audience-analysis

Systematic audience research — demographics, psychographics, Jobs-to-be-Done, buyer personas, and voice-of-customer analysis. Use when understanding target audiences for products, services, or content.

Model	Source
sonnet	pack: recon

Full Reference

Audience Analysis

Announcement

┏━ 🔍 audience-analysis ━━━━━━━━━━━━━━━━━━━━━━━━━┓ ┃ your friendly armadillo is here to serve you ┃ ┗━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┛

Audience analysis is structured research to understand who buys, why they buy, what language they use, and what obstacles they face. Output is an Audience Brief — a single reference doc that drives messaging, positioning, and product decisions.

Reference Index

I want to…	Reference File
Apply JTBD, persona templates, psychographic profiling, or segmentation frameworks	frameworks.md
Mine reviews, forums, social media, or interviews for raw VOC data	voice-of-customer.md
Format the final Audience Brief deliverable	output-format.md

Research Process

Phase 1: Scope Definition

Before any research, lock down:

Decision	Why
Product / service being analyzed	Frames all subsequent research
Primary vs. secondary audiences	Avoids conflating distinct segments
Research depth (light / standard / deep)	Light = 2-3 sources; Standard = 5-7; Deep = 10+ with interviews
Existing data available?	CRM, analytics, past surveys — use before going external

Ask the user ONE question if scope is unclear:

▸ What product or service are we researching, and do you have any existing customer data (reviews, CRM exports, survey results)?

Phase 2: Data Collection

Run these in parallel where possible:

1. Review mining       → Amazon, G2, Capterra, Trustpilot, App Store
2. Forum research      → Reddit, Quora, niche communities, Facebook Groups
3. Social listening    → Twitter/X, LinkedIn, TikTok comments
4. Competitor analysis → Their positioning, reviews, messaging gaps
5. Interview synthesis → If transcripts or notes are provided

Deep review mining: Use firecrawl to scrape review pages (Amazon product reviews, G2 category pages, Trustpilot company pages) for full VOC data. Rate limit: max 5 firecrawl pages per review platform.

Source priority: direct customer language > third-party reviews > social > inference

Never invent quotes. Flag all inferences with [inferred].

Phase 3: Framework Application

Apply these in sequence — each builds on the last:

Demographics     →  Who they are (age, income, role, geography)
Psychographics   →  How they think (values, beliefs, identity)
JTBD             →  What they're hiring the product to do
Pain Catalog     →  What's broken before finding this solution
Objection Map    →  What stops them from buying
Trigger Events   →  What causes them to start searching

Full framework details: frameworks.md

Phase 4: VOC Extraction

Pull exact customer language from raw sources:

Emotional phrases (frustration, aspiration, relief)
Before/after language (“I used to… now I…”)
Specific pain descriptors (not paraphrased)
Feature requests stated as problems

Methodology: voice-of-customer.md

Phase 5: Audience Brief Assembly

Synthesize everything into the standard brief format.

Template and field guidance: output-format.md

Quick Persona Snapshot

For fast turnarounds, generate a single primary persona:

Name:           [Archetype name — e.g., "Overwhelmed Ops Manager"]
Age range:      [25-34]
Role/context:   [Who they are in relation to the product]
Primary job:    [The main JTBD — functional]
Emotional job:  [How they want to feel]
Top pain:       [Single biggest friction point]
Top trigger:    [What causes them to search for a solution]
Top objection:  [What nearly stops them from buying]
VOC quote:      ["Exact words from a real customer"]

Depth Guidelines

Depth	Sources	Time Estimate	Output
Light	2-3 sources, no interviews	30-60 min	Single persona + pain list
Standard	5-7 sources, optional interviews	2-4 hrs	Full brief, 2-3 personas
Deep	10+ sources + interviews	1-2 days	Full brief, 4-6 personas, segment map

Default to Standard unless user specifies otherwise.

Common Research Sources by Business Type

Business Type	Best Sources
SaaS / software	G2, Capterra, Reddit r/[category], Product Hunt comments
E-commerce	Amazon reviews, Trustpilot, Reddit, TikTok comments
Local services	Google reviews, Yelp, Nextdoor, Facebook Groups
B2B / agency	LinkedIn posts, case study comments, industry forums
Consumer apps	App Store reviews, Reddit, Twitter/X threads
Healthcare / wellness	Healthgrades, Reddit r/[condition], forums

Output Checklist

Before delivering the Audience Brief:

At least one direct VOC quote per persona
JTBD stated in customer language (not product language)
Trigger events identified (not just pain points)
Objections mapped with severity
No invented quotes — all inferences flagged
Messaging implications section complete

Full template: output-format.md