A data driven approach to pricing

Since the paid subscriptions for Fast Rhymes were launched over 3 years ago, the prices have remained unchanged at 2.99USD/month and 24.99USD/year. Pricing is hard. It’s impossible to know what your product is worth to your customers. Even they don’t know.

A/B testing

A/B testing is a great method to test out the hypothesis's of a new pricing increasing the LTV per customer. LTV per customer is the lifetime value per customer, which in simple terms is how much revenue on average a customer brings in. This is usually what you want to optimize for. Since this is a freemium app, most customers aren’t paying, so it makes sense in this case.

Fast Rhymes uses RevenueCat for handling the subscriptions across the App Store and Google Play. They have a feature called experiments. This allows you to run price experimentation across the App Store, Google Play and even Stripe in a single place. This is the A/B testing suite used in this price experimentation.

The first experiment

The first hypothesis was that the pricing was way too low and that if a user pays 2.99USD/month, they would most likely also pay 4.99USD/per month. Same with the annual pricing. If you’re willing to pay 24.99USD/year, 39.99USD/year might be in your range as well. So these products were set up and configured in RevenueCat, and then an experiment was started.

After looking through the results, the change was overall more than double than before with an increase of 162,5%. Digging further into the results, the change on Android was 0%, so the results were all coming from iOS with an increase of 530%. From an LTV per customer of 0.23USD to 1.45USD. This is huge for later experimenting with Apple Search Ads, since the initial ROAS is likely to be significantly stronger.

The second experiment

The second hypothesis was that the pricing was a bit high for the annual plan. This came from seeing that a lot of the trials failed to convert because of billing errors, or cancelled before they were charged. So a new experiment was set up where the annual pricing was decreased from 39.99USD to 34.99USD. The monthly pricing was still kept at 4.99USD, which makes the annual plan seem like a better deal as well.

The results for the second experiment further increased the LTV, this time on both platforms with results on Android being more significant this time. The resulting LTVs for the treatment are a bit off since these experiments were run in different periods. It’s also important to consider that this is considered low-traffic testing which isn’t statistically significant. However, it does give a good indication of the result.

Putting a lot of hours in price experimentation and A/B testing with low traffic is probably not the best idea for low traffic, but performing some can give good indications. Then following up on the longer term effects is important. In this case a couple months after these experiments the revenue for the app has nearly doubled which is game changing for how much can be reinvested in further growing the app.

Conclusion

Price experimentation proved to be highly valuable for Fast Rhymes, demonstrating that initial pricing was suboptimal. It significantly increase in LTV per customer, particularly on iOS, which shows the importance of data-driven decision making when it comes to pricing strategy. While statistical significance might be harder to achieve with low traffic, even directional insights can lead to meaningful revenue improvements when acted upon thoughtfully.