UserTesting launches machine learning-powered Friction Detection for enhanced behavioral analytics

UserTesting, a company that helps organizations test their products and services with end users, announced today the latest updates to its Human Insights Platform. These updates include a new feature called Friction Detection.

Friction Detection uses machine learning to analyze video recordings of user sessions and identify moments when users encounter difficulty or confusion while performing a task or navigating a workflow. The feature aims to help product designers and developers pinpoint areas that need improvement and enhance the overall user experience.

The announcement comes after UserTesting went private in a $1.3 billion deal in October 2022, in which it merged with UserZoom, another user experience testing company. The merger, which was completed on April 3, combined UserTesting’s video-based approach with UserZoom’s various tools for measuring user behavior and feedback.

Andy MacMillan, CEO of UserTesting, said in an interview with VentureBeat that the merger would enable the company to offer a more comprehensive view of user experience and generate more data for its machine learning capabilities.

“The idea of the platform is to have more transaction volume and more test data, which is really interesting for our machine learning prospects,” he said.

Also Read : What we’ve learned about Sony PlayStation’s new smartphone games team

UserTesting is one of several new companies that use machine learning to augment human insights and provide more actionable recommendations for product development. Others include FullStory, which analyzes user interactions on websites and apps, and ContentSquare, which tracks user behavior across digital channels.

Why friction detection is much more than sentiment analysis

There are several reasons why a company would use a service like UserTesting in the first place.

Sometimes it’s to learn about how users feel about a product or service. According to MacMillan, often it’s also about understanding the user experience overall. While sentiment is important, as it can identify if a user is happy or perhaps angry about the experience, there are many factors that can lead to that sentiment. For example, if a user experiences friction in a process, that is, some kind of barrier or hurdle that makes the process harder or perhaps less enjoyable to execute, that’s not a good thing. Friction could also potentially mean a user is not able to complete a process like a purchase, which ultimately means less revenue for a vendor.

To date, the way that companies found the points of friction was by manually searching for them in a testing video where the user had trouble. But that’s not a scalable approach.

MacMillan said that what UserTesting discovered is that with the large volume of data it has, it could build a machine learning model to detect the friction. The model can analyze and determine where it is that the user conducting a test ran into trouble trying to complete a task. The attributes that could indicate friction are excessive scrolling or clicking behavior and other forms of delayed activities that don’t lead the user to the next step in a workflow.

“It’s one of these things where we need to boil it down to something simple, which is, the user is   frustrated and not finding what they’re looking for,” MacMillan said. “What we’re really doing is helping people to zoom in to those moments.”

How friction detection works

The UserTesting system has long had an approach known as interactive path flows, which track the user journey as they go through testing.

MacMillan said that UserTesting first overlaid basic sentiment analysis on top of the path flow with a color-coded system of red, yellow and green indicating user satisfaction. The next piece is something UserTesting refers to as an intent path. This defines the intent the user has when they are using a service, whether they are shopping or just collecting information. The friction detection is the new piece on top. It identifies where a user is struggling as they go through the path flow.

The friction detection machine learning model is a combination of using multiple assets within the UserTesting interactive path flow portfolio and applying an analysis.

“The whole goal here is to take a bunch of different assets that we’ve had available in our customer experience narratives and deliver them in a simple straightforward way to somebody who’s maybe not an experienced researcher to show them where people struggled,” MacMillan said. “The power of machine learning and where we’re going is actually to take complicated things to make them feel simple.”

Originally appeared on: TheSpuzz