Tracking Lifetime Metrics in Paid UA: Cohort Analysis


One of the biggest issues in paid UA is that it can be difficult to accurately measure the true impact of a particular campaign or channel over time. In performance marketing, we typically evaluate campaigns and channels based on their immediate results.
However, this approach can be shortsighted and lead to hasty decision-making that fails to take into account the bigger picture.
Due to current tracking limitations on iOS and extended user conversion flows, it may take some time for users to take the desired action. This highlights the importance of avoiding hasty decisions based solely on immediate results in performance marketing. One effective practice is to review previous reports and track the performance of campaigns over time.
For instance, a campaign that showed a higher Cost Per Action (CPA) than the benchmark in a particular month could show improved results in the following days, weeks, and months.
While this approach may not (yet) apply to SKAdNetwork, it remains relevant for tracking performance on Android and Apple Search Ads and for consenting users from iOS campaigns on META, TikTok, etc.
AppsFlyer, Adjust, and other MMPs offer the opportunity to measure results over time. For example, Adjust refers to this as ‘Cohorted Events,’ while AppsFlyer uses the term ‘LTV Based Metrics’ (don’t confuse with Life Time Value).
As AppsFlyer puts it, “LTV data are the events performed during the lifetime of a user who converted/installed the app in a specific date range. For example, use LTV data to see the ROI up to the present day of users who installed your app during a specific month. “ |
Our Data Examples
One of our clients operates an audiobook app that primarily relies on subscription-based revenue. To entice potential users, they offer the first chapter of each audiobook for free. Based on our cohort analysis, we found that most users convert to a paid subscription within the first three days of signing up.
However, we’ve also observed that some users opt to convert at a later time. As such, it seems unfair to discount the value of these later conversions in our CPA report. After all, our campaigns may still have played a critical role in persuading these users to convert, even if it wasn’t within the initial three-day window.
As Apple Search Ads are exempt from tracking limitations, they can provide us with lifetime measurements of installers from a specific date range via MMPs. We usually send monthly reports to our clients by the third day of the following month.
After analyzing the reports we submitted for July, August, and September 2022, we reviewed the overall performance of those months in January 2023 again and found that the overall CPA had decreased significantly. This led to a 67% improvement in the CPA. We anticipate that this improvement will continue to grow over time.
For Facebook Android campaigns, you can easily track lifetime development. However, we didn’t run any Android ads for this client during that particular timeframe. Regarding Facebook iOS campaigns, we performed the following calculations to estimate the cohort development:
- Since we can access the lifetime development data of iOS users who have given consent only, we calculated the conversion rate (installs to subscriptions) of those users and extrapolated it to SKAN installs.
- When we reanalyzed and estimated the CPA in January 2023, the results showed a 38.45% improvement for July, August, and September 2022 installers.
Side Note
Currently, if the privacy threshold is passed, SKAN is able to report the in-app events/revenue for up to 72h post install. With SKAN 4.0, it will be possible to measure in-app events/revenue for up to 35 days. While this still presents a significant limitation in lifetime measurement, it will for sure strengthen our ability and reliability to make longer-term estimations. |
Cohort Analysis Takeaways
Make it a habit to track the development of your cohorts overtime regularly. Remember that some users may take longer than others to convert, so it’s important to be patient and gather valuable insights from this type of analysis.
While Apple Search Ads and Android Campaigns may not pose any limitations, don’t forget to analyze the data of iOS consenting users from channels that are subject to SKAdNetwork (Facebook, TikTok, etc.) during the reporting period. This will help marketers obtain accurate numbers when estimating how the figures evolved over time.
Eventually, your objective is the one to determine how much you can afford to pay for a converting user. Therefore, you should combine the CPA development analysis with the LTV calculation to realistically estimate how effective your marketing efforts are and that you don’t miss out on scaling possibilities due to a too much conservative strategy.
Need help measuring the true impact of a campaign? Our team of experts will help you measure, manage and analyze marketing performance data to understand the effectiveness and improve ROI.