Note: Examples below use fictional brands (Acme, Lumi, Helm). Replace with your own brand context.

/audience-research -- Know Who You're Talking To

Every marketing skill gets better when you know your audience. Generic copy happens when you write for "everyone." Specific copy that converts happens when you write for a real person with real problems.

This skill builds that audience profile. It's the foundation — 8 of 11 content skills read audience.md to shape their output. Running this first makes everything downstream sharper.

No SaaS tools needed. Systematic research plus web search.

On Activation

1. Check if brand/ directory exists in the project root.
2. If it does, read available files: voice-profile.md, positioning.md, audience.md, competitors.md, creative-kit.md, stack.md, learnings.md.
3. Apply any loaded brand context to enhance output quality — skip questions the user has already answered.
4. If brand/ does not exist, proceed without it — this skill works standalone.

Iteration Detection

Before starting, check whether ./brand/audience.md already exists.

If audience.md EXISTS --> Update Mode

Do not start from scratch. Instead:

1. Read the existing audience profile.

2. Present a summary of current personas:

   EXISTING AUDIENCE PROFILE
   Last updated {date} by /audience-research
   
   Personas:
   ├── {Persona 1 name}    {one-liner}
   ├── {Persona 2 name}    {one-liner}
   └── Primary: {primary persona name}
   
   Watering holes: {N} mapped
   
   ──────────────────────────────────────────────
   
   What would you like to do?
   
   1. Refine existing personas with fresh research
   2. Add a new persona segment
   3. Deep-dive into watering holes
   4. Full rebuild from scratch
   

3. Process the user's choice:

   • Option 1 --> Re-run research for existing personas, update with new data
   • Option 2 --> Identify the new segment, build persona, merge into existing file
   • Option 3 --> Focus on community mining for existing personas
   • Option 4 --> Full process from scratch

4. Before overwriting, show what changed and ask for confirmation.

If audience.md DOES NOT EXIST --> Full Research Mode

Proceed to the full process below.

The core job

Build a complete picture of who buys this product, why they buy it, where they hang out, and what language they use. The output is structured enough for other skills to parse but human enough to be useful for strategy.

Approach selection

Choose based on what's available:

Quick Profile (L0 — zero context)

When no brand files exist and minimal information is provided. Ask 5 key questions:

1. What does your product do? (one sentence)
2. Who currently uses it? (or who do you want to use it?)
3. What problem does it solve? (the pain before, the relief after)
4. What alternatives exist? (what would they do without you?)
5. What's the price point? (free, $10/mo, $500/mo, enterprise?)

From these 5 answers, generate a working persona. It won't be perfect but it gives every downstream skill something to work with.

Persona Build (L1-L2 — some context)

When you have basic product info or existing brand files. Systematic persona construction using demographics, psychographics, and jobs-to-be-done.

Community Mining (L3-L4 — full research)

When Exa MCP or web search is available. Deep research into where the audience lives online, what language they use, what they complain about, and what they aspire to.

The audience research process

Step 1: Identify the transformation

Same as positioning — start with the outcome, not the product.

• What does the customer's life look like BEFORE? (the pain state)
• What does it look like AFTER? (the desired state)
• What's the gap between those two states?

The gap defines who your audience is: people stuck in the BEFORE state who want the AFTER state but haven't found a good way to get there.

Step 2: Build buyer personas

For each distinct audience segment (usually 1-3), build:

Demographics (who they are):

• Role/title, company size, industry
• Age range, income level, location
• Technical sophistication

Psychographics (how they think):

• Goals and aspirations
• Fears and frustrations
• Values and beliefs about this category
• How they make decisions (data-driven, social proof, gut feel)

Jobs-to-be-done (why they hire your product — see references/jtbd-framework.md for the full framework, Forces of Progress, and job mapping template):

• Functional job: What task are they trying to accomplish?
• Emotional job: How do they want to feel?
• Social job: How do they want to be perceived?

Language patterns (how they talk):

• Words they use to describe their problem
• Jargon level (none, industry-standard, expert)
• Tone they respond to (casual, professional, irreverent)

Step 3: Find watering holes

Where does this audience already gather? This is where you'll distribute content.

Search process (use Exa MCP when available):

1. Search for "[problem/topic] + reddit/forum/community"
2. Search for "[job title/role] + newsletter/podcast/blog"
3. Search for "[competitor name] + reviews/alternatives"
4. Look at who engages with competitor content on social platforms

Map the landscape:

──────────────────────────────────────────────────

  AUDIENCE WATERING HOLES

──────────────────────────────────────────────────

  Online Communities
  ├── r/[subreddit] — [activity level, relevance]
  ├── [Forum/Discord] — [activity level, relevance]
  └── [Facebook Group] — [activity level, relevance]

  Content They Consume
  ├── [Newsletter] — [subscriber count if known]
  ├── [Podcast] — [relevance]
  └── [Blog/Publication] — [relevance]

  Social Platforms
  ├── [Platform] — [how they use it]
  └── [Platform] — [how they use it]

  Events & Conferences
  └── [Event] — [relevance]

──────────────────────────────────────────────────

Step 4: Map pain points and objections

For each persona, identify:

Pain points (what hurts today):

• What triggers the search for a solution?
• What's the cost of NOT solving this? (time, money, stress)
• What have they tried before that didn't work?

Objections (what stops them from buying):

• Price objections: "Is it worth the money?"
• Trust objections: "Will this actually work for me?"
• Effort objections: "Is this going to be hard to implement?"
• Timing objections: "Is now the right time?"
• Alternative objections: "Can I just do this myself?"

Step 5: Identify language mines

Search for actual quotes from the audience — Reddit posts, review sites, forum threads, social media. These are gold for copywriting.

Look for:

• How they describe their problem (use their exact words)
• Emotional language they use (frustrated, overwhelmed, stuck)
• Aspirational language (wish I could, dream of, finally)
• Comparison language (better than, unlike, switch from)

Output format

Write ./brand/audience.md with this structure. The ## Archetypes block at the top is the structured contract consumed by Studio's archetype-pulse-cards and any future skill that needs a card-shaped persona summary. The fuller persona blocks below it carry the long-form context.

---
title: Audience Profile
type: audience-research
skill: audience-research
date: [ISO date]
personas: [number]
primary_persona: [name]
archetypes: [number]
---

# Audience Profile — [Product/Project Name]

## Archetypes

Card-shaped summary of every persona. One archetype per persona. Required fields per archetype: `name`, `one_liner`, `demographic`, `top_pain`, `top_desire`, `watering_hole`, `language_quote`. Do not skip fields. If you do not have evidence for one, write "not measured" rather than fabricating.

### Archetype: [Name]

- **one_liner:** [Who they are and what they want, one sentence]
- **demographic:** [Role, company size, technical level — single line]
- **top_pain:** [The single biggest