We wanted a clean number for the impact of selling subscriptions on the web instead of via in-app purchases. Rather than guess, we set up two cohorts: web checkout signups and in-app purchases. For each cohort we tracked gross revenue, platform fees, gateway fees, CAC, and retention.
The math was straightforward: net_revenue = gross - platform_fees - gateway_fees. Then we computed cohort LTV at 30, 60, and 90 days. Beyond immediate net uplift from lower fees, we saw indirect gains: web-acquired users gave us email for win-back campaigns and we could experiment on pricing and upsells faster, which improved 90-day LTV.
How do you structure your comparison cohorts so the data isn’t biased by traffic differences?
We seeded equal quality traffic to both flows and tracked cohort LTV at 30 and 90 days. Subtract platform fees and compare net margin per cohort.
Having the web funnel made it trivial to change price and retest. Web2Wave helped us iterate fast enough to feel confident in the numbers.
Run parallel tests with the same creative and landing pages where possible. Keep traffic consistent and measure net LTV not just conversion. Web funnels let you change offers instantly and see the effect on revenue and retention which makes the comparison meaningful.
We compared cohorts from the same ad sets split to web vs app flows. That reduced bias and showed the real net uplift after fees. It was more work up front but worth it.
Run split traffic then compare net LTV
To avoid bias, randomize traffic at the ad level so users with similar intent hit either the web checkout or the in-app flow. Track gross and net revenue, CAC, and retention per cohort. Don’t forget to control for geography and device.
Report LTV both including and excluding platform fees. The delta shows the theoretical uplift but look at retention too because cheaper acquisition via web can alter cohort quality.
We matched cohorts by source and device. If you cannot randomize at ad level then use propensity matching on baseline signals like session depth and landing page.
We saw clearer margins after subtracting store cuts. Make sure you compare cohorts from the same campaigns.