Personalized Cold Outreach

personalized-cold-outreach

Purpose — Generate 1-to-1 personalized outreach messages at batch scale. Each message quotes the recipient's own words back to them, references a specific pain they've expressed, and offers a concrete next step.

Triggers — "personalize cold emails for ", "write outreach for this CSV", "1-to-1 message for each row in "

Why this beats a mailmerge — Standard mailmerge replaces {{name}}. This skill generates 3–5 personalized sentences per recipient referencing their actual public pain statements. Reply rates are typically 3–10× standard cold email.

---

Inputs

• Seed CSV from [competitive-intel-mining](../competitive-intel-mining) — must include handle, source_url, exact_quote, pain_category
• Your pitch — 2–3 sentences describing what you'd show the recipient if they replied (not your full product tour)
• Message channel — email / X DM / GitHub issue / LinkedIn
• CTA — what action you want them to take (reply / book call / click pre-filled issue / watch 3-min Loom)

Outputs

• A CSV enriched with columns: subject, opening, pain_mirror, pitch_line, cta, pre_filled_url (if applicable), send_batch
• A review list of the 5 best and 5 worst generated messages for manual tuning before batch send

Steps

1. Read row + source URL — For each recipient, pull the exact_quote and (optionally) re-open the source URL for 1–2 adjacent sentences of context.
2. Draft opening — 1 sentence referring to the specific place you found them ("saw your Paperclip #447 thread"). Never generic ("I noticed you").
3. Draft pain mirror — 1–2 sentences quoting their own words and naming the underlying pain ("you said 'agent panics when sandbox rejects bash' — that's the class of bug that comes from X").
4. Draft pitch line — 1 sentence only. What you'd show them if they replied. Concrete, not abstract.
5. Draft CTA — 1 specific next step with near-zero friction. Examples:

• "Want a 3-min Loom tailored to your case? Reply y/n."
• "Open a pre-filled issue with what breaks today — "
• "Reply with a screenshot of your current agent config"

1. Sign-off — First name only. No title, no links in signature.
2. Quality check each message against the rules below. Regenerate if any fail.
3. Batch output — Group into send batches of ≤10 per day per channel to avoid spam filters / rate limits.

Quality bar (per message)

• [ ] ≤120 words total
• [ ] Zero generic phrases ("I hope this finds you well", "I came across your profile")
• [ ] Exact quote from their words appears verbatim once
• [ ] Subject line is

Hey {{name}},

Saw your {{source}} where you wrote: "{{exact_quote}}"

That's the class of pain that happens because {{one-line technical/strategic take}}.

I've been working on {{one-line pitch}} — here's what I'd show you if it helps: {{prefilledurl}} (30 seconds, no signup)

Worth a look?

—{{yourfirstname}}

## Anti-patterns

- **Don't use "I think" or "I believe"** — they read as low-confidence. State the observation.
- **Don't close with "let me know if you're interested"** — it's a weak CTA. Use a specific verb.
- **Don't send on Monday morning** — inbox is overloaded. Tuesday–Thursday 10am recipient local time performs best.
- **Don't A/B 10 variants** — pick one good message, iterate after 20 replies.

## Related

- Skill: [`competitive-intel-mining`](../competitive-intel-mining) — source of the input CSV
- Skill: [`landing-to-issue-funnel`](../landing-to-issue-funnel) — for the `pre_filled_url` pattern
- Template: [`templates/cold-email/`](../../templates/cold-email) — raw message templates
- Playbook: [`01-cold-email-campaign`](../../playbooks/01-cold-email-campaign.md)

## Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

- **Author:** [Celina-create](https://github.com/Celina-create)
- **Source:** [Celina-create/X-Studio](https://github.com/Celina-create/X-Studio)
- **License:** MIT

Install and usage instructions live in the source repository linked above.