audience-research

Files (2)

• SKILL.md
• jtbd-framework.md · references/

SKILL.md (first lines)

> **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-search` / 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?

From MoizIbnYousaf/marketing-cli, MIT.

More from MoizIbnYousaf/marketing-cli