A Discrete Choice Take on App Building
Updated: Jul 11, 2020
Last week, a good friend shared a stat she heard at a conference. It was incredibly surprising, and we both knew that discrete choice could help.
Across all industries, 71% of all app users churn within 90 days.
This is according to research from @Localytics in their article – Mobile Apps: What’s A Good Retention Rate by Justina Perro
What does this mean? Well, by definition, churn means leave. Adios. No longer an active user. Etc.
What does mean for the apps of the world? That all the time, energy, man power, and not to mention, money, that goes into creating an app and getting people to download it, has a high likelihood of being obsolete if you don’t have features that keep your users “using”.
One solution for increasing retention and decreasing churn would be to ask your users what features would drive usage and to cultivate a culture of continuous innovation and product development. One app that comes to mind as a great example of this is SnapChat.
While I’m the first to admit I am not an expert on their platform, I do know how to do research! When filing for their IPO, SnapChat shared their product development timeline. You can see continuous innovation over the years with no sign of slowing down.
But how does SnapChat decide which features to implement? And how do they know those features will resonate with users? I’d like to think that they are using a combination of user feedback and in-house creativity to come up with a list of feature ideas. And then they might even use a MaxDiff to prioritize their roadmap. For example, a MaxDiff survey might look as follows:
In this example, SnapChat is asking users about four different features at a time, from a list of 10-30 to maybe even 100 different features. Users choose which feature they like the Most and the Least from each subset. They see multiple screens with different subsets of items and by forcing users to make choices, we can uncover what truly interests them.
At the end of data collection, we can run a model and generate a score for each feature tested. These scores can be transformed into ratio-scaled scores, meaning that a feature with a score of a 10 is twice as preferred as a feature with a score of a 5.
Using these MaxDiff results, SnapChat could sit down with its engineers and developers and trade-off user wants with developer time to implement and truly optimize their app roadmap.
Want to chat more about MaxDiff and its potential benefits to you and your business? Email us at firstname.lastname@example.org.
Note – Ideas for SnapChat features in the example were drawn from a Tech Crunch article here