---
name: static-ad-designer
description: Activates when the user wants to design Meta or Instagram static ads, plan creative variations, or build a creative matrix for paid social testing. Handles the strategic and compositional layer of static ad creation. Calls the nano-banana-ads skill for actual image generation. Always references brand-kit.md for brand identity.
---

# Static Ad Designer Skill

This is the strategy + layout layer of Pillar 1's Static Ads Engine. It defines what makes a good Meta static ad, plans the creative matrix, and provides the rules that the nano-banana-ads skill follows when generating images.

## Activation Rules

Activate when:
- The user asks to "design ads," "plan a creative matrix," "build static ads," or anything related to Meta/Instagram static ad creation
- The user runs any of the slash commands defined below
- Another skill (creative-strategist, weekly-creative-report) recommends new ad concepts to produce

## Prerequisites

- `brand-kit.md` exists in project root
- `nano-banana-ads` skill is installed (this skill calls it)

---

## Layout Rules (applied to every generated image)

### Visual hierarchy (mandatory)

Every static ad must have a clear visual hierarchy in this order:
1. **Headline** (most prominent element)
2. **Supporting visual** (product, lifestyle scene, or graphic)
3. **CTA button or label** (least prominent but immediately readable)

Avoid: Multiple competing focal points, decorative text that distracts from the headline, CTAs that get lost.

### Text rules

- Maximum 10 words on the image itself
- Minimum 48px equivalent for headline text (must be readable on a phone in a feed scroll)
- Always 20% padding from edges (no edge-bleeding text)
- Never put text over busy backgrounds without a solid or semi-transparent overlay
- Use brand headline font from brand-kit.md
- Maintain WCAG AA contrast (4.5:1 minimum)

### Color rules

- All colors must come from brand-kit.md
- Never invent new shades
- High contrast between text and background (always)
- CTA buttons in the brand accent color

### Composition rules

- Product (or main subject) is the visual anchor
- Use rule of thirds for product placement (avoid dead-center compositions unless the brand explicitly calls for symmetry)
- Negative space is your friend; don't fill every corner
- Asymmetric compositions outperform symmetric ones for stopping the scroll

### Format requirements

- Default aspect ratio: 1:1 (1080x1080)
- Also generate 4:5 (1080x1350) for max feed real estate
- 9:16 (1080x1920) only when explicitly requested for stories/reels
- PNG format, RGB color space
- Resolution: 1K for testing, 2K for production

### What to avoid (anti-slop checklist)

- Stock-photo aesthetic (overly polished people smiling at the camera)
- Cluttered layouts with more than 3 visual elements competing for attention
- Tiny text that disappears at thumbnail size
- Generic headlines that could apply to any product in the category
- Centered composition by default (lazy, predictable)
- Pure white backgrounds without any texture, shadow, or depth
- "AI-generated" looking lighting (perfectly even, no shadow direction)

---

## The Creative Matrix Method

The matrix is the strategic backbone of this skill. Random generation produces slop. Matrix generation produces a structured testing plan where every ad has a specific hypothesis behind it.

### Matrix structure

```
HOOKS (4 standard angles):
1. Problem-centric: lead with the pain point
2. Benefit-centric: lead with the positive outcome
3. Social proof: lead with a customer result or testimonial
4. Curiosity: lead with a surprising fact or question

PERSONAS (3-4 from brand-kit.md):
[Use the personas defined in your brand kit]

VISUAL STYLES (2-3 standard):
1. Minimal: white/light background, product centered, bold typography
2. Lifestyle: product in real-world use, warm lighting, contextual
3. UGC: phone-camera aesthetic, casual composition, "user-generated" feel
```

Cross-multiply: 4 hooks × 4 personas × 2 styles = 32 unique concepts.

Each concept gets:
- Concept name (using brand-kit naming convention)
- Headline text (10 words max)
- One-sentence visual description
- Primary text (the Meta ad copy field, 2-3 sentences)
- CTA button label

---

## Slash Commands

### `/build-matrix`

Interactively build a creative matrix from the user's brand kit.

**Workflow:**
1. Read brand-kit.md
2. Confirm or extract: hooks (default 4), personas (from brand kit), styles (default 2)
3. Generate all combinations
4. For each combination, generate the headline + visual description + primary text + CTA
5. Save to `./creative-matrix.json`
6. Show the matrix table for review
7. Prompt user to edit any concepts before generation

### `/build-matrix-from-spy [spy-report.md]`

Build a creative matrix informed by competitor research from Pillar 5 (Impressions Spy).

**Workflow:**
1. Read the spy report
2. Identify top-performing hooks, formats, and offers across competitors
3. Build a matrix that:
   - Replicates the top 3 winning hook patterns from the spy report
   - Tests the top untested gap angles identified in the spy report
   - Uses the personas from brand-kit.md
4. Output `./creative-matrix.json`

This is the strategic compounding move — research → production → distribution.

### `/refine-matrix [feedback]`

Iterate on an existing creative matrix based on performance data or strategic feedback.

**Usage:**
```
/refine-matrix "Drop the curiosity hook variants — none of them performed last month. Add more variations of the social proof hook."
```

### `/generate-from-matrix`

Hand off to the nano-banana-ads skill to generate all images from the current matrix.

This is a one-line wrapper around `/generate-ads-batch ./creative-matrix.json` that ensures the static ad designer rules are applied to the prompt before generation.

---

## Hook Templates

When generating headlines for each hook category, use these structures as starting points (then customize for the specific persona and product):

### Problem-centric

- "Still [doing painful thing]?"
- "The [thing your audience hates] no one talks about."
- "Why [common method] doesn't work for [persona]."
- "[Number] reasons your [thing] keeps [bad outcome]."

### Benefit-centric

- "[Specific outcome] in [timeframe]."
- "Finally, [thing they want] without [thing they don't want]."
- "How [persona] gets [outcome]."
- "[Outcome] is closer than you think."

### Social proof

- "What [number] [people like persona] discovered last month."
- "'[Direct customer quote, ≤8 words].'"
- "[Specific result] is what [persona] is getting."
- "Why [number]+ [persona] switched."

### Curiosity

- "The [thing] that [unexpected outcome]."
- "What your [expert/authority] forgot to tell you."
- "There's a faster way."
- "[Counterintuitive claim that makes you stop scrolling]."

---

## Visual Description Templates

For each concept, the visual description goes to the nano-banana-ads skill. Format:

```
A [composition type] showing [main subject] in [setting/context]. 
[Lighting description]. [Color palette reference from brand-kit.md]. 
[Typography placement]. [Anti-slop notes if applicable].
```

**Example:**
```
An asymmetric overhead shot showing the VitaPure jar on a textured 
linen surface in the bottom-right third of the frame. Soft warm 
afternoon light from upper-left. Cream and forest color palette 
from brand kit. Headline in top-left corner using Fraunces font, 
primary forest hex (#2d3a2c). Negative space dominates the upper-left. 
Real shadow from the jar — no flat lighting.
```

---

## Cross-Skill Integration

**Calls into this skill:**
- The user (directly via slash commands)
- creative-strategist (Pillar 3) when recommending new batches based on performance data
- impressions-spy (Pillar 5) when building matrices from competitor research

**This skill calls:**
- brand-kit (read brand identity)
- nano-banana-ads (for actual image generation)

**Output flows to:**
- Pillar 2 (Bulk Uploader) for distribution to Meta Ads Manager
- Pillar 3 (Creative Analytics) once ads have run for 14+ days

---

## Output Format

After running `/build-matrix`, return:

```
Built creative matrix with [N] concepts.
Saved to: ./creative-matrix.json

Concepts:
[Table showing first 8 concepts with concept name, headline, persona, hook, style]

[Show "... and N more" if matrix exceeds 8]

Estimated generation cost (1K, both aspect ratios): $[N * 0.07 * 2]

Next steps:
1. Review the matrix above
2. Edit any concepts that don't fit
3. Run /generate-from-matrix to produce the images
```

---

## Notes

- The matrix size of 32 (4×4×2) is the recommended starting point. Larger matrices spread budget too thin to learn anything. Smaller matrices don't generate enough variations to find winners.
- After your first 14 days of testing, use Pillar 3's creative-strategist skill to identify which hook/persona/style combinations won. The next matrix should be weighted toward winners.
- This skill produces strategic concepts. The nano-banana-ads skill produces images. The creative-analyzer skill (Pillar 3) tells you which ones worked. The loop is the system.