Blog
What is the 3-2-2 dynamic testing framework?
What is the 3-2-2 dynamic testing framework?
The 3-2-2 dynamic testing framework is an operator matrix: three concepts, two bodies per concept, two hooks per body. Twelve files, one variable changing at a time. It is not a Meta product and it has no published win rate. Use it to stop random tests; Produce the matrix in eonik; read it in Ads Manager with stop rules you wrote first.
What is the 3-2-2 dynamic testing framework?
The 3-2-2 dynamic testing framework is an operator matrix: three concepts, two bodies per concept, two hooks per body. Twelve files, one variable changing at a time. It is not a Meta product and it has no published win rate. Use it to stop random tests; Produce the matrix in eonik; read it in Ads Manager with stop rules you wrote first.
Why a small matrix beats a pile
Random creative dumps mix hook, offer, and actor. You cannot say what moved. 3-2-2 forces isolation: concepts first, then body, then opening.
Twelve variants is enough to learn if you can actually ship them. A 50-cell spreadsheet you never produce is theater.
How to run it
Sandbox at a budget you can afford to read. Pre-write pause rules. Promote only when your readout (hook rate, hold rate, then your CPA guardrail) beats control on the same window.
eonik assembles the twelve on-brand files. You approve. You do not outsource the decision.
- 1. Pick three concepts you can film or already have footage for.
- 2. Lock two bodies per concept, then two hook families per body.
- 3. Assemble all twelve in eonik with brand kit on. Approve each.
- 4. Sandbox. Read relative hook rate and hold rate. Graduate by human click.
When to shrink the matrix
If you cannot ship twelve this week, run 3-2-1 or 2-2-2. A smaller matrix you produce beats a larger one you fake.
Neither 3-2-2 nor 3-3-3 is Meta-certified. Pick the grid your studio can finish.
3-2-2 vs dumping ads
| Approach | What you learn | Risk |
|---|---|---|
| 3-2-2 matrix | Which concept / body / hook moved attention | Needs production discipline |
| Random pile | Almost nothing attributable | Spend without a readout |
| One hero only | No comparison | Fatigue with no backup |
What usually breaks
- Planning twelve cells and shipping three mixed files.
- Letting Advantage+ mix the whole matrix on day one.
- Using an invented hook-rate gate as a universal kill line.
When the readout is fatigue, hook rate, or velocity, the next move is consider what to make. Then make the cut in eonik and approve it. The loop is how to generate AI ads. It never touches spend.
“Almost any question can be answered, cheaply, quickly, and finally, by a test campaign. And that’s the way to answer them, not by arguments around a table.”
That principle still holds. The next on-brand cut is yours to approve.
Sources
Questions
What is the 3-2-2 dynamic testing framework?
It is an operator matrix: three concepts, two bodies per concept, two hooks per body, with one variable changing per comparison. It is not a Meta product. It has no published win rate. The point is isolation you can actually produce; Twelve on-brand files, not a random dump; Read hook rate and hold rate against control, then you decide what continues.
Is 3-2-2 better than 3-3-3?
3-2-2 is smaller and faster to ship. 3-3-3 is wider. Pick the matrix you can actually produce this week. Neither is a Meta-certified method; A 27-cell grid you never export teaches less than twelve files you sandbox with stop rules; Shrink without guilt if supply is tight; Isolation beats cell count.
Does eonik run 3-2-2 in the ad account?
No. eonik is the Mac app that assembles the twelve on-brand files from your footage and clips. You approve every export. You upload, you sandbox, and you graduate winners by human click; It never writes to Ads Manager and never spends; The framework lives in your log plus the cuts, not in an autopilot.
Free, no card · macOS 15+ · Apple silicon