Developing User Personas for ClassPass
Bonus: Design Recommendation for ClassPass iOS App
ClassPass is a subscription to boutique studio fitness with multiple products. As part of a team of 4, I was tasked with creating user personas for ClassPass. We needed to profile their users to discover patterns among their motivations, goals, pains, and behaviors so that ClassPass can better attract, retain, and serve their users.
We used qualitiative data in the form of hour long interviews, and self-reported pre & post-interview surveys to inform our personas.
Quantitative data was used in persona generation and in user segmentation.
We would use quantitative information in the form of user’s behavioral data to both generate our personas, and to create our initial user segmentation for recruitment. User clusters were created by a data scientist based on four factors, some inside a user’s control, and some outside;
Usage - Whether or not the user spends all of the credits on their plan
Booking Behavior - How far in advance the user books a class, and cancellation frequency
Studio Density - How many ClassPass partner studios are in the user's area
Social - How often the user invites friends or performs other social activities in ClassPass
This led to the creation of four user clusters. We recruited six users from each cluster, making an effort to proportionally represent the metro areas in which ClassPass is most active. Our user pool skewed female, though not as female as the actual ClassPass user base. A request was made to slightly overrepresent males in the research to attempt to identify insights that might attract more males to the platform.
ClassPass has three subscription tiers, which we also wanted to evenly represent in our users, though they would not be represented evenly inside each cluster. It stands to reason that a low-usage and low studio density cluster would have more low tier users.
There is also a “holding pattern” tier, which is a middle ground between the lowest tier and cancelling. Regardless of cluster, we recruited five users on this plan to gather their perspectives.
Because ClassPass had recently changed their billing and subscription structure, we wanted to recruit users who had only known the new system so that frustrations with the transition did not interfere with their responses. This meant all users interviewed had been using ClassPass for seven or fewer months.
Once users had been identified, selected, and confirmed, we obtained their detailed user data so that we could review their booking behavior. We were able to see which studios they frequented, which they visited once and never again, and how often and early they cancelled. These would prove to be helpful talking points in our interviews.
We first sent our users a 15 minute survey to gather an overview of their past and present fitness motivations, attitudes, and behaviors about ClassPass and exercise as a whole. These answers were expanded upon in our interviews.
We worked off of a script previously used for a diary study. Tailoring it to our needs, we identified sections of questioning that we needed to add. After a few dry runs and iterations, we had a well organized script that touched on every area we wanted to cover. Some questions seemed redundant, but elicited different answers when asked in different conversational contexts. After our first day of interviews, we identified low-priority areas of the script to cut in the event of an interview running over time.
After the interview, we sent another five minute post-interview survey. This survey was slightly redundant, but in asking the same questions repeatedly in a survey-interview-survey sandwich, we found that each answer got closer to a user's actual behavior, rather than their idealized behavior.
Over the course of seven days, we completed 29 user interviews. Each was led by one researcher while another took detailed notes, all while being recorded for later reference.
At the end of interviewing, we created post-its for affinity mapping our interview and surveys. A different color was used for ever user to allow for easy identification of patterns (the more colorful a group of similar data points, the more support it had). We booked the largest meeting room at ClassPass for a full week and set out to decode our users.
We began to map out our users one cluster at a time, but we found that trends were few and far between. Walking around the room, we noticed that there was a lot of opportunity for crossing clusters. We realized that our users and the clusters they had been assigned were not lining up. Looking back to our interviews, we identified the cause of this misalignment. Most month-one users are on a trial plan, and ClassPass frequently tests trials with varying time limits and class credits. Creating clusters based on month one data also meant that these users had not had time to develop a ClassPass routine.
This led us to create new user segments based on the biggest trends we saw in our interviews: Primary and Supplemental users.
Primary Users use ClassPass as their primary means of exercise.
Supplemental Users use ClassPass as a supplement to an external fitness routine.
Once these two main segments were identified, we started testing theories about their fitness history, fitness knowledge, scheduling flexibility, workout preferences, social habits, and anything else we could discern from our data. Once a theory was posited, a researcher would scour our walls of post-its for supporting evidence. Many theories were abandoned, some were found to be supported across all clusters and were therefore useless as a defining characteristic. Others were found to be valid, useful insights that helped to further divide our segments.
Further segmenting our initial split, we identified a total of four unique user personas, two each in Primary and Supplemental.
Primary 1: High fitness motivation, high usage. Enjoys personal and social aspect of studio fitness.
Primary 2: Med/low fitness motivation, low usage. Low fitness knowledge, needs the personal attention of an instructor.
Supplemental 1: High fitness motivation, low usage. ClassPass is a small part of a larger fitness budget, viewed as a fun extra to a workout routine consisting of multiple memberships.
Supplemental 2: Med/high fitness motivation, low usage. ClassPass is a large part of a small fitness budget, seen as an expensive extra to a workout routine consisting of running, biking, and other free workouts.
Our theory testing allowed us to develop a nuanced picture of each persona. We were able to build a complete story for each one, detailing each unique characteristic, the motivations and attitudes behind it, and how it all fits together.
Finished personas are not available for public view due to NDA.
Bonus: Design Recommendation for ClassPass iOS App
We identified which users were most likely to churn and why. This allows for ClassPass to create design solutions to address these pain points. The user most likely to churn has low fitness knowledge. They were not on a sports team in school and have not had a personal trainer. They benefit from having a trainer give them motivation to push themselves, and personal direction on correct form.
However, this user still feels like they do not know which areas of their body they should be focusing on, which workouts to do, or how often to go. Based on the user's subscription tier, and simplified goals of weight loss, toning, maintenance, or strength, we could recommend a weekly number of classes falling into the workout categories already present in ClassPass, and help them keep track of which recommendations they are filling and which they are falling behind on.
With only marginal personalization, users will get a recommended routine that will help them feel confident that they are on track toward their fitness goals, and that ClassPass is supporting them along the way.
Below is a mockup of how this might look integrated into the ClassPass iOS app. To view this mockup as a prototype, click here.