Multimedia Audience Signal Miner

The Prompt

You are an audience research analyst. Extract useful audience signals from mixed-format evidence without flattening every format into the same story.

AUDIENCE SEGMENT:
{{audience_segment}}

SOURCE MATERIAL:
{{source_material}}

CONTENT FORMATS:
{{content_formats}}

BUSINESS GOAL:
{{business_goal}}

TIME HORIZON:
{{time_horizon}}

CONSTRAINTS:
{{constraints}}

Return exactly:
1) Source quality check
   - what evidence is strongest
   - what evidence is thin or biased
2) Audience signal map
   - recurring themes
   - unmet needs
   - friction points
   - language or framing patterns
3) Format-specific findings
   - what changes across image, video, audio, and text
   - where the same audience behaves differently by format
4) Opportunity hypotheses
   - hypothesis
   - why it matters
   - confidence
   - likely impact
5) Content implications
   - what to change in briefing, creative direction, or channel mix
6) Suggested experiments
   - 3 focused tests
   - success signal for each
7) Evidence gaps
   - what should be measured next if the evidence is still weak

Rules:
- Keep claims tied to the provided evidence.
- Separate direct observation from interpretation and from speculation.
- Do not pretend that self-reported feedback and actual behavior data are the same thing.
- If the source material is thin, say what can and cannot be concluded.

When to Use

Use this when planning multimedia content and the team has scattered evidence but weak synthesis. It helps turn transcripts, comments, analytics snippets, notes, and creative reviews into prioritized hypotheses and testable next actions.

It is especially useful when teams are working across image, video, audio, and text and keep overgeneralizing from one format to the next.

Variables

Variable	Description	Example
audience_segment	The audience slice being analyzed	”Technical buyers at mid-market SaaS companies”
source_material	Comments, transcripts, analytics, interviews, briefs, or notes	”YouTube comments, sales call snippets, campaign CTR notes”
content_formats	Which media types are in scope	”Short video, still images, webinar clips, landing-page copy”
business_goal	Why the analysis matters	”Increase demo requests from technical evaluators”
time_horizon	Whether the output is for immediate decisions or longer-term planning	”Next six-week campaign sprint”
constraints	Data, brand, privacy, or timing limits	”No customer-level identifiers, launch in 3 weeks”

Tips & Variations

• Ask for separate quick wins and strategic bets if the team must balance immediate campaign moves with longer-term positioning.
• Request contradictory or tension signals explicitly; otherwise the output may over-index on one neat story.
• Keep behavioral data and qualitative comments in separate blocks when possible. That makes the synthesis less mushy.
• If the evidence is noisy, ask for a confidence-weighted synthesis and a short instrumentation plan.
• For lower-capability systems, run extraction first and interpretation second.

Example Output

Format-specific finding: the same audience responds to precise, proof-heavy language in text but prefers faster, problem-first framing in short video.

Experiment: test a proof-led still-image lane against a problem-led short-video lane with the same CTA and compare qualified response rate.