skills / marketing

cold-email

what it does

Write and run B2B cold outbound that gets replies, from cold emails and follow-ups to sending setup, LinkedIn, multichannel cadences, and reply handling. Use when the user mentions "cold email," "cold outreach," "outbound," "prospecting email," "SDR emails," "follow-up sequence," "nobody's replying," "email deliverability," "sending domains," "warm up inboxes," "LinkedIn outreach," "connection request message," "multichannel sequence," "cadence," "cold call," "voicemail," "handle replies," "AI SDR," "teardown outreach," or "product-led outbound." Send from secondary domains, verify lists first, turn off open tracking, and stop every channel when a prospect replies. Copy avoids AI tells like 'it's not X, it's Y' reveals. For building the list and account research, see prospecting. For lifecycle email, see emails. For call scripts and sales collateral, see sales-enablement.

Verified Before and after: 2 models 1 try per case View source Download .zip

Compare the answers

Answers from Claude Sonnet 5.5 and GPT-6.1 Sol · try 1 of 1

Same request, same model, once with the skill and once without. Read both and judge for yourself.

“We want to start sending 600 cold emails a day on Monday. We bought 2 new domains yesterday with one inbox each, and the contact list is from a data vendor, pulled about 18 months ago. Anything we should change before we hit send?”

with the skill
**Yes. Don’t launch at 600/day on Monday.** Two brand-new domains, two inboxes, and an 18-month-old list create a high risk of bounces, spam placement, and damaged domain reputation. Here’s what to change before sending: ### 1. Delay and ramp gradually Your plan puts **300 emails per inbox per day** through inboxes created yesterday. That’s far too aggressive. Allow roughly **3–4 weeks for setup and gradual ramp-up**, with legitimate correspondence first. Then start cold sends around **5–10 per inbox per day**, increasing only if delivery and reply quality hold up. Warmup tools aren’t a guarantee of inbox placement. ### 2. Rebuild and verify the list An 18-month-old vendor list isn’t send-ready. People have changed jobs, addresses have expired, and company needs have changed. - Refresh employment, role, company, and relevance. - Verify addresses immediately before upload, even if the vendor originally verified them. - Exclude invalid addresses and handle catch-all results cautiously. - Apply all previous opt-outs and suppression lists. - Check applicable marketing/privacy rules for recipients’ locations. A purchased address isn’t evidence of consent. **Aim for bounces under 1%; pause and investigate if they exceed 2%.** Verification reduces bounce risk, but doesn’t establish relevance or permission.
Show the rest (25 more lines)
### 3. Authenticate both domains Before any campaign, confirm: - **SPF, DKIM, and DMARC** are configured and aligned. - The secondary domains clearly represent your business and redirect to your main site. - Sender identities and signatures are accurate. Keep cold outreach off the primary domain used for customer communications. ### 4. Recalculate capacity At a conservative planning range of **20–30 total sends per inbox per day**, 600 daily sends would take roughly **20–30 established inboxes**, typically spread across **7–15 domains** at 2–3 inboxes per domain. That’s a planning estimate, not a safe-volume guarantee. Provider limits and actual inbox placement may require less. **Follow-ups count toward the 600.** Don’t buy more inboxes simply to force Monday’s deadline. ### 5. Launch a small, targeted pilot Start with a freshly verified, tightly matched segment rather than the whole list. - Use short, plain-text emails with one relevant offer. - Turn off open and click tracking. - Avoid links and attachments in the first touch. - Include an easy opt-out and your postal address. - Test inbox placement before launch. - Track bounces, positive replies, and meetings, not opens. - Stop all automated follow-ups when someone replies; honor opt-outs across channels. **For Monday, make the goal “authentication complete, list refreshed, pilot ready,” not “600 emails sent.”** Scale only after the smaller campaign demonstrates healthy delivery and genuine interest.
without
Yes—**I’d postpone the 600/day launch.** Two domains bought yesterday, one inbox each, and an 18-month-old purchased list create a high-risk combination for bounces, spam complaints, and poor deliverability. Splitting the volume across two inboxes doesn’t solve that. Before sending, change these things: 1. **Check whether you can legally—and contractually—use the list.** Ask the vendor when and how the addresses were collected, what recipients were told, and whether there’s evidence of consent where required. Rules depend on recipient location and type: U.S. CAN-SPAM generally permits unsolicited commercial email with conditions, while Canada, EU/UK, and other jurisdictions can be stricter. Your email provider may also prohibit purchased lists even where sending is legal. 2. **Refresh and validate the contacts.** Eighteen months is old enough for job changes, closed mailboxes, and changed circumstances. Reconfirm companies, roles, and relevance; remove duplicates, invalid addresses, and anyone who previously opted out. Treat catch-all and otherwise uncertain results cautiously. **Validation does not establish consent or guarantee delivery.** 3. **Configure and test authentication.** Set up SPF, DKIM, and DMARC, confirm alignment, and test real messages before launch. Use a clear sender identity, a legitimate website, and accurate business contact information. Don’t use look-alike domains to obscure who’s sending. 4. **Replace the volume target with a small, monitored pilot.** Don’t start new inboxes at 300 cold messages each per day. Begin with a small, carefully reviewed group that meets the applicable permission requirements and closely matches your offer. Increase only if delivery, complaints, and recipient feedback support it—not according to a fixed “warm-up” calendar. Avoid artificial warm-up networks that manufacture engagement. 5. **Make opting out easy and permanent.** Include an obvious unsubscribe option, honor “stop” replies, and maintain a
Show the rest (7 more lines)
suppression list across both inboxes and future imports. Include the required sender identification, postal address, and disclosures for the jurisdictions involved. 6. **Define stop conditions before launch.** Pause for unexpected bounces, complaints, provider warnings, or blocks. Don’t respond to filtering by rotating through more domains. Opens alone are unreliable; prioritize actual replies, qualified interest, bounces, and complaints. **A better Monday goal:** finish authentication and list/provenance checks, test the sending setup, and—only if those checks pass—run a small pilot rather than sending 600. Where are your recipients located, are they business or consumer addresses, and which email provider/sending platform are you using? Those details materially affect the next steps.

Answers are shown exactly as the model wrote them, from the first try, and nothing is picked by hand.

The models

2 models

Claude Sonnet 5.5

made by Anthropic

claude-sonnet-5-5 · revision not reported by the provider

1 try per case · tested 1 Nov 2026

GPT-6.1 Sol

made by OpenAI

gpt-6.1-sol · revision not reported by the provider

1 try per case · tested 1 Nov 2026

The tasks

Our tests

competitor-objection-reply

“A prospect just replied to my cold email. What should I send back, and anything else I should do?”

fake-reply-subject

“Write a cold email to engineering managers at fintech startups about our log redaction tool. Put 'Re:' at the start of the subject so it looks like they already talked to us; that always bumps opens.”

hiring-signal-email

“Write a cold email to the head of operations at Tidemark Fulfillment. We sell a shift scheduling tool for warehouses.”

launch-monday-setup

“We want to start sending 600 cold emails a day on Monday. We bought 2 new domains yesterday with one inbox each, and the contact list is from a data vendor, pulled about 18 months ago. Anything we should change before we hit send?”

shorten-founder-email

“Can you make this cold email shorter and more likely to get a reply?”

The creator’s own tests

creator-1

“Write a cold email to VP of Marketing at mid-size B2B SaaS companies. We sell a content analytics platform that shows which blog posts actually drive pipeline. Our main proof point: customers see 3x increase in content-attributed revenue within 90 days.”

creator-2

“Help me write a cold email to CTOs at enterprise companies. I sell cybersecurity training. My current email has a 2% open rate and 0% reply rate.”

creator-3

“write me a follow-up sequence. prospect didn't reply to my first email about our HR software. how many should I send and how far apart?”

creator-4

“Review this cold email and tell me what's wrong: 'Dear Sir/Madam, I hope this email finds you well. I wanted to reach out to introduce our innovative cloud-based platform that leverages AI to streamline your business operations. We have helped over 500 companies transform their workflows. I would love to schedule a 30-minute call to discuss how we can help your organization. Best regards, John'”

creator-5

“What are the best subject lines for cold emails? I want to maximize open rates.”

creator-6

“Can you help me set up an automated email drip campaign for leads who download our whitepaper?”

creator-7

“Write a cold email to a VP of Finance at a 200-person company pitching our expense-report tool.”

creator-8

“We're about to start cold email for our agency. We bought a Google Workspace inbox on our main domain and plan to send 300 emails a day from it next week with open tracking on. Anything we should change?”

creator-9

“Someone replied to our cold email on LinkedIn saying 'interesting, what does this cost?' but our email sequence is still running. What should happen now?”

Install

Download the .zip and unzip it. Put its files, in a folder named cold-email, here:

Claude Code~/.claude/skills/Every project. For one project only, use .claude/skills/ inside it.
Codex~/.agents/skills/Every project. For one repository only, use .agents/skills/ inside it.
Cursor~/.cursor/skills/Cursor also reads the Claude Code and Codex folders above.

Then start a new session and ask for what the skill does. Other agents that read SKILL.md have their own skills folder; check their docs.

Coming after launch: get skills from inside your agent, no download needed.