How we enabled users to use our latest feature
The mobile team at Lyst had been iterating on the “discover” feature, focusing recently on personalised feeds (new products your brands, brand recommendations, etc.). This personalisation was powered by users telling us their preferences by following brands. The problem was that very few users were following brands, and almost nobody was following more than three.
The ultimate goal was to expose as many users as possible to the new personalised content. This was to be achieved by getting users to follow brands they like during onboarding.
We didn’t want to negatively impact the existing sign-up rate if possible. This was a fine balance as we would be adding some friction to the onboarding experience by asking users what brands they liked.
The existing onboarding had multiple paths dependent on the user’s region (EU, US, ROW) for GDPR related reasons. This was difficult to maintain, so we wanted to simplify this if possible.
Finally, we needed to communicate to new users that we would be personalising their experience based on their preferences.
A three phased approach was used (we tested at each stage before moving to the next):
1. Phase one was the MVP which would focus on removing existing parts of the onboarding and adding a simple brand following mechanic.
2. Phase two tightened up the landing and loading screens and improved the brand following experience by recommending brands to follow. iPad support was also added.
3. Finally, sign up was made mandatory and users were forced through our new onboarding flow. Signing up was skippable up until this point.
This section made up the entirety of the existing onboarding experience and was reworked to be simplified and as logical as possible:
- Most signups were from Facebook and Google so we placed those at the beginning of the flow
- Design of landing page was brought more in line with rebrand work that had taken place previously
- The T&C acceptance was changed to be implied (after discussion with the legal department) and added to the landing screen
- Email preferences were changed from a checkbox to buttons
- Men/women preferences were pushed further along the flow as they felt like “preferences” like brands on the next screen
The final version of this screen presents the user with two courses of action: search for a brand they like or add a brand from the provided suggestions.
- We prompted users by asking them what specific brands they were wearing, which most target users knew. This proved a good way to get the ball rolling, rather than asking users what their “favourite brands” were. This copy changes after a user adds their first brand
- Initial suggestions are recommended based on popular brands for the user’s gender and region
- Further recommendations are injected based on brands the user adds (e.g. if a user follows Gucci, we inject Prada and Louis Vuitton below the Gucci cell)
A series of short loading screens were added to help set context. These told users that we were personalising content for them, and also contained the names of the feeds that were being created (e.g. “Your daily drop”), and where to find them.
The MVP/phase 1 release tested surprisingly well with users following around 10 brands on average according to our analytics. Phase 2 tested even better, with users following around 14 brands on average.
Mandatory sign-up was A/B tested and found to have an impact of less than 5% on signup rate, so it was shipped as well.
In conclusion we felt we could proceed with a degree of confidence that new users were being exposed to an area of the application that delivered value to them and was useful.