Data Science for Product Managers

On 22nd September 2018, Abhishek Rathore, head of product and analytics at Rakuten and formerly the director of product management at Snapdeal, shared his insights with us on the burgeoning field of data science and the impact it has on the product management community.

The following are the key excerpts from our session with him:

In today’s business environment, why is data science important for Product Managers?

There is an increasing trend of hiring searches for data scientists in the past 4–5 years, and this is about to surpass searches for product managers. The skillsets of data professionals too are evolving into multi-faceted ones that can tackle both data and business aspects, thereby enabling organisations to rely more heavily on them. Traditional product managers therefore should try to stay ahead of the curve and pick up data science skills.

Why do PMs need to worry about data science?

• Conversational products are on the rise. While in India this might still be at a nascent stage, it is picking up all over world and it is only a matter of time it becomes a predominant feature of the Indian market too.
• Data scientists themselves are expanding in to the product management domain so PMs need to learn new skillsets to continue playing a role in the digital stack. They will need to learn data sciences to stay relevant going forward.
• The makeup of future product teams will also be highly reliant on data. They ideal teams of the future might always comprise a mix of product and data personnel so it is imperative for PMs to be able to be fluent in data to work in these teams.

“Product managers need to evolve to be continue being relevant”

What role do PMs play with data science models?

While classical PM role remains the same even in the data science world, a data science sphere will also be thrown into the mix in coming years. What is now coming up is an AI/ML PM role that can act as an expert translator between user/business needs and AI/ML algorithms that can achieve them.

What do you mean by digital-focused team structures? And how will they differ from what we have today?

Problem solver-teams is the need of the day. Traditionally teams work towards roles or functions, but what we are moving towards is a system where problems are identified and teams are setup to solve them. Teams having a mix of personnel such as data science, product, tech, design, etc. which help solve the problem together.

More than 50% time is spent in data gathering in big organisations. PMs play a huge role in the gathering data part while data scientists play a role in data parsing.

In your experience, how has the collaboration between a PM and data scientist worked?

It’s about identifying problems and solving them one at a time. A PM identifies the problem and goes to data scientist for help in solving it. The PM then monitors the impact of the solution and decides how to move forward.

In a startup environments where resources are always crunched, if and how should PMs decide to invest in data-centric approaches?

Data science is not always about complex algorithms. It is about identifying the right metrics and approaches. In a startup environment the requirements for heave data-centric algorithms are generally not required, and PMs should focus on simpler data approaches.

Some simple hacks to succeed at data science?

• Rule-based systems can help solve many data problems easily. These are about setting clear guidelines on how to sort and redirect data.
• Collecting user feedback is the key to learning from your data. While it is not always possible to ask your users for feedback directly, there are many ways to indirectly capture their response to a feature or service.
• Correct labeling of data is crucial for any machine learning endeavour. Without properly labelled data it is quite difficult to properly study what is happening with your data sets and draw relevant conclusions.
• Keep an eye out for false positives as they can really skew your algorithm efficiency.

How do PMs solve data problems of legacy organisations?

It is not uncommon that legacy organisations have data in different silos and different formats due to the varied working styles of different teams. To solve this companies are now adopting a technical product managers whose main objective is to help define a unified view of the organisation’s data.

And finally, what are the key traits you look for in product managers today?

• Technical aptitude
• Data analysis skills
• Problem simplification skills

This workshop was brought to you by ProductTank Delhi, largest product community in NCR bringing stories from product teams across India.

Subscribe to receive our top stories and stay tuned on our events here.