cold-outreach

Files (1)

• SKILL.md

SKILL.md (first lines)

# Cold Outreach

## When to Use
Activate when a founder needs to write cold emails or LinkedIn messages to prospects, potential customers, investors, or strategic contacts. Also use when the user says "nobody replies to my emails," "how do I reach out to X," "write me a cold email," or "help with outbound."

## Context Required
From `startup-context` or the user:
- **Target prospect** — Name, role, company, and why them specifically
- **Research signals** — Recent news (funding, launches, hires), LinkedIn activity, company growth data, or role/industry context
- **Sender positioning** — Who you are, what you offer, your unique credibility
- **Platform** — Email, LinkedIn, or both
- **Batch size** — Single prospect or multi-prospect campaign

Work with whatever the user provides. A strong research signal and clear value prop is enough to draft. Note what would strengthen the message but do not block on missing inputs.

## Workflow
1. **Gather context** — Read startup-context if available. Ask for missing info on prospect, value prop, and proof points.
2. **Research the prospect** — Conduct web searches for recent signals. The core principle: 10 minutes of research transforms a cold message into a warm one. Rank signals by strength:
   - **Tier 1 (strongest):** Recent news — funding rounds, product launches, key hires
   - **Tier 2:** LinkedIn activity — posts, comments, job changes
   - **Tier 3:** Company growth signals — hiring trends, tech stack changes
   - **Tier 4 (weakest):** Role/industry awareness only
3. **Assign personalization tier** — Based on research signals found:
   - **Tier 1 (custom):** Named signals across multiple research sources — fully personalized message
   - **Tier 2 (templated + personalized):** Company info and role context — template with personalized elements
   - **Tier 3 (volume template):** No signals found — use volume approach with strong value prop
4. **Select mode based on scope:**
   - **Quick:** Single connection request + follow-up for one prospect
   - **Standard:** Four-touch sequence for a prospect (default)
   - **Deep:** Multi-prospect system with A/B variant messages
5. **Draft the sequence** — Write messages following the structure and rules below.
6. **Self-critique pass** — Before delivering, validate that personalization connects to the problem. If you remove the personalized opening and the message still makes sense, the personalization is not working. Rewrite.

## Output Format
Deliver all of the following:
- **Connection request** (LinkedIn, max 300 characters) or **Subject line** (email, 2-4 words, lowercase)
- **Primary message** — the full outreach text (emails under 125 words, InMails under 500 characters)
- **Follow-up sequence** — with timing and a new angle per touch
- **Personalization notes** — what to customize per recipient if sending to multiple prospects
- **Tier label** — which personalization tier this message uses and why

## Frameworks & Best Practices

### The Core Principle
The word "cold" is the problem. Every message should feel like it comes from someone who understands the prospect's world. Research is what makes that possible.

### Message Structure
- **Connection request (LinkedIn):** Max 300 characters. Reference something specific. Never pitch in the request.
- **First message (24-48 hours after connection):** "Thanks for connecting" + bridge to a research signal + value statement + question. Keep it conversational.
- **Follow-up 1 (Day 7):** Introduce a new angle — different problem, proof point, or insight.
- **Follow-up 2 (Day 14):** Share something valuable (article, data, framework) with a soft reconnect.
- **Break-up (Day 21):** Professional close — "Closing the loop. If timing is ever right, I'm here."

### Writing Principles
- **Write like a peer, not a vendor.** Use contractions. If it sounds like marketing copy, rewrite it.
- **Every sentence must earn its place.** If it does not move toward a reply, cut it.
- **Lead with their world, not yours.** "You/your" should dominate over "I/we."
- **One ask, low friction.** Interest-based CTAs ("Worth exploring?") beat meeting requests.
- **Every message must reference a specific research signal** or explicitly default to Tier 3. This is a hard rule.

### Email Frameworks
- **Observation-Problem-Proof-Ask** — You noticed X, which usually means Y challenge. We helped Z with that. Interested?
- **Trigger-Insight-Ask** — Congrats on X. That usually creates Y challenge. We have helped similar companies. Curious?
- **Story-Bridge-Ask** — [Similar company] had [problem]. They [solved it this way]. Relevant to you?

### Subject Lines
- 2-4 words, lowercase, no punctuation tricks
- Should look like an internal email ("quick question," "re: [their company]")
- No product pitches, no urgency, no emojis

### What to Avoid
- Opening with "I hope this finds you well" or "My name is X and I work at Y"
- Jargon: "synergy," "leverage," "best-in-class," "leading provider"
- Feature dumps — one proof point beats ten features
- HTML formatting, images, or multiple links in cold emails
- Fake "Re:" or "Fwd:" subject lines
- Asking for 30-minute calls in first touch
- Sending identical templates with only the name swapped
- Pitching in a LinkedIn connection request

### Founder-Specific Advantages
- Founder-to-founder or founder-to-exec emails get 2-3x higher reply rates
- Lead with "I built this because..." — more compelling than "our company offers..."
- Offer what reps cannot: personal onboarding, product roadmap input, advisory relationships

## Related Skills
- `lead-scoring` — use to prioritize which prospects to reach out to first
- `sales-script` — use when the outreach lands a meeting and you need a discovery call or demo script

## Examples

**Example prompt:** "I need to reach out to VP Engineering at mid-market SaaS companies about our API monitoring tool. We reduced downtime by 73% for Acme Corp."

**Good email output (Standard mode, Tier 2):**
> Subject: api alerts
>
> Hi [Name],
>
> Saw your team just shipped the new payments integration — nice work. Launches like that usually surface a wave of edge-case API failures that are tough to catch with standard monitoring.
>
> We built a tool that catches those failures before customers notice. Acme Corp cut their API downtime by 73% in the first month.
>
> Worth a quick look?

**Good LinkedIn connection request:**
> Hi [Name] — saw the payments launch. We help engineering teams catch API failures before customers do. Would love to connect.

**Follow-up (Day 7, new angle):**
> Hi [Name], quick thought — after launches like yours, the #1 issue teams tell us about isn't downtime, it's the silent failures that slip through alerts. Happy to share what patterns we see across 50+ engineering teams if useful.

From mkurman/zorai, MIT.

More from mkurman/zorai