# Plan Customer Reach

> Plan consent-aware customer research recruitment and draft-only outreach without sending messages. Use when defining interview segments, recruiting participants, selecting channels, or preparing 고객 인터뷰 대상자·모집·컨택 초안. Do not use for sending the outreach, running the interview, or synthesizing completed transcripts.

- Skill: `kimsanguine/plan-customer-reach` (Agent Skill, multi-file: 3 files)
- Install (CLI): `npx skillmds@latest add kimsanguine/plan-customer-reach`
- Raw SKILL.md: https://api.skillmd.com/api/skills/kimsanguine/plan-customer-reach/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Research & Search
- Author: kimsanguine (https://skillmd.com/u/kimsanguine)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/kimsanguine/plan-customer-reach

---


# Plan Customer Reach

Create a research recruitment plan that separates who to learn from, why their experience matters, and what requires human approval.

## Inputs

Require:

- a research question;
- candidate segment hypotheses;
- inclusion and exclusion criteria;
- channel and capacity constraints;
- consent, privacy, incentive, and retention policies.

If the research question asks which feature people want, restate it around a past behavior or unresolved decision before planning recruitment.

## Workflow

1. Define the decision the research will inform.
2. Describe candidate segments by recent behavior, context, and role. Do not define a segment only by demographics.
3. Separate B2B buyer, user, approver, and administrator roles when they can experience different jobs.
4. State inclusion, exclusion, and sampling-bias risks.
5. Choose channels based on where the target behavior is observable.
6. Draft a short invitation that states purpose, time, incentive, data use, and voluntary participation.
7. Create a recruitment log with status, consent state, and no unnecessary personal data.
8. Stop at drafts. Ask for explicit approval before any email, direct message, form submission, or public post.

## Boundaries

- Let the model propose segments, channels, and wording.
- Use deterministic checks for required consent fields, duplicated candidates, private-data patterns, and `external_send=false`.
- Require a person to approve targets, incentives, channel rules, and every external send.
- Do not invent response-rate benchmarks or treat a participant count as universal evidence sufficiency.
- Do not scrape or enrich private contact information.

## Outputs

Create:

- `reach-plan.json`
- `contact-drafts.md`
- `recruitment-log.csv`

Read [references/output-contract.md](references/output-contract.md) before writing them.

## Stop conditions

Stop and report the missing decision when:

- consent or retention policy is absent;
- the segment cannot be connected to the research question;
- recruitment depends on restricted data;
- an invitation makes a product or payment promise;
- the user asks to send without naming targets and approving the final draft.

## Verification

Run:

```bash
python3 scripts/stg.py scan-privacy contact-drafts.md
```

Confirm that every external action remains a draft and that observed facts, assumptions, and recommendations are labeled separately.