of the experiment” how do you as an analyst or data scientist, know you have delivered something of value to the Sr. Leader? Sure you can offer up a number to them, but where is the value behind it?

1.      The answer is 25……

2.      The answer is a qty of 25, up 10% from the previous week accounting for 1.34% of the total gross YTD to an accuracy of 15%. We are hoping to attain a 5% accuracy raking by year end if warranted.

What answer is better? Obviously the 2nd, but you would never offer that information unless you had the discussion with the Sr. leader about what would identify this as a successful experiment.

Let’s role play a little bit here…and have a Sr. leader come to you and ask, I want to know “how many red apples we sold last week”. Seems like a pretty easy task, but you start to ask questions like…

1.      Do you care about all apples, or just red apples? Can we compare them?

2.      How many other kinds of apples do we even have?

3.      Why red? All red, gala, and delicious….others?

4.      Why are we only looking at last week? Should we look longer term, or was there an event that sparked this question?

Ok this could go on…but what you’re now doing is building a relationship and bringing this person into the experiment. They feel more comfortable, they want to help……but even still with these questions..

Were any these the “right question”?

The answer….no, there all just interesting questions….none really drive actions or even answer the big one…..”So What”? You need to ask, if I get you the information you’re looking for “the qty of red apples from last week”, what are you even going to do with that? Is there a target we should be meeting, or is this just something nice to know? If the Sr. leader now knows we sold 13 apples……..”So What” and this is where we get closer to the right question. Why does that person care, and what's the impact to the business, knowing this bit of information?

How does selling X apples impact the business from a strategic perspective?

If you get to this point when asking the above question and the Sr. leader doesn’t know the answer….it will start to make them now think themselves if this is a good use of company time. Now if you tell them it will take you 3 days to get the answer, they will again start to question their motives, and next time they will be more prepared too when looking for an answer. Not all analytics questions warrant your time, many can be a complete waste of your skills….and this is not anyone’s fault, it’s simply a sign of a less mature organization in the ways of data and analytics.

Truthfully, even the above might not be the “right question”, but this is a hypothetical situation with a made up company selling apples. Hopefully your own company has more depth….but I hope the example gets to the point on how you need to dig in and think of the larger picture. Leveraging tools from business analysis are great like using the 5 whys or SMART will help you get to the point as well, proving that you don’t need to be a PhD in stats to be involved in data science.

Don’t get me wrong though….some Sr. leaders will come and ask you to just get them the numbers, but that’s not data science….that’s just order taking…and unfortunately, sometimes part of the job.

This all goes back to building strong core relationships with your partners so you can have some hard conversations. No one wants to waste time and resources, on silly questions, especially your Leadership teams. It’s easy to get caught up in the actions of daily work, but with analytics you will always get better results when you take a step back, refocus your efforts and align to a clear target with tangible benefits. 

I hope you like my post, it was just something on my mind that I see a lot of in the industry.

John Maschke