Songkick iOS Onboarding
Improving acquisition and activation in the first time use of Songkick's iPhone app, in order to improve retention and increase MAUs.
The problem
The Songkick iPhone app was failing new users. There was significant drop-off at key moments of acquisition and activation - resulting in a poor user experience, and a negative impact on Songkick’s retention rate and MAUs.
The context
Onboarding in the app consisted of up to 6 screens and 30 interactions, aimed at acquiring and activating new users.
In order to be considered 'activated' there are a number of conditions Songkick requires a user to meet. These conditions increase the likelihood of a rich, personalised experience on first time use, and return visits in future.
(Note! In a separate project, I've also worked on shortening onboarding and reducing the threshold for activation)
Understanding the problem
Using Mixpanel, I created a funnel for onboarding which included these steps of activation - to understand where the biggest drop-offs were occurring, and therefore where we could have the most positive impact. There were a couple of clear takeaways:
- 12% of new users weren’t making it past the first screen of the app - the library scan. Given this represented the top of the funnel, it was a key moment to investigate and address.
- 45% of new users were dropping off at the sign-up screen. With MAUs in mind, this stood out as a primary problem to solve.
- 40% of new users weren’t allowing push notifications when prompted. Unsurprisingly, allowing push notifications had shown to be a key characteristic of users retained beyond their first month.
Staying scrappy
Data from Mixpanel had exposed the moments of biggest drop-off, and therefore we could narrow our focus and hone in on the areas most impactful to MAUS.
Onboarding can be a bit of a honey trap for product teams; there’s often a strong appetite to emulate other apps, or to rewire the entire experience into something fluid and beautiful. I made sure that as a small team we were still testing in the leanest, scrappiest ways in order to validate hypotheses and find the impact we were looking for.
Learning fast
To help better understand the problems and opportunities at the key points in this funnel, I set about gathering fast, qualitative feedback from users.
This took the form of remote usability sessions using Ping Pong, and surveys targeted at specific cohorts of users (namely, those who had opted out of push notifications) using a combination of BigQuery, Sendgrid and Typeform.
From this feedback we were able to craft a number of hypotheses that we could easily test at scale with new users in the app.
Hypotheses
- If we add a welcome screen at the top of the funnel, then a higher proportion of new users will allow access to their library and their location because they will understand what's happening, and the value will be more contextualised.
- If we move the push notification permission screen after signup, then a higher proportion of new users will allow push notifications because they will have seen the artists they've tracked and recognised the value of receiving concert alerts for those artists.
- If we redesign onboarding screens to have a clearer content hierarchy and use more engaging, less committal language, then a higher proportion of users will sign up and activate because the experience will be easier to understand and complete.
Experimenting
As a product team, we decided the fastest way to learn would be to productise these changes outright.
- At the time, the app didn't have an A/B testing framework that allowed for switching the order of screens.
- Onboarding conversion was very stable and didn't show any noise or spikes that could skew our understanding of the changes.
- Given the experience of the tech team in iOS at the time, and the unknown release and approval cycle with Apple, we agree we'd see faster results if the changes were shown to 100% of users - and benchmark against the average of the past 3 months.