Meta Ads Creative

---
name: meta-ads-creative
description: Build and judge Meta ad creatives — deconstruct a winning reference, write the hook, apply the build and design rules, and produce a creative spec with variants ready to test. Use when asked to make a Meta ad, design an ad creative, write ad hooks, build a video ad for Reels, design a static ad, judge or review an ad creative, fix ad fatigue, diagnose why ads stopped working, plan a creative refresh, or someone says 'my ads suck'. For 100+ copy variations at volume, see bulk-ad-variations. For campaign objectives and budget, see meta-awareness-planner. For the page the ad clicks through to, see landing-page-copy.
metadata:
  version: 1.0.0
---

# Meta ads creative

**A weak result is usually a weak creative, not a small budget.** Before anyone
touches the budget, this skill fixes the thing people see: the hook, the build, the
design, and the variety.

The goal in one sentence: turn a brief into a creative spec that passes the checklist
before a rupee is spent, and diagnose a live creative before anyone blames the money.

## Before you start

| Input | Why |
| --- | --- |
| Product, price, offer | Every claim on the creative comes from here |
| Brand lock — logo path, hex colours, font | Brand lock is absolute; a creative that breaks it does not ship |
| Format target — static feed, Reels/Stories video, or both | The build rules differ: 1080×1350 static vs 9:16 video under 15s |
| A winning reference ad, if one exists | Step 1 deconstructs it; without one, start from the persona instead |
| Real social proof, if any — ratings, reviews, logos | Used only if supplied. Never invented |

If the brand lock or the product details are missing, stop and ask. A creative built
on guessed colours and invented claims is rework, not progress.

## Step 1 — Deconstruct the reference

If a proven winner exists, reverse-engineer why it works before creating anything.
Study six things:

1. Layout grid and visual hierarchy — what the eye hits first, second, third
2. Hook and headline placement and size
3. Subhead, CTA button, and offer badge positions
4. Social proof — ratings, reviews, logos
5. Colour system, background treatment, product placement
6. **The job each element does, and why it is there**

Output of this step is a written spec of the proven skeleton, reusable with fresh
messaging. No reference? Build the persona first — the angle comes from their
Problems and Questions, not from the product sheet.

## Step 2 — Write the hook

The first 3 seconds of a video, or the first thing the eye hits on a static. Three
checks, from the creative checklist:

| Check | Tip |
| --- | --- |
| Something grabs attention in the first 3 seconds | Motion, a face, or a bold line |
| It works with the sound off | Most people watch on mute |
| The brand shows up early, not only at the end | You want it *remembered* |

The hook carries one angle. A hook mixing two arguments lands as neither.

## Step 3 — The build

| Check | Tip |
| --- | --- |
| One clear idea, not five messages | Awareness is about one memory |
| Video: vertical full-screen 9:16 | Fills the phone on Reels and Stories |
| Video: under 15 seconds | Ideal for ThruPlay — holds attention to the end |
| Static: 1080×1350 (4:5) | Unless the placement demands otherwise |
| Feels native and real, not like a TV ad | Raw and relatable wins in feed |

## Step 4 — The design pass

Three rules that separate clean creatives from amateur ones, with the numbers:

1. **Keep text away from the edges.** Wide ads: at least **10% of the ad width** as
   padding. Square or vertical ads: at least **15%**. On a 2400px-wide ad that is
   240px minimum — measure it, do not eyeball it.
2. **Clean line height.** The litmus test: a capital **E should fit snugly** between
   two lines of text — measured from the bottom of the top line's descenders to the
   top of the bottom line's ascenders.
3. **Fill lines wide before wrapping.** Lump the text into solid chunks rather than
   ragged fragments.

Plus the brand lock: correct logo, colours, and font on every creative. Never
distort the product photo. Legible on a phone, no spelling errors.

## Step 5 — Variety

One creative is not a plan. From the checklist:

| Check | Tip |
| --- | --- |
| At least 2–3 versions to test | Let the audience pick the winner |
| Different hooks across the versions | Test what stops the scroll |
| A refresh planned every 2–3 weeks | Beat ad fatigue before it hits |

Name every file `persona_angle_variant.png` so a winner in Ads Manager traces back
to the exact angle that won. And the shortlist rule: **volume is for ideas, not for
the live account** — rank, then test 15–20 distinct angles at most.

## Step 6 — Judge a live creative

When performance data exists, read it against the benchmarks before changing spend:

| Signal | Threshold | Verdict |
| --- | --- | --- |
| ThruPlay rate | 20% is the benchmark; above 15% is strong | Healthy creative |
| ThruPlay rate | Below 5% | Weak creative — fix the video, not the budget |
| Cost per person who remembers you | Above ₹85 (≈ $1) | Refresh the creative |
| Strong start, decaying results | 2–3 weeks in market | Fatigue — rotate the next variant in |

## Output

```
# Creative spec: <product> — <angle>

## Reference deconstruction        (if a reference was supplied)
Layout: ...   Hook placement: ...   Proof: ...   Colour: ...

## Hook
<the first 3 seconds / first eye-hit, one angle, works on mute>

## Build
Format: <9:16 video <15s | 1080x1350 static>
Idea:   <the one message>

## Design check
Padding: <n>% (min 10% wide / 15% vertical)   Line height: E-test pass
Brand lock: logo / colours / font confirmed

## Variants
| File (persona_angle_variant) | Hook | What it tests |
[2-3 rows minimum, different hooks]

## Refresh plan
Next rotation: <date, 2-3 weeks out>
```

## When it breaks

| What you see | What it means | The fix |
| --- | --- | --- |
| Results weak, first instinct is more budget | The creative is weak, not the spend | Run step 6 before touching money |
| ThruPlay under 5% | The hook loses people immediately | Rewrite the first 3 seconds; check it works on mute |
| Strong for two weeks, then decay | Ad fatigue | Rotate a fresh variant; plan refresh every 2–3 weeks |
| Creative looks like a TV ad | Polished but not native | Raw and relatable wins in feed; shoot for the phone |
| Five messages in one creative | No single memory to keep | One idea per creative; move the rest to other variants |
| Text looks cramped or amateur | Padding under 10%/15%, or ragged line height | Apply the step 4 numbers; measure, do not eyeball |
| All variants perform identically | Same hook restyled three ways | Different hooks per variant, not different colours |
| Winner cannot be traced to an angle | Files named final_v2_FINAL.png | Rename to persona_angle_variant and relaunch tracking |
| Asked for ratings or reviews you do not have | Proof does not exist yet | Drop the proof element. Never fabricate it |

Never respond to a weak creative by raising the budget, and never retry a failing
hook by re-rendering it prettier. Diagnose against the checklist first.

## Rules

- **Never invent social proof.** No fake ratings, review counts, or testimonials —
  a fabricated claim in an ad is a legal problem, not a creative choice.
- **Never fake a dashboard or screenshot with an image model.** Real interfaces get
  real screenshots; generated images are for concepts.
- **Brand lock is absolute.** Change the layout, never the logo, colours, or font.
- **One idea per creative,** because awareness buys one memory, not five.
- **Never ship a single creative.** 2–3 versions minimum with different hooks; the
  audience picks the winner, not your gut.
- **If the brand lock or product facts are missing, stop and ask** — guessing
  produces confident rework.

## Related skills

- **bulk-ad-variations** — when one good creative needs to become 100+ on-brand
  variations (Persona × Angle × Offer at volume).
- **meta-awareness-planner** — whether to run the campaign at all, which objective,
  and how much budget; judge results there, creatives here.
- **landing-page-copy** — the page the ad clicks through to; message match between
  ad and page is decided there.