The short answer
A creative matrix maps your ads on three axes (narrative angle, reader awareness, format) so you can see which combinations you have never tested. On our board of 2,443 winning Meta ads, the most crowded cell (Root Cause x Unaware) holds 318 ads with an 8% top-tier rate, while Systematic Breakdown x Solution-Aware reaches the top tier 22% of the time with 99 ads.
Why a matrix
Teams that test a lot of ads often test the same ad a lot. Twenty new creatives a week can still be twenty problem-aware UGC videos with a root-cause story. Meta's retrieval system (Andromeda) is widely reported to group look-alike ads, so that output behaves more like one test than twenty. Practitioners call this a creative similarity problem. The term is industry shorthand, not an official Meta metric.
A creative diversity matrix makes the problem visible. You cross the dimensions that change an ad's argument, drop every ad into a cell, and look at where the gaps are.
The two axes that matter most:
- Narrative angle: the argument. Root cause, systematic breakdown, story, expert, wrong question, cautionary tale, insider, accidental discovery.
- Reader awareness (Eugene Schwartz's five levels): how much the viewer already knows. Unaware, problem-aware, solution-aware, product-aware, most-aware.
Format (video or static, and the production style) is the third layer, applied inside each cell.
The matrix from 2,443 winning ads
Here is the matrix for the whole Ad Radar board on October 9, 2026: 2,443 live Meta ads, each above $50k in estimated spend, placed by primary angle and awareness level. Each cell shows ads / share that reach the top tier (Ad Radar's top 10% by winner score; about 10% across the board).
| Angle \ Awareness | Unaware | Problem-Aware | Solution-Aware | Product-Aware | Most-Aware |
|---|---|---|---|---|---|
| Root Cause | 318 / 8% | 186 / 9% | 175 / 9% | 49 / 8% | 112 / 6% |
| Systematic Breakdown | 132 / 8% | 55 / 20% | 99 / 22% | 22 / 0% | 35 / 11% |
| Story/Case Study | 100 / 12% | 60 / 5% | 32 / 12% | 9 / 22% | 9 / 44% |
| Expert-Based | 70 / 4% | 50 / 18% | 42 / 12% | 11 / 9% | 6 / 0% |
| Wrong Question | 68 / 15% | 48 / 4% | 59 / 10% | 12 / 8% | 11 / 9% |
| Cautionary Tale | 13 / 15% | 13 / 15% | 14 / 7% | 0 | 2 / 0% |
| Insider/Whistleblower | 11 / 18% | 6 / 0% | 5 / 0% | 1 / 0% | 2 / 0% |
| Accidental Discovery | 1 / 0% | 0 | 0 | 0 | 0 |
Cells with fewer than about 20 ads are small samples. Read their rates as hints, not findings. The product-aware and most-aware columns are mostly small.
318
ads in the most crowded cell (Root Cause x Unaware)
22%
top-tier rate, Systematic Breakdown x Solution-Aware (99 ads)
20%
top-tier rate, Systematic Breakdown x Problem-Aware (55 ads)
18%
top-tier rate, Expert-Based x Problem-Aware (50 ads)
What the matrix says
The crowd is in one corner. Root Cause ads make up 840 of the 2,443 on the board, and 318 of them talk to unaware readers. It is the default concept in DTC: "here's the hidden reason you have this problem". It works, which is why it's crowded. But its top-tier rate (8%) sits below the board average.
The middle columns hold the best rates. Systematic Breakdown ads (a step-by-step "here's how this works" argument) reach the top tier 20 to 22% of the time when they talk to problem-aware or solution-aware readers, against 8% for unaware ones. Expert-Based ads show the same shape: 18% for problem-aware, 4% for unaware. An explanation lands better on someone who already knows they have the problem.
Wrong Question works cold. The "you're asking the wrong question" angle does best with unaware readers (15%, 68 ads), where reframing the problem is the whole job.
Some cells are nearly empty. Accidental Discovery has one ad as a primary angle. Insider and Cautionary Tale angles are rare past solution-aware. Empty cells are either bad ideas or untested ones. Your category's own matrix tells you which.
Examples from the high-rate cells
This bread bag isn't for everyone.
- Est. spend
- $1.3M
- Days live
- 301
- Format
- Video, 55s
- Variants
- 2
Systematic Breakdown x Problem-Aware. The reader knows bread goes stale. The ad goes through each storage method (plastic, paper, freezing), says why each fails, then shows how this bag differs.
This is the biggest mistake cat owners make when their cat starts spraying.
- Est. spend
- $187K
- Days live
- 151
- Format
- Video, 139s
Expert-Based x Problem-Aware. The owner already has the problem; the expert frames the usual response as the mistake. In the top tier at 151 days.
What's cheaper, this mason genie or throwing out half a bag of spinach every time?
- Est. spend
- $962K
- Days live
- 439
- Format
- Video, 57s
Wrong Question x Unaware. The viewer wasn't thinking about food waste as a cost until the opener reframed it. Over 14 months live.
This is Maisie, and our story began three years ago.
- Est. spend
- $2.2M
- Days live
- 53
- Format
- Video, 40s
Story x Most-Aware, a small cell (9 ads) with a high rate. A dog owner tells the story of adopting her foster puppy before the product appears. Cheap to make, and the cell is almost empty.
Add format as the third layer
Within each awareness level, the format split shifts too:
| Awareness | Ads | Image share | Top-tier rate, images | Top-tier rate, videos |
|---|---|---|---|---|
| Unaware | 825 | 21% | 7.5% | 8.9% |
| Problem-Aware | 488 | 17% | 16.0% | 9.1% |
| Solution-Aware | 496 | 18% | 12.2% | 11.3% |
| Product-Aware | 125 | 16% | 5.0% | 8.6% |
| Most-Aware | 205 | 28% | 5.2% | 8.8% |
Problem-aware statics stand out (16%, with 83 images). A static image can name a known problem and offer a fix in one frame. For unaware readers, video does slightly better, which makes sense: introducing a problem the viewer hasn't noticed takes more than a headline. See video vs image ads.
Build your own matrix in 30 minutes
CREATIVE DIVERSITY MATRIX Brand: ______ Period: last 90 days
Step 1. List every ad that spent > 1x target CPA.
Step 2. Tag each: primary ANGLE, AWARENESS level, FORMAT (V/S), WINNER (Y/N).
Step 3. Fill the grid (write count, winners in brackets, formats):
| Unaware | Problem | Solution | Product | Most |
Root cause | | | | | |
Systematic | | | | | |
Story / case | | | | | |
Expert | | | | | |
Wrong question | | | | | |
Cautionary tale | | | | | |
Insider | | | | | |
Step 4. Mark: ■ = 3+ ads □ = 1-2 ads (blank) = never tested
Step 5. Pick next tests:
- 1 blank cell that is a high-rate cell on the board (see table above)
- 1 blank cell adjacent to your best winner (same angle, next awareness)
- Format flip: your best video cell -> static version
Step 6. Brief each as a NEW CONCEPT (one dimension changed vs the parent).
Two rules keep the matrix useful:
- One primary angle and one awareness level per ad. If it's ambiguous, pick what the first 15 seconds do. Our awareness levels guide and ad angles guide help with borderline cases.
- Hook pattern and creator live inside the cell. Ten hook variations of one root-cause ad are still one cell. That is the difference covered in concept vs variation.
The fastest way to fill step 2 for competitors is to let something else do the tagging. Ad Radar tags every winning ad on these dimensions, so you can filter your niche by angle and awareness and see which cells your competitors' winners occupy before you brief a gap. Bring the grid to your weekly creative review and update it there.
Figures marked as estimated spend come from Ad Radar's model of engagement on public Meta Ad Library ads. They are estimates, labeled as such, and are best used to rank ads against each other.