The Four-Layer Data Framework

This diagram was literally built the night before my workshop at RevenueCat App Growth Annual but I already explained it in one of the articles I wrote for this amazing company (DM me if you can’t find it).

I’ve always believed that education through visualization makes things easier to understand — and diagrams like this have helped me clarify my own ideas. Hopefully, this one helps you to understand my framework much faster.

The framework goes from top to bottom, depending on how aggressive you want to be with UA:

👉Layer 1: SKAN
Your main source of truth — deterministic and directly reported by Apple. We all know SKAN underreports, but I’ve seen cases where SKAN gave me a CAC below my limit. That’s the best-case scenario: if you’re here, you can scale your campaigns confidently with minimal risk.

👉Layer 2: Adjusted Data
When SKAN doesn’t give me the green light, I use an intermediate scenario between SKAN and probabilistic data. I apply an adjustment factor (formula’s in the article) to the numbers from my probabilistic sources (AEM/ADC/whatever) since those are normally higher than SKAN.

The adjustment factor gives me a scenario which is still conservative, but a bit more aggressive than SKAN. If you are in such a place with good margins, you will likely be safe to scale too but you have to monitor blended to ensure scalability brings incremental growth.

👉Layer 3: Probabilistic Data
If you need more room to spend, this is your next step. You can use probabilistic data directly as reference and if the CAC is still within target, scale SLOWLY while checking blended results to confirm that extra spend = extra conversions.

There’s still great growth potential here, but you need to move slower than on layer 1 or 2 — your campaigns may be slightly overstating performance.

👉Layer 4: Blended
The most aggressive scenario as you just count everything as part of UA. I like using this when the app relies heavily on paid traffic.

If numbers don’t look good here — STOP your campaigns and analyze your fundamentals:

-Are your conversion rates within benchmarks?
-Any major drop-offs during onboarding?
-Are campaigns properly set up? (maybe you need help here 😉)
-Do you just have a bad product?

You’ll find at least one “no” in those answers, and that’s your direction to fix things.

On the other hand, if you see viability with blended, you can still scale but in this case you must do it even SLOWER as in the layer 3. You’re taking higher risks of over-attribution and possible cannibalization of your organic traffic, so it’s essential to keep monitoring blended and ensure your growth is truly incremental.

Hope this helps all the new devs out there just starting with paid UA!
See you in the next one 🫶