Bridging the Gap between Business & Data Science

Original article was published on Artificial Intelligence on Medium

Data Science is used to provide a “data-driven solution”. But before we can solve a problem, we need a problem! And problems come from the pain points of the different business teams. (I will be using the term “Business Teams” for the various teams which are part of every organization -> Marketing and Sales, Customer Services, R&D, Production & Distribution, IT, HR, Finance, etc.)

In my previous post, I tried to bring together the important stages of a Machine Learning Pipeline, helpful for anyone new in this field. In this post, let’s try to understand the problems in the business and what is the gap between the data science and business teams.

The business teams can collaborate with the data science team to resolve various pain points like — grouping similar customers together, retaining the customers, forecasting sales, creating marketing campaign models, recommending products to the customers, price optimization, analyzing customer feedback, etc. But there is always a little hesitation in doing so.

So, is there a gap between the business teams and the data science team?

Well, business and data science can be considered as two sides of the same coin. Both of them want the best for their company and the end customers. The only difference is, business knows their products and the customers personally, whereas the data science team knows the report card of products and customers (the data), and can provide insights on why a particular business strategy might/might not work.

So let us understand from both the sides, why is there a gap and what can we do to reduce this gap.

The Business

These are the people who’ve been running the company for a long time and want the best for the company. During the early days of trying out data science, when this field wasn’t so stable, everyone was exploring different approaches, and because of which some of the problems/misconceptions born were:

  1. It needs technical knowledge and additional effort
  2. It requires a lot of money
  3. It is a black box. Doesn’t explain “why?”
  4. Data collection is an issue
  5. I know the business and the customers (I talk to them!)

But day-by-day, the teams are working together to solve these problems, and data science teams are also gaining stability:

  1. Data Scientists are willing to step in at all the areas to understand the pain points and can deliver solutions as a product, which can be understood by the least tech-savvy people.
  2. Thanks to the Pay-as-you-go services(Microsoft Azure, AWS, etc.), there is no need to invest a huge amount in advance. You can pay as you use the services and can terminate when not needed, which makes them inexpensive but very useful.
  3. Trust me, sometimes it’s difficult even for a data scientist to understand the “why”! But a data science solution is not a “black box” anymore. We have different interpretability techniques, which would help understand, for example, why would some customers leave, whereas the others would stay in a customer churn problem.
  4. Most of the organizations have been dumping the data in their storage for a long time now (for example, transaction details, product details, customer details, sales information, etc.). And the Data Science Team can make sure that the dumped data becomes useful for the future of the organization.
  5. Yes, the business teams know their products and customers. But when the company scales up/expands, there will be a different type of customers attached to the company, there could be a different product demands at different times of the year, some promotional strategies could be more useful than the others, etc. And to understand these different aspects, the insights from data science teams can be of help.

The Data Science Team

For a data science team, the first task is to convert a business problem into a data science problem. And to make that happen, they should step in the shoes of the business team and interact continuously with them to understand the following areas:

  1. Background of the problem
  2. How are they handling it currently?
  3. Who all are impacted by the problem? (at all levels)
  4. How will the solution add value to the company?
  5. How will the solution fit into the existing business framework?
  6. Which other teams can be approached to understand the problem better?
  7. And at last, various datasets that can impact the problem?

The business needs to trust the data, and the insights coming from it, and at the same time they can use their domain knowledge to be a good critic. The data science team should understand the actual problem before wrapping it around the data. Both teams need to collaborate with each other to get the best solution out because both are aiming for the same goal.

Business will always be the decision-makers, however, with the help of data science techniques they can make those decisions more confidently and accurately

The things I wrote here are from my experience in this field so far, and I would be more than happy to get feedback and suggestions from you all! 🙂