At Sprocket, we’re big fans of using consumer insight to fuel innovation and business growth, and we’re always pushing the envelope on methods that help organizations gain competitive insight. One method we’re seeing big returns from is analyzing online reviews using machine learning and natural language processing.
Online reviews have been available for a long time, but only recently have they become accessible to analysis in large numbers. Due to advances in computational power and machine learning tactics, we can now analyze very large numbers of online reviews and uncover meaning from them
One powerful way to use online review analysis is to fuel innovation.
Here are three ways we’ve been applying modern natural language analysis (NLA) to drive business value:
- Understand customer jobs-to-be-done (JTBD). We’re fans of the jobs-to-be-done framework when it comes to uncovering and classifying customer goals. Knowing what a customer is trying to accomplish when they consider your solution as well as alternatives is a fundamental step in framing the opportunity for future product & service innovation. Analyzing online reviews allows us to pull out what customers were trying to accomplish when they purchased the product/service – and the insight is highly objective because online reviews are what people say “when you’re not in the room”.
- Know how customers are measuring success. Even customers with the same job-to-be-done have different needs in how they want to accomplish the job. Consider the job of getting to your next meeting in downtown Manhattan. Some customers want to minimize the logistical steps required (they may choose Uber) where others may want to cut down on their CO2 footprint while they do it (they may bike/walk). Performing NLA on online reviews allow us to pull out these different ways customers measure success in accomplishing their job.
- Create meaningful and actionable customer segments. The old school approach to segmentation is to do so based simply on economics (how much customers spend/use), demographics (age, gender, geography, etc.), or attitudes. While these approaches to segmentation may produce some value, they’re highly sub-optimized at best. A more appropriate way to segment customers is to do it based on their differing needs (2. above). A thorough analysis of online reviews can culminate with an objective view of how your customers should be segmented so that your segments are highly actionable for product, marketing, and CX teams.
How well do you understand what your customers are trying to accomplish and how they measure success? If the answer is somewhere around “not very well”, consider analyzing online reviews to give you an edge (just like our friends at Frontpoint and Webroot decided to do).