We’ve talked a lot about how companies can use machine learning and predictive analytics to make better creative business decisions, but we haven’t specified how to develop products based on user data. Product Managers—this one is for you.
Product Managers Have to Figure Out How to Innovate With Data Science
With data science being such a big part of understanding how customers interact with your product, Product Managers or Head of Products are now given the task of quickly finding solutions with their data and integrating machine learning into their workflow and their products.
A past post we did on how 5 startups are creatively using machine learning to solve 21st-century problems, told a story of how these companies are creating data-driven products. These startups all have great product teams that innovated their industry by adding AI to their products, but what the post didn’t tell was how these startups were able to tap into and read their data to be able to apply it when developing an innovative product.
Product teams are forced to interact with data science to gain a competitive edge, but may not have the resources or budget to bring on a full data science team. This responsibility now falls into the lap of the PM.
But don’t worry you, product pioneer, you. We will try to enlighten you on turning data science into a skill you already have: User Science.
Understanding User Science
We touched upon how the Product Manager is in charge of adding innovating tech to their workflow, but the modern PM should also be able to use that tech to tap into their customer’s wants and needs. This is where user science comes in.
What is User Science?
User science is essentially the study of understanding how the user interacts with your product or company. It starts with data. Capturing the churn rate, purchase data, customer personas, etc., are just some of the aspects of user data that gives you a better understanding of user science. Another aspect of user science is creating predictions on what your customer will do if such and such happens or if you modify your product in certain ways.
A part of a Product Manager’s job description is interpreting customer behavior and this isn’t a new discipline. Each PM has explored several learning curves to implement innovative tools into their products.
The 4 Learning Curves of User Science
Luckily, for Product Managers, there are many tools and techniques to understand customers better—almost too many. PMs have to know which tools will work most effectively and they can draw value from.
Brent Tworetzky, SVP of InVision, divided user science into 4 learning curves:
Awareness - PMs should learn what the tools are.
Quality - PMs should learn how to use the tools correctly.
Appropriateness - PMs should learn when to use the tools.
Value - PMs should learn how to use the tools effectively.
Again quoting Tworetzky, “User Science includes the vast fields of User Research and Analytics. Product User Research encompasses a range of tools to directly learn from users: hands off observation, direct inquiry, putting products in users’ hands, and the many forms in between. Product Analytics ranges from A/B testing to exploratory analysis (searching for patterns) to predictive analysis. Both fields specialize beyond Ph.D. levels, across such titles as human factors engineering and usability research on one side, to statistics and data science on the other.”
For this post let’s focus on the product analytics side, encompassing data science and statistics.
How to Turn Data Science into User Science
The obvious answer is machine learning and predictive analytics. Plugging customer data into predictive models can help PMs understand user science in a new way.
We talked about how to predict customer churn, various price points, or even ways to predict where you should set your Airbnb room prices in different neighborhoods around New York City. There are just so many possibilities to hack user science with predictive analytics, making machine learning extremely valuable for Product Managers.
In terms of products, PMs can harness behavior data and make predictions to improve how users interact with their products.
Some Use Cases
- When PMs conduct interviews or surveys to collect data, they can collect quantitative and qualitative data to train an algorithm to analyze the relationship between intent and action.
- Using product data, you can see which customers are most or least likely to make an in-app purchase based on purchase data and optimize your product offering accordingly.
- To better understand a user’s needs, PMs can A/B test their subjects and use the data to predict which users are likely to choose an option over the other.
- PMs can see at which touch points a user is likely to churn and predict exits based on demographic data and optimize their products to better fit the persona who churns.
Become a User Scientist, Not a Data Scientist
The average Product Manager has a lot on their plate. They’re trying to innovate, manage teams, and build the best product possible. They don’t usually have time to be a data scientist and interpret user behavior. PMs should also know where to allocate resources. Taking the time to hire a data scientist to build and train models cut down on the agility and resources of a company.
The Product Manager should focus on user science and actions focused on making the product better. Implementing quick machine learning tools into your workflow to create a better product is essential to a great PM.