Launched in May, 2016, the SessionCam Customer Struggle (CS) Score initially used a gesture-based approach to measure user behavior on a website. While this detection method had proven to be successful, we have now updated the CS Score to continuously improve its accuracy by measuring web analytics and error data.
The new update…
- Uses more ways to calculate the customer struggle drawing data from not just gestures/user behavior, but also web analytics and errors too
- Has access to, and conducts analysis of, thousands of new data points to predict what is considered ‘normal’ user behavior, to calculate each session’s ranking with more accuracy
- Uses recent traffic patterns and other data produced by a website to continually relearn and refocus on the areas that may indicate struggle
- Enables you to review a recommended list of web pages that need to be optimized in order of priority
Read on to find out more about what this means for our clients…
The original concept
Our initial customer struggle detection method was designed to answer the question: Can we detect customer struggle from user behavior alone?
Our Chief Product Officer, Richard Churchill explains in more detail: “Originally, we wanted to bring something new to session replay. We wanted to give our clients the ability to detect customer struggle in an automated fashion to focus efforts on watching only those replays that demonstrate what is obstructing a customer’s interaction with your website.”
Working in the background, SessionCam’s CS Score is able to analyze thousands of customer sessions and use this data as a framework to keep learning about user behavior on mobile and desktop devices. Each session is then ranked high-low, a score of five indicates the highest level of struggle, with a score of zero indicating no struggle.
“To calculate this, we use a machine-learning algorithm,” says Richard. “A method to evaluate and score the likelihood that a user is struggling, the algorithm handles vast amounts of data from individual sessions and automatically gives behaviors – such as mouse position and scroll speed – a weighting depending on how strongly they correlated with customer struggle factors.”
The new and improved CS Score
Instead of only analyzing user interaction from mouse or touch gestures to make the struggle predictions, the CS Score now takes into account factors such as dwell-time, distance, speed of navigation, form interactions, clicks and the cause and effect of errors. “Feeding this new data into the algorithm means we can better describe what is ‘normal’ behavior for each web page, and what effect these two sources have on the user journey. This new information, in addition to the gestures, gives the algorithm a physical grounding in each session and we can make better decisions on the customer journey,” details Richard.
The introduction of these two new data feeds helps to provide clients with the ability to both optimize individual touch points and the end-to-end customer journey. “This is known as journey analytics, which embraces the ever-evolving nature of customer service,” he adds.
Tracking data from these three feeds allows us to produce features that give an understanding of how individual customer journeys are affected. “What is a problem for one customer journey is not always a problem for another, even when the same behavior is present. One may be using a different device, for example, so the reason behind the struggle could be different,” Richard highlights.
Knowing more about the journey gives our machine-learning algorithm more information to work from to continually relearn and refocus on areas that may indicate struggle. “It evolves alongside any changes you make to your website over time and learns what is expected user behavior and what is a struggle,” he adds. “For example, the algorithm can draw upon previous experiences with each URL/site/client based on the three feeds to determine how a user behaves on each page compared to the users before them.”
What do we mean by errors?
Errors and messages in a journey can be useful indicators of where users may encounter struggle. However, errors themselves are not the whole story. There is a broad range of error types, and the ways these are interpreted may or may not impact a journey.
“How many times have you been asked to re-enter your postcode, or look up your address and encountered an error? The CS Score can now observe the cause and effect of these kinds of errors and messages, giving each one an appropriate weighting to identify struggle areas that are often missed,” explains Richard.
These are measurable / tangible things that happen when on a page, such as how far the mouse moves and how long is spent on a page. How these are stacked up and balanced is the product of extensive research and is a close-kept secret.
CS Score in-action
International insurance group, Ageas is one of many clients already citing our CS Score as an invaluable customer experience analytics tool. Using the CS Score, the company has been able to find high-value insights faster and focus their time and efforts on fixing the most important problems.
“With SessionCam, we can sit and watch people scrolling around, clicking, and typing into fields on the website,” says John Crichton, Head of Ecommerce at Ageas. “It’s highlighted UX problems which lay in our court, but it’s also revealed some extraordinary user behavior, things you’d never guess people would try to do.”
SessionCam is also used by Ageas’ retail partners, one of which enjoyed a 2,700% return on their investment in our solution over a period of three months.
John adds: “SessionCam has become an integral part of our website improvement program, we use it every day. We can go straight to the Customer Struggle Score and see exactly where the blocks in the funnel are and why.”
Richard from SessionCam concludes: “Our clients are finding value in better understanding customer behavior to ultimately drive a bigger portion of users through the conversion funnel. We want to remain at the cutting-edge as innovators that make the most of the wealth of user behavior information to deliver a measurable return on investment for them.”