| name | avatar-builder |
| description | Use this skill when the user asks Claude to define, build, or refine customer avatars for a brand. Goes beyond demographics into psychographics, beliefs, current behavior, and target behavior. Outputs 2-3 distinct avatars per brand that downstream modes can target with avatar-specific ads, copy, and content. |
Avatar Builder
Purpose: Build specific, actionable avatars. An avatar isn't a demographic — it's a person with a name, a life, beliefs, and current behavior the brand can change.
WHAT A USABLE AVATAR LOOKS LIKE
Bad avatar (demographic)
"Women 35-65, household income $50K+, interested in wellness"
This is useless. You can't write a hook for this person. You can't pick a scene for an image. You can't decide what objection to handle.
Good avatar (specific persona)
The Crafters & Makers Avatar
Sarah, 52, runs a small Etsy ceramics business from her garage studio. Hands cramp after 2 hours of throwing or sanding. She's tried KT tape, ice baths, OTC anti-inflammatories. She doesn't trust "wellness gadgets" but will pay for tools that work. She follows hand-care creators on Instagram. She buys quality once instead of cheap repeatedly. Currently believes hand pain is just part of getting older + part of doing what she loves.
This you can write hooks for. Visuals for. Objections for.
THE 8 AVATAR FIELDS
| Field | What goes here |
|---|
| 1. Name (fictional) | A name that makes the avatar feel real |
| 2. Demographic snapshot | Age, location type, household structure (briefly) |
| 3. Identity / role | What they call themselves first ("I'm a ceramicist" not "I'm 52") |
| 4. The pain (specific) | Not "hand pain" — specific physical and contextual pain |
| 5. Current behavior | What they've tried, what they're using, what they've given up on |
| 6. Beliefs (current) | What they believe about their pain and about solutions |
| 7. Beliefs (target) | What they need to believe to buy |
| 8. Where they hang out | Platforms, communities, content they consume |
CONSTRUCTION PROCESS
Step 1 — Read source material
Brand pack, customer reviews (raw/customers/ if brain exists), competitor angle analysis, survey data, interview transcripts.
Step 2 — Cluster customers by pain context
Don't cluster by demographic. Cluster by:
- WHY they have the pain (occupation, habit, life stage)
- WHEN it shows up (morning, after activity, chronic)
- HOW they currently relate to it (resigned / fighting / desperate / curious)
Step 3 — Name 4-6 candidate avatar clusters
Each cluster gets a working name (e.g., "Crafters & Makers", "Keyboard Athletes", "Morning Stiffness 50+", "Caregivers", "Sports Recovery").
Step 4 — Score each candidate
- Size of segment (rough estimate)
- Brand fit (does this avatar align with the brand's voice and positioning?)
- Reachability (can this avatar be targeted on Meta/TikTok?)
- Conversion potential (specific enough that creative for this avatar would convert?)
Step 5 — Pick 2-3 to develop fully
Generally:
- Primary avatar (the bullseye — biggest segment + best brand fit)
- Secondary avatar (a near-adjacent segment worth testing)
- Optional third (a stretch test — different pain context, same product)
Step 6 — Build the 8-field doc for each
PRIORITY ORDER (for ad sequencing)
When Jake's running a test campaign, ad sets should map to avatars in priority order:
- Primary avatar — highest budget, most creative variants
- Secondary avatar — meaningful budget, 3-4 creative variants
- Stretch avatar — small budget, 2-3 variants — testing the angle, not optimizing yet
OUTPUT FORMAT
Save to brands/<brand>/avatar-sheet.md:
# Avatars — <Brand>
**Last refined:** <YYYY-MM-DD>
**Total avatars:** [count]
## Avatar 1 (PRIMARY): <Name + Identifier>
### Snapshot
[3-4 sentence vignette — like the Sarah example above]
### Demographic
- Age range: ...
- Location: ...
- Household: ...
### Identity / role
- "I am a [thing they call themselves]"
### The pain (specific)
- Physical: ...
- Contextual: ...
- When it shows up: ...
### Current behavior
- Has tried: [list]
- Currently using: [list]
- Has given up on: [list]
### Beliefs (current)
- About the pain: ...
- About solutions: ...
- About brands in this category: ...
### Beliefs (target — what they need to believe to buy)
- [Belief 1] — see necessary-beliefs.md #N
- [Belief 2] — see necessary-beliefs.md #N
- ...
### Where they hang out
- Platforms: ...
- Communities: ...
- Content they consume: ...
### Ad set notes
- Hook tone: ...
- Visual tone: ...
- Awareness level: ...
---
## Avatar 2 (SECONDARY): <Name + Identifier>
[Same structure]
---
## Avatar 3 (STRETCH): <Name + Identifier>
[Same structure]
---
## Rejected candidates (with reason)
- [Avatar] — rejected because [too small / off-brand / not reachable / etc.]
## [VERIFY] flags
- [Anything based on assumption, not primary data — needs survey or interview]
INTEGRATION
necessary-beliefs.md → cross-reference which beliefs apply to which avatarsawareness-matrix.md → map each avatar to their awareness stagestatic-ad-generator/ad-concept-generator → reads avatars to generate concepts per avatarmeta-ads-operator/campaign-launcher → builds ad sets per avatartiktok-slideshow/hook-bank-generator → tags hooks by target avatardtc-second-brain → after performance data, validate which avatars actually convert
