50% to 1% churn rate - 8 things that worked for us

We built an AI application called LevelFields.ai to mirror the event-driven investing styles of the top hedge funds. It’s like a weather app for market events, telling you where the major events are happening and forecasting how they will play out so you can easily trade and make money.

It took a team of PhDs, data scientists, and engineers two years to build so we weren’t able to spend a lot on the front end of the system.

When we soft launched, we realized we had a major problem with the UX after seeing 50% churn in our monthly subscriptions. But doing these 8 things enabled us to almost eliminate churn:

1. We Made Everyone Pay.
   We tried offering a free subscription option thinking we would convert as a freemium model. It didn’t work. Few converted because we were not giving access to the entire app. Even worse, we didn’t get any reason as to why people didn’t sign up. It was only when we made people pay did we start to hear the kind of complaints necessary to enact productive changes.

2. We Targeted The Right Users .
   When we launched LevelFields, we mostly used an existing list of friends and family, thinking they would be the easiest adopters. But few signed up for anything that required paying, and those that did provided zero feedback. They just paid and never used the application. As a result, we learned nothing. We made a tactical change to begin targeting people with a small ad campaign who showed interest in investing and saw an immediate change in user activity on the app. Finally we had some data to review.

3. We Changed Our Pricing Structure.
   Initially when we launched LevelFields, we offered a subscription for 59/mo and a free option with limited access. A lot of users came in to explore, poach ideas, or just thinking the AI was going to do all of the trading for them. Most people who canceled the subscription did so within 24hrs of signing up. This told us that we had an onboarding problem. But we did not know why. We opted to eliminate the monthly option entirely for a bit of time to see if that made users more willing to spend the time reading the documentation on the app we had put on there before trying to cancel. It worked to some degree, but we still got a lot of cancellations very quickly and needed to understand why.

4. We Removed The Cancellation Button .
   Because we needed to know the reasons users were canceling, we made them email customer support requesting to cancel and had our agent politely inform them they were required to gather a reason for cancellation prior to canceling their subscriptions. It pissed some people off, but we got a lot more information about user expectations this way and we began to see the patterns. Half of the users quit because they couldn’t figure out how to use the app, 25% because it wasn’t cherry picking the best recommendations, and 25% because they couldn’t afford the app and had buyers remorse. But we needed to go deeper as to what users were expecting.

5. We Did In-depth Interviews.
   We asked LevelFields users to help us improve the experience and got a few to volunteer their time. We then asked them to walk through the app and explain what they were seeing and how to use it. They fumbled and skipped over all the documentation we had written out describing in detail how to use it.

We saw they kept trying to search for things when there was intentionally no search. We saw they were looking for stocks they owned instead of new opportunities we were providing. And we gathered that they didn’t know what to do with the information they were seeing. But the sample size of users was still too low to determine the needed changes to the app, so we had to offer something else to get more users to do interviews.

6. We Offered Free Access In Exchange for Feedback.
   We blasted an offer to waive subscription fees to anyone willing to spend 30 minutes on a Zoom call describing the issues they were having and what they wanted to see. We got a few dozen to take the offer and gathered the feedback in a large spreadsheet over the course of several months. This offer worked because we did not start with a free service.

Some of the issues users noted were browser-related technical issues but most had to do with expectations of key features we didn’t create that users had come to expect. The biggest takeaway was users already had their own way to invest in stocks and were looking for ways to expedite THEIR process. We were providing an entirely new process, and it was too stark of a difference for them to overcome.

7. We Built The Top 5 Most Requested Features.
   We analyzed all of the requests in a spreadsheet and looked at the top 5 most requested items. About 75% of the users we spoke to requested the same features. Four out of six features were very easy to build. We got to work on those right away and had them in the app within a month.

The other features were harder, and would require a complete rework of the application, so we got to work on the designs and are about to implement them.

8. We Doubled Down on Customer Service.
   We created template responses to common questions so we could respond quickly when customers wrote into the application. We included screenshots of the application that were marked up to visually show users how to access features since earlier insights revealed they didn’t want to read lengthy bits of help text. The responses were usually very good, as some user expected they’d never hear from customer support and we tried to answer them within the same hour they emailed.

After a few months, we dropped churn to almost nothing and we’re seeing rapid month over month growth. We’re also getting a lot of emails about how much money they are making using the application, and how much the ideas and trading tips provided in the weekly newsletter have helped them avoid losing thousands of dollars as the market crashes.

In the coming days, we’ll be releasing a monster upgrade to the application that enables a lot more personalization. We plan to raise the prices from $228 to $2000 per year for new LevelFields users to offset the rising costs of, well, everything, and that will be the next test to see if users are really price sensitive.

If you’re interested in supporting the project or collaborating, please visit www.levelfields.ai.

Thank you,
Andrew

Related Posts:
0-400K in 12 months. Then We Launched. https://www.indiehackers.com/post/0-400k-in-12-months-then-we-launched-heres-what-happened-next-dc5d40cb89