Machine Learning: When Primary Research Becomes Secondary

By Carmel Dibner

What's your go-to strategy when you’re in-search of actionable insights to inform new product development?

If you’re like most companies, the answer is primary research. Primary research among your target audience allows you to gather detailed insights around specific research questions of interest. It’s a tried and true technique for uncovering impactful new insights, but it’s no longer the only option. The alternative? Mining existing data using machine learning.

In collaboration with researchers at the MIT Sloan School of Management, Applied Marketing Science (AMS) has used a new machine learning algorithm to mine big data for insights. It can be used on the abundance of data about your product or service that already exists including online product reviews, user forums, call-center data, open-ended survey responses, and much more. The result? A detailed list of insights in customers’ own language.

In a pilot study, AMS used machine learning to uncover a full database of customer insights around blenders. Specifically, the machine mined through thousands of user comments from product reviews written about these household appliances and identified informative comments detailing consumer requirements and unmet needs. We used the output to develop a full list of attributes and considerations that consumers have when it comes to blending -- ranging from blending quality to cleanliness, odor, noise, health benefits, ease of use, durability, and more.

Both the number of insights and quality of insights collected through machine learning were consistent of with what we typically observe in the hundreds of primary research studies we’ve conducted throughout our 30-year history. And the best part? Compared to primary research, machine learning is faster and a fraction of the cost.

While machine learning isn’t appropriate for all research questions and all product categories, in cases where a rich data source exists, it’s a great place to start. Primary research can be used to supplement machine learning, helping you get more mileage out of your research.

Don’t miss our webinar on demand where Carmel Dibner provides an overview of how machine learning can be used to uncover new insights quickly and cost-effectively. She also discusses tips on how to incorporate machine learning in your market research toolbox to stretch your research dollars.

Think you’re sitting on a goldmine of data waiting to be tapped? Contact us to learn how you can use our cutting-edge machine learning algorithm to uncover valuable insights.