Improving Umi's core loop
Umi is a language learning app with 100k+ downloads that helps users learn a language with tv shows and movie clips.
Before Resubscribe
In order to accelerate their growth, the team wanted to discover why they had a lower-than-desired conversion rate. Many users were downloading the app, giving two lessons a try, and then not paying when presented with a paywall. The founders didn't know what to build next to move the needle.
We helped Umi understand this problem with Resubscribe's conversion interviews.
Umi's attempt to gather feedback via surveys
You might be reading this and thinking to yourself "Why do I need Resubscribe? I can just send my users a survey". Well long story, short — the Umi team tried. And it didn't work.
They ran surveys via email and saw a few things:
- Low engagement rate (1-4%)
- Non-actionable information: the answers were often "the app is buggy" or "the price is too high".
- Short responses: People often don't fill out the "Other" field on surveys. The medium of the feedback was failing
- Adverse selection: their super users loved giving feedback, but they were missing a core group… the people who didn't like the app.
Surveys are great for hypothesis driven questions such as "Should I build feature A or feature B", but aren't so great when you're trying to gather user feedback that can help you iterate to product-market fit.
Here's a neat graph that shows where surveys and Resubscribe fit on the spectrum:
After Resubscribe
To gather the right feedback to know what to build next, The Umi team implemented Resubscribe to run discovery-driven conversations at a critical point — right after users decide not to pay.
Resubscribe also got better engagement rates than their previous email channel. With Umi, they saw an improvement from 1-4% to 15-25% engagement rate from users, resulting in 48 user conversations per day!
Results: discovering unknown unknowns
Talking to users is important because of unknown unknowns — questions that you don't even know that you should be asking. It's why opened ended follow up questions can reveal so much.
See an example conversation of unearthing an unknown unknown:
🤖: Would you mind sharing what you think Umi is missing?
🧑: When I got the app I was asked if I know hirigana and katakana. I know some (most of hirigana and a tiny bit of katakana) but there was only a yes or no answer. Unfortunately I fall somewhere in between.
Results: finding UX issues
Users also love to complain. When you stop to ask, they'll tell you what's wrong. Our AI knows how to ask the right questions to uncover exactly what the problem is.
See a real example of our AI digging into this UX issue:
🧑: I believe it was a good concept but the app had too many bugs. At first the app wouldn’t load properly only until I’ve redownloaded it multiple times it would work. There is also the matter of when I would complete my lessons it wouldn’t show that I’ve completed that lesson. I enjoy the concept but if only the app wasn’t so buggy I would’ve fully enjoyed it.
🤖: I'm really sorry to hear about the technical issues you faced. It must have been quite frustrating, especially when you couldn't see your lesson progress. Can you share a bit more about these issues? For example, were there specific lessons where this issue occurred or was it common across all?
🧑: It was commune across all
Clearly the user misspelled "common" there, but you get the point.
Results: uncovering trends
Umi didn't know they had a large cohort of middle and high school users. These users don't have an ability to pay, and so instead of asking these users to pay directly, Umi found a way to incentivize referrals in this cohort.
🤖: That's great to hear! Do you mind sharing what factors influenced your decision not to subscribe or purchase Umi at this time?
🧑: cause I dont have that much money because I'm still in high school
In summary
Umi started their journey without a firm understand of what to build next, and Resubscribe helped them set out again with confidence and conviction.
