I'm looking to improve my testing and scaling process for paid ads on Meta and TikTok for a subscription app. Right now I feel like I'm either leaving money on the table or discarding too early some interesting ads.
What's your workflow for:
What metrics do you look at to decide "this one's ready to scale"? (Apart from the Purchase event)
How do you structure your testing?
I usually go with 1 testing campaign, 1 adset, and only 1 ad per adset so each one gets equal opportunities. But lately with higher testing volumes, I've started adding multiple ads with the same concept to a single adset to identify which one performs best.
Would love to hear what's working for you guys. Thanks!
I currently promote apps with fast-converting in-app purchases rather than subscription models. So I´m not sure if its good for subscriptions as well.
My setup is designed for fast validation and quick scaling decisions.
Testing approach:
I allocate a fixed test budget of €300-500 per creative cluster (10 creatives). Each test ad group usually contains around 10 creatives running in parallel, allowing clear performance comparisons within a short time frame.
Core evaluation metrics:
CTR
CPC
CPI
CPA
Retention
Since the key KPI purchase typically happens within the first 24–48 hours, I can make data-driven decisions very quickly.
Beyond early conversion signals, I also analyze D1–D7 retention to evaluate user quality and long-term performance potential.
It has always worked very well for me so far.
more or less same as @bungabunga
My top KPI to look at is the cost per event but it really depends on each case. In gaming, you also have to check retention rates for example while in sub apps, you have to check conversion from trial to paid if you are optimizing towards trial start event.
I do also monitor placement and age distribution in order to move a testing asset into a BAU ad group: If most of the impressions are served in my most valuable segments and placements, that's an extra reason to try to scale
.png)
