Published by: Ujjyaini Mitra

Last week I wrote about ‘What takes a Data Scientist to be a Great Data Scientist’ and I mentioned that ‘Knowing WHY we do what we do’ is the key differentiation and this drives ‘How we work’ and which in turn determines ‘Work culture’ of an organization. It starts with you.

Many people reached out to me post that and specifically asked 2 questions –

  1. Do leaders themselves know WHY, so that they disseminate to their junior team members?’
  2. ‘How to effectively ask the WHY question?’

If a Data science leader or senior member doesn’t know WHY, then I suggest they practice asking it. Thus, it all boils down to that one question – ‘how to ask a WHY question?’ So, today’s post is dedicated to that.

First let’s understand why it’s so difficult to ask ‘WHY’ question.

  1. Surprisingly enough, when an analyst or data scientist asks ‘WHY’ question to a Sr member or to business functions – they end up taking it to their ego! If it’s a highly hierarchical organization, you may often hear back the answer ‘Because I have asked’!! In such situations people are scared to ask a WHY question ever again.
  2.  Many times, you get the answer like ‘We are not sure, we want to see the data and think what we can do’ – a purpose less exploration, which may end up boiling the ocean. And if you are specially in a big data industry, then you know how burning experience this could be, because, in this manner, you are searching a needle in a haystack. And you don’t know when and how will you find it. [Note: I am not saying that bottom up analysis is not beneficial, but one need to plan when to leverage this, as this requires a luxury of time and computation cost]
  3. At times, they just repeat the same thing again in another way. For example, you found that month on month renewal rate of subscription is 40%, and they asked you break it down for people less frequent to platform vs high frequent to platform. You ask ‘WHY’, and they reply ‘See we want to know if high frequent people are renewing better than others’. That’s not an answer. That’s an obvious Pareto statement! In such cases, clearly the person also may not know the true WHY.

Because, to answer WHY question, one need to have clear vision, and that comes, only when you know – what business problem you are solving for. And most of the times, a clear vision is rare to find.

So, You as a data scientist, or a Sr person in Data function – what do you do now? Ask the why question a better way. Here are 3 tips for you.

Never ask ‘WHY should I do this analysis or WHY you asked to do this analysis?’

Instead ask – Could you please share the background or context of this analysis? This would give me a bigger picture, which would allow me to think better. And knowing the background I might be able to bring in additional analysis which would help you solve it better. – So, you place yourself as partner in the game, and the other person find a support in you. This helps to get rid of the ‘Egoistic irruption’. This also helps you build the rapport with the person for future conversation. Always remember, Data Science and business functions have to work hand in hand, as without your intelligence they can’t take informed and objective decisions, same way, without them you won’t be able to put your best work/ model to action. So, rapport building helps in getting the WHY answers.

Go agile in case it’s Boiling the Ocean

If it’s really the case of boiling the ocean, don’t argue with teams. Let’s understand what happens in such situations. Team may have asked you to do 20 different analysis, and you know this will take good 2 weeks’ time. Don’t wait to finish all of them before you share the results with teams. After 2 weeks’ of hard work, when you show case the results, they may only discuss till the 3rd analysis, which brings a great insight and they want you to do the remaining deep dive for a specific segment only. E.g. You have been asked to do various consumer behaviour analysis. 20 points which business asked, and you have done. One of the early analysis shows – people who come to the platform through a specific source never returns, however, that’s a source which marketing team uses to onboard 40% of the new users. Now, instead of looking at remaining analysis, they asked you to do all the deep dive only for this segment of users…. For obvious reason you feel frustrated and you feel that if you knew it earlier, it could have saved more than 6-8 days’ of your hard effort…

So, in such cases, Go AGILE is the mantra. Every day or every alternative day, start sharing the analysis, as much done and spend 30-60 minutes in explaining what your analysis telling. This is another good way to build rapport with business functions and they end up valuing you more, as a data intelligence partner in the game. They will call you more often in their discussion now, and you save lots of your effort and time in doing specific analysis which has direct connection to business problem.

Help them answer your WHY question

In the 3rd situation, help them by asking same question in a different way. Suppose they say 'What do you mean by context?', you could say ‘Well assume you have results from all these analyses tomorrow, what will you do with it? How will you use it?’. That question helps them think in right direction. You may take one analysis at a time and help them answer this question. You will find at the end either they have a clear story line, or it’s all random.

If a story line emerges, then bingo, you have your answer with you, start the analysis. This collaborative effort would give you a clarity where and how to start with. If you find it’s totally random, seek help from someone senior – from your Data team or from the business function. Instead of mentioning time and effort – mention that all these analyses will take huge computation and that would cost X amount. Are we okay to go ahead? Or should we re-consider? If they say ‘It’s okay, do them’ fall into the previous Agile option.

I am sure you have come across many such situations, ultimately, remember 2 things:

  • Never give up. Keep going back and keep asking. It takes time for this culture to be built, so hold your patience.
  • Do not always expect that the other person knows it all. If they knew, they would have shared. So, go Agile and help the vision unfold.


Source: Ujjyaini Mitra via Linkedin