We launched improved cohort comparisons in ‘Charts’
These additions to RevenueCat Charts make it even easier to analyze data over time
RevenueCat acts as your single source of truth for in-app purchase data. Our Charts feature automates reporting for over 15 success-driving metrics like monthly recurring revenue (and its movement), subscriber retention, and conversions. Comparing different cohorts across these metrics in Charts is a valuable exercise in evaluating performance.
To make sure developers can measure performance accurately over time, we’ve launched two new additions to Charts: Incomplete Period Styling and Conversion Timeframe selection.
Visualizing incomplete periods in Charts
Charts now supports distinct styling for incomplete periods to make analyzing your data as easy as possible. Periods where cohorts have not fully matured will be styled with dashed lines and semi-transparent areas based on the chart type, and the corresponding cells in the data table will have a hashed overlay.
For charts which are cohorted by event date, the incomplete period is a function of the current day, where the current day/week/month/year is the incomplete period.
For charts which are cohorted by a customer’s first seen date, such as our Realized LTV charts, the incomplete period is a function of the current day and the specified Customer Lifetime, so that you can quickly see whether a period has had enough time to reach full maturity to be comparable with prior periods.
Accurately compare cohorts with Conversion Timeframes
Similarly, our Initial Conversion and Conversion to Paying charts now allow for Conversion Timeframe selection so that you can limit the timeframe that a conversion might occur within for it to be included in this chart.
Like the Customer Lifetime setting in our Realized LTV charts, this allows you to compare prior periods accurately, and quickly see if your most recent periods are still incomplete.
Equalizing periods for comparison is particularly relevant for conversion charts, as it lets you limit how much time a specific cohort has to mature and be accurately compared against a more recent cohort.
For example, an older one-year-old cohort that had an entire year to complete the desired action shouldn’t be compared with a one-month-old cohort which has had much less time to mature. Such a comparison would lead to the one-year-old cohort appearing to perform better, which may not necessarily be true as this cohort has just had 12x the time to convert.
With Conversion Timeframes and incomplete period styling, our goal is to make it easy to do these accurate performance comparisons so that you can truly understand what’s driving your business — and do more of it.
You might also like
- Blog post
Implementing in-app purchases and monetizing your Roku app: A step-by-step guide
A comprehensive guide to implementing and monetizing your Roku channel, from choosing the right strategy to technical tips and integration best practices.
- Blog post
How we built the RevenueCat SDK for Kotlin Multiplatform
Explore the architecture and key decisions behind building the RevenueCat Kotlin Multiplatform SDK, designed to streamline in-app purchases across platforms.
- Blog post
Inside RevenueCat’s engineering strategy: Scaling beyond 32,000+ apps
The strategies and principles that guide our global team to build reliable, developer-loved software