Growth design case study
Ultimate Guitar
The most popular guitar tablature service in the USA.
I worked in a growth team whose purpose was to test hypotheses quickly and increase ARPU by improving user engagement and product usage.
Core objective
My main task at UG was to develop hypotheses for growing ARPU. Based on previous experiments and market trends, I concluded that personalisation was one of the strongest areas. My task was to determine which personalisation method to add to the product.
My plan consisted of the following steps
- Study and decompose the audience that needed to be influenced.
- Understand their motivations and challenges.
- Create a solution that personalised the experience of selected audience groups.
What do users dream about?
Analytics showed that 60% of the audience we needed to influence were newcomers to music. Qualitative research identified two core motivations: learning to play a favourite song and being able to play popular songs for friends.
Main problems for beginner musicians
- The learning process feels too long and progress is difficult to see.
- It is hard to practise regularly without a clear goal.
- A favourite song may be too difficult, so the user underestimates the effort and quits.
Winning idea
Six music teachers gave the same advice: practise regularly and devote a minimum number of hours each week to make progress. Based on this research, I developed around 40 ideas for personalising the experience of novice musicians.
The best result across 15 A/B tests delivered a 14% increase in ARPU.
The winning combination contained a schedule for future classes and personalised levels for each musical direction. Users selected how many hours per week they could practise, and the graph showed how many months it would take to reach the next level.
Winning experiment
Talk and context
Additional background on the work. Kept secondary to the case-study visuals.
Other experiments
Ideation process
The most effective experiments appeared when the same idea was supported by evidence from different sources. For each hypothesis I documented what part of the product changed, the before and after states, how users reached the screen, what we would keep, the proposed copy, and why the change should work.
View my ideation processPersonal challenge
I had to generate 10 hypotheses per week, with at least 3 strong enough to test. At the same time, I designed and implemented around 5 A/B tests each week. Because Ultimate Guitar had no UX research culture, I had to find different sources of knowledge to make those experiments successful.
Read about my personal challenge