Linkedin opener

LinkedIn Opener

Generate 5 first-line variants for a LinkedIn message, each under 25 words, with zero AI tells. Load voice.md and icp.md before drafting.

When this skill triggers

• User asks for an opener, first line, or hook
• User asks to draft a LinkedIn DM or InMail
• User pastes signal data and asks "how do I open this?"
• User asks for a "cold DM" or "outreach message" on LinkedIn

Inputs required

1. Signal that fired — job change, LinkedIn post engagement, funding, hiring, etc.
2. Target person — name, title, company
3. Personalization data — one of: their recent post, mutual connection, company news, podcast appearance, team growth signal, product announcement

If personalization data is missing, ask the user for it before drafting. Never draft a "cold" opener with no personalization hook — that's what makes cold email dead.

The 5-variant structure

For every request, produce five distinct openers using different angles. Number them 1–5. Mark the strongest with ★.

The five angles to rotate:

1. Direct observation — reference the specific signal or content, phrased like a human noticed something (never using the word "noticed")
2. Curiosity question — open with a question tied to their situation, not a pitch
3. Third-party credibility — reference a mutual context (customer, community, event, publication)
4. Contrarian angle — challenge an assumption gently, in a way that opens dialogue
5. Utility offer — offer something specific and free (an idea, a template, a comparison), tied to their signal

Rules

Never use these phrases:

• "I noticed"
• "I saw"
• "I stumbled on"
• "Reaching out"
• "Hope this finds you well"
• "Just wanted to"
• "Quick question"
• "Circling back"
• "Bumping this"
• "Wanted to introduce myself"
• "Let me know if you have any questions"

Never do these things:

• Compliment their profile or title ("Impressive background")
• Reference the platform ("Saw you on LinkedIn")
• Reference the fact that you're reaching out cold
• Mention your product in the opener (that's the second sentence, not the first)
• Use exclamation points (one, at most, and only for genuine warmth)
• Ask for their time in the opener ("Do you have 15 minutes")
• Use hedging language ("I might be reaching out at a bad time")

Every opener must:

• Be under 25 words
• Reference the specific signal data provided (not generic)
• Pass the "would I send this to a friend?" test — if it sounds like a template, rewrite
• Match the voice from voice.md

Output format

Return in this structure:

LINKEDIN OPENERS — [Person Name] @ [Company]

Signal: [signal type from input]
Personalization hook: [what you're referencing]

1. [Direct observation variant]
   Angle: Direct

2. [Curiosity question variant]
   Angle: Question

3. [Third-party credibility variant]
   Angle: Credibility

★ 4. [Contrarian angle variant]
   Angle: Contrarian
   Why strongest: [1-sentence rationale]

5. [Utility offer variant]
   Angle: Utility

Always mark exactly one variant with ★ and explain in one sentence why it's strongest for this specific person + signal combination.

Examples

Example 1 — LinkedIn post as personalization

Input:

• Person: Sarah Chen, VP of Sales at Acme (B2B SaaS, 300 emp)
• Signal: She posted yesterday about "the pain of managing 4 outbound tools that don't talk to each other"

Output:

LINKEDIN OPENERS — Sarah Chen @ Acme

Signal: LinkedIn post about outbound tool sprawl
Personalization hook: Her exact post about "4 outbound tools that don't talk to each other"

1. Your post about 4 outbound tools not talking to each other hit a nerve — we ran a Frankenstein stack for 18 months before we consolidated.
   Angle: Direct

2. Curious — when you say the tools don't talk, is it the enrichment-to-send handoff or the reply data going nowhere?
   Angle: Question

3. Same problem hit us at [comparable company] last year — happy to share how we consolidated to 2 tools without losing coverage.
   Angle: Credibility

★ 4. Most consolidation stories focus on the CFO win. The one nobody talks about is that your SDR team gets back 6 hours a week.
   Angle: Contrarian
   Why strongest: Reframes the problem from cost-saving to time-back, which is what a VP of Sales actually cares about. Also — she'll want to know the "hours a week" claim.

5. If it helps, I mapped the 3 handoff points where outbound stacks usually leak — happy to send it over if you're doing this exercise.
   Angle: Utility

Example 2 — champion job change

Input:

• Person: Marcus Reid, new Head of Growth at Vector.io (Series B, 80 emp)
• Signal: Marcus was previously at Beta (a former customer of ours)
• Notes: Started at Vector 2 weeks ago

Output:

LINKEDIN OPENERS — Marcus Reid @ Vector.io

Signal: Champion job change — Marcus was at Beta (former customer), now at Vector
Personalization hook: The Beta relationship + his new role

1. Congrats on the Vector move — hope the first 2 weeks have been survivable.
   Angle: Direct

2. Curious what Vector's outbound looks like now compared to what we built together at Beta.
   Angle: Question

★ 3. Beta's outbound engine held up through the Q4 push last year — happy to share what to keep and what to leave behind at Vector.
   Angle: Credibility
   Why strongest: Grounds the outreach in a shared, positive history. Signals continuity of the relationship without pitching.

4. Every new head of growth wants to rebuild the stack in month one. The interesting ones wait 60 days and audit first.
   Angle: Contrarian

5. Wrote up a 90-day playbook for growth leaders starting fresh — want me to send it? Zero pitch, just the framework.
   Angle: Utility

Example 3 — funding announcement

Input:

• Person: Priya Sharma, CEO at Nova (just raised $12M Series A)
• Signal: Funding announcement 3 days ago in B2B SaaS
• Notes: 25-person company, marketing/martech vertical

Output:

LINKEDIN OPENERS — Priya Sharma @ Nova

Signal: Series A funding announced 3 days ago
Personalization hook: The $12M raise + Nova's stage

1. Congrats on the $12M — assuming your inbox has been chaos.
   Angle: Direct

2. Curious what Nova's rebuild looks like in the first 90 days — outbound is usually the first thing that changes.
   Angle: Question

3. Watched a few Series A martech teams rebuild in the last 6 months — happy to share what worked and what wasted a quarter.
   Angle: Credibility

4. Series A pitches are almost always about "scaling the machine." The teams that pull it off spend the first month tearing the old one down.
   Angle: Contrarian

★ 5. If it helps, I have a Series A GTM audit template — the one I wish I'd had at my last company. Want me to send it?
   Angle: Utility
   Why strongest: Priya's inbox is a firehose right now. Every message is "congrats + let me sell you." Leading with utility (specifically a template) cuts through and asks for nothing.

Common failure modes

Failure: opener sounds "polished" instead of human. Fix: Rewrite in a shorter, more clipped rhythm. Read it out loud — if you'd never say it in person, it needs to change.

Failure: opener references the signal too directly and sounds surveillance-y. Fix: "Saw you visited our pricing page 3 times" → surveillance. "Nice to see [category] hitting the radar for you" → same signal, human framing.

Failure: opener is generic despite having personalization data. Fix: The specific detail must appear in the opener, not just be referenced. If the data is "she posted about tool sprawl," the opener needs the word "tool sprawl" (or a close paraphrase). If it doesn't, the personalization was decorative not functional.

Failure: all 5 variants sound like the same person wrote them slightly differently. Fix: The 5 variants need to be 5 DIFFERENT angles. If two feel similar, replace one.

Failure: opener pitches the product instead of setting up a conversation. Fix: The opener's job is to earn the second sentence. If your product name appears in the first line, you've collapsed the funnel.

Related skills

• linkedin-connection-note — if they're not connected yet and you need a note under 300 chars
• follow-up-sequence — for the second and third touch if no reply
• objection-response — for handling the reply when it comes