Peloton and the paradox of choice
A product idea inspired by Netflix’s “Play Something” feature
If you’ve ever used Peloton, you’ll know that it can be extremely overwhelming to pick what class to take.
Peloton’s app offers thousands of classes, with a few different criteria: type, length, weights required, instructor, and music.
One of Peloton’s value propositions is time; you don’t need to physically go anywhere, and there are workouts of varying durations, so they fit perfectly into your busy schedule. It’s all about flexibility.
With the amount of content they produce, Peloton is effectively now a media company. Their CEO, John Foley, even describes them as a “Netflix library, but for fitness classes.” But as Peloton scales and adds more and more classes weekly, the problem emerging for users is the paradox of choice.
Having many options to choose from, rather than making people happy and ensuring they get what they want, can cause them stress and problematize decision-making.
As someone who recently started taking Peloton’s classes, I found myself spending more time deciding what class to take than actually doing the class itself!
When you load up the classes list, here’s what you’re faced with:
5,966 classes! And that’s only the Cycling ones.
Then you try the filters…
Imagine being a first-time Peloton user (like I was) and landing on this screen. You have no idea where to begin.
Do you just endlessly scroll through the list? Do you filter the results? How do you know what instructor or class type you’ll like?
This friction harms new user activation, as well as engagement and retention for existing users. New users may cancel their subscription or return their Peloton bike, while existing users may opt to do other workouts instead.
If this paradox of choice sounds familiar, it’s because another consumer company has started facing the same dilemma: Netflix.
Netflix has upwards of 6,000 titles available in a given country. And while they have an excellent personalization algorithm that curates content they think you’ll like, it’s still mentally exhausting to scroll through all those tiles to find exactly what you’re looking for.
It’s why Netflix recently launched “Play Something”; a new feature that just picks something for you to watch, with one click.
As Elena Naira, a professor at the University of Catalonia, says:
The world of never-ending content rows brings too many alternatives for our brains to handle, and that is when decision fatigue kicks in. We start to feel overwhelmed. And if you can’t make up your mind, frustration can make you leave and go to another platform.
That last sentence is the crux of the problem for Netflix. Ironically, by offering too much choice, users may abandon Netflix in favour of something simpler — like cable TV.
And so, “Play Something” tries to mimic precisely that: the feeling of channel surfing on a regular TV.
The experience was enhanced, too; it always starts at the beginning of a program, rather than in the middle (unlike watching TV), and all the content is curated based on your preferences.
That brings me back to Peloton.
Imagine if, upon opening the Peloton app, it simply asked you how long you want to workout. And from there, it launched you right into a class (of that duration) that it thinks you’ll like.
If you didn’t like the selection, you’d have the option to skip to a different class, or do a different type of workout.
Now, how could Peloton decide what class is the best fit for you? Well, for existing users, Peloton has a ton of historical data about their usage.
It knows their age, gender, what music they like, who their favourite instructors are, their ratings of classes they’ve previously taken, which classes they love to take repeatedly, and which classes they haven’t taken yet.
For new users, Peloton could take a page out of Spotify and Netflix’s playbook by simply asking users a few short questions in their onboarding flow.
All they’d need to know is what kind of music you like, what equipment you have, and how long you want your workouts to be.
Peloton’s curation algorithm can then use this sample dataset to generate a personalized stream of classes they think you’ll like. Users could give feedback along the way if a class was the right pick or not – which helps improve their model.
Apple Fitness+ has started doing this already, striking the right balance of letting users do more of what they like, while also introducing them to new content. Apple is incentivized to do this so that you learn to enjoy more and more of their content over time, which increases engagement and lowers the risk of churn.
Imagine being a new Peloton user. You open up the app for the first time, tell Peloton that you like 2000’s Pop music, have a set of light dumbbells, and want to work out for 30 minutes. You’re immediately launched into a 30-minute bike bootcamp with Cody Rigsby. And after you complete it, you’re shown suggestions for similar classes to schedule for your next workout.
Or imagine being an existing user. You launch the app, and are shown a curated stream of classes — much like Netflix or Spotify’s “For You” sections — and spend way less time deciding what class to take.
Over time, this allows Peloton to leverage a flywheel: the more effective their personalization is, the greater their user experience, which leads to more user growth. Again, just like Netflix:
You can apply this model to other forms of content consumption where time is a constraint — like podcast listening.
Measuring success
Onboarding
I’d run an A/B test on new Peloton users, where half of them see the new-user onboarding flow, while the others are popped into the full class list. I’d want to measure the difference in free-to-paid trial conversion rates, as well as 1-month retention post-signup. I’d also want to measure new-user engagement by the median number of classes a user took in their first month.
Personalization
I’d run an A/B test on existing Peloton users, where half of them see a personalized class list, while the others see the default class list, sorted by date. I’d want to measure the difference in 1-month retention, as well as in engagement; the median number of classes a user took in the next 30 days.
Relevancy
Finally, I’d want to measure the effectiveness of the recommendations. Each time a user is launched into a “Workout Now” (my feature name for this!) session, they’re able to skip to the next class if they didn’t like the selection. I’d want to measure the % of classes that get skipped past. After the class, users would be asked if they liked the recommendation, and could provide a simple 👍 or 👎. I’d want to ensure >80% of recommendations receive a thumbs up!
In summary, here’s how I think Peloton could solve the “paradox of choice” problem:
- Implement a new-user onboarding flow to learn users’ preferences
- Personalize your class list based on these preferences
- Add a “Workout Now” feature that launches you into a class you’ll like
This is a series of blog posts on the tech industry and product management. I’m on Twitter at @samir_javer if you want to say hello! 👋
