---
name: consumer-ai-creator-research
description: Build and score creator/influencer target lists for Consumer AI Factory venture validation, including platform discovery, fit scoring, contact-path checks, and first-wave outreach prioritization.
tags:
  - consumer-ai
  - creator-research
  - influencer-research
  - distribution
  - partnerships
  - validation
---

# Consumer AI Creator Research

## When to use

Use this skill when Antoine asks to find influencers, creators, newsletters, podcasts, specialists, or communities for a Consumer AI Factory venture, especially for early validation or distribution tests.

Examples:

- Busy mom / working mom / single mom creators for Alma.
- Relationship creators for HappyCouple / Emma.
- ADHD, productivity, anti-doomscrolling, beauty, parenting, or other niche creator lists.
- Paid UGC candidate lists where creator fit matters more than celebrity reach.

For HappyCouple-specific outreach operations, also load `happycouple-creator-partnership-agent` because it contains relationship-domain language, outreach rules, and logging conventions.

## Core stance

Creator research is not just a list of large accounts. For early consumer AI validation, prioritize creators whose audience already voices the exact pain the product is trying to solve.

Strong candidates usually have:

- repeated audience pain in comments or captions
- practical advice content rather than pure aesthetics
- trust with a narrow persona
- a reachable contact path
- plausible willingness to try a tiny experiment
- low likelihood of already building the same AI/software product

Do not over-index on follower count. Mid-sized and niche creators are often better first-wave targets than category celebrities.

## Discovery workflow

1. Define the venture, persona, and one-sentence user pain.
2. Split discovery by niche, not just platform:
   - pain educators
   - routine/system creators
   - day-in-the-life creators
   - newsletters/podcasts/communities
   - UGC creators for paid tests
3. Search using official sources first:
   - creator website/about page
   - TikTok/Instagram profile pages where indexed
   - podcast/newsletter pages
   - media kit/contact pages
   - creator marketplaces only as secondary evidence
4. Capture contact paths, but classify quality:
   - official creator email
   - agency/manager email
   - official form
   - DM only
   - unverified/blocked
5. Separate first-wave outreach candidates from benchmarks.

## Required output fields

For each candidate, return:

```text
name
brand_or_handle
niche
persona_fit
platforms
primary_url
secondary_urls
rough_audience_size_if_available
recent_relevant_content_or_source_evidence
why_fit
cautions
contact_path
recommended_experiment
classification: outreach_now | paid_ugc_candidate | later | benchmark_only | avoid
```

If the list is quick and tactical, it is fine to omit scoring and provide a short ranked list. If it will drive outreach or spend, score it.

## Scoring rubric

Score each 1-5:

- Persona fit: does their audience match the venture's target user?
- Pain evidence: do posts/comments/captions show the target pain?
- Trust: does the creator provide practical help the audience likely believes?
- Distribution: can they plausibly move enough users for signal?
- Contactability: is there a credible outreach path?
- Experiment fit: can the product be tested through a tiny creator scenario?
- Brand safety: low controversy, no obvious mismatch, no clinical/regulatory boundary risk.

Use the score to prioritize, not to pretend the research is more precise than it is.

## Classification rules

- `outreach_now`: strong persona/pain fit, reachable, plausible small test.
- `paid_ugc_candidate`: useful for creative/ad tests, even if their organic audience is weaker or smaller.
- `later`: good fit but too large, too hard to reach, or needs better product proof first.
- `benchmark_only`: use for language/category learning, not first cold outreach.
- `avoid`: wrong audience, risky brand, poor contact path, or likely competitor conflict.

## Recommended first ask

Early creator outreach should ask for a tiny private test, not a big launch or sponsorship.

If the product's core activation path is not live yet (for example SMS/WhatsApp verification is pending), still talk to creators, but only for discovery and private validation. Do not ask them to post, send their audience, or accept paid distribution until the user path works. Use the downtime to ask whether the promise feels useful, safe, and specific enough, then line up a small test wave for when activation is live.

Shape:

```text
Your audience already deals with [specific pain]. I’m testing [product], a lightweight tool for [specific job]. Would you be open to trying 3 real scenarios and telling me if it feels useful?
```

For Alma-style busy-mom tools, the better pitch is:

```text
Your audience already carries tiny household tasks in their head all day. I’m testing Alma, a text-message assistant where a mom can send one messy thing — dinner, school email, grocery list, schedule chaos — and get a useful next step back. Would you be open to trying 3 real mom scenarios and telling me if it feels useful?
```

### If the core product action is blocked

If the product's primary action is not live yet, for example Alma SMS awaiting Twilio toll-free verification, creator outreach can still start, but only as **discovery and private validation**.

Do:

- Ask for the creator's read on whether the product would resonate with their audience.
- Ask what would make the product feel useful, safe, and trustworthy.
- Build a warm list for private testing once the blocked action works.
- Offer to send a quick example or 3 scenarios, not a public link.

Do not:

- Ask them to post publicly.
- Pay for distribution.
- Send their audience to a dead CTA or broken activation loop.
- Interpret weak creator response as product rejection if the core experience cannot yet be tried.

Use a low-pressure note like:

```text
Hey [Name],

I’m building Alma, a text-message assistant for busy moms.

The idea: a mom can send one messy thing, like dinner, a school email, groceries, or schedule chaos, and Alma sends back the next useful step.

We’re still prototyping, so I’m not asking you to post anything. I’d love your honest read on whether this would resonate with your audience, and what would make it feel more useful or trustworthy.

Open to me sending a quick example?

Best,
[Name]
```

## Pitfalls

- Do not present a list as outreach-ready if contact paths have not been verified.
- Do not mix celebrity benchmarks with first-wave targets without labeling them.
- Do not rely only on creator marketplace pages; verify identity through official site/social links when possible.
- Do not pitch “AI” generically. Anchor the ask in the user's concrete moment.
- Do not send outreach autonomously unless Antoine explicitly approves the batch in the current conversation.
- Do not log every quick brainstorm as a permanent partnership queue entry; only outreach/intended-test lists need durable operational logging.

## Support files

- `references/alma-busy-mom-influencer-seed-list.md`: seed target list and buckets from the Alma busy-mom influencer research session.