Cold email still works in B2B if you can solve the two problems that kill most campaigns: poor personalization and inconsistent follow-up. I've built automated cold email workflows with Claude Code that consistently book 15–25 qualified meetings per month for clients across software, consulting, and agency verticals. The key is treating it as a system, not a one-off send.
Here's how I automate B2B cold email outreach with Claude Code — from lead scraping to personalized copy to follow-up sequences — without burning sender reputation or sounding like a robot.
The Foundation: Lead Scraping and Enrichment
Every cold email campaign starts with a target list. Most people build these manually or pay for a database that's 40% outdated. I use Claude Code to scrape and enrich prospect lists from LinkedIn, company websites, and industry directories — all tied to specific ideal customer profile (ICP) criteria.
The workflow looks like this:
- Define the ICP: company size, industry, location, tech stack signals, job titles
- Use Claude Code web scraping to pull LinkedIn search results or company directory listings
- Enrich each record with secondary data: recent company news, hiring signals, tech stack from BuiltWith or similar sources
- Validate email addresses using a catch-all checker to avoid hard bounces
The result is a CSV with 200–500 highly targeted leads, each row containing the prospect's name, title, company, email, LinkedIn URL, and 2–3 data points that can be used for personalization. This entire process takes about 90 minutes — compared to 8+ hours if done manually.
Why Enrichment Matters for Reply Rates
The difference between a 2% reply rate and a 15% reply rate usually comes down to relevance. Generic outreach gets ignored. Outreach that references a specific trigger event — a recent funding round, a new product launch, a job posting for a role you can help fill — gets opened and replied to.
Claude Code makes it possible to inject those personalization tokens at scale. For every prospect, I pull:
- Recent LinkedIn posts or company blog updates
- Job openings that suggest a pain point (e.g., "hiring a CRM admin" = they're struggling with their current system)
- Technology adoption signals (e.g., they just started using HubSpot, which means they might need implementation help)
This data gets stored in the CSV and feeds directly into the email template engine.
Email Personalization at Scale with Claude Code
Once the list is enriched, the next step is writing emails that don't sound mass-produced. The pattern I use: a templated structure with dynamic personalization fields that Claude Code fills in based on the enrichment data.
Here's a real example from a campaign I ran for a SaaS client targeting VP of Sales roles:
Subject: {{firstName}}, quick thought on {{companyName}}'s SDR hiring
Hi {{firstName}},
Saw you're hiring SDRs at {{companyName}} — congrats on the growth. Quick question: are you planning to build out email sequences in-house, or is that something you'd consider automating?
We help B2B SaaS teams set up outbound email systems that book 20+ meetings/month without adding headcount. Happy to share a quick breakdown if it's relevant.
Worth 15 min?
The variables — firstName, companyName, hiring signal — all come from the scraped data. Claude Code reads the CSV, generates a unique email for each row, and outputs them into the sending tool (I typically use a combination of SMTP + a warmup service to protect sender reputation).
The key constraint: every email must pass a "could this have been written by a human?" test. If the personalization feels forced or irrelevant, it gets flagged and rewritten. This is where the Claude Code advantage over ChatGPT shows up — the extended context window allows me to feed it 50+ examples of good vs. bad personalization, and it learns the pattern.
Automated Follow-Up Sequences
Most cold email campaigns fail because of weak follow-up. One email gets a 5–8% open rate. A well-sequenced follow-up series can push total campaign engagement to 30–40%.
I build follow-up sequences with conditional logic based on recipient behavior:
- No open after 3 days → Send a subject line refresh with a different angle
- Opened but no reply after 5 days → Send a value-add follow-up (case study, tool, insight)
- Replied with "not interested" → Tag as closed-lost, remove from sequence
- Replied with interest → Trigger a Calendly link or meeting booking automation
Claude Code handles the logic layer. I define the rules in a simple YAML config file, and it generates the sequence, schedules the sends, and logs every interaction. For a typical 5-touch sequence, this means 4 automated follow-ups per prospect, each one contextually aware of what happened before.
The result: reply rates go from 3–5% on the initial send to 12–18% across the full sequence. That's the difference between 6 replies and 36 replies from a 200-contact campaign.
Protecting Sender Reputation
Cold email automation fails hard if you ignore deliverability. If your domain gets flagged as spam, your emails stop landing in inboxes — no matter how good the copy is.
The safeguards I build into every Claude Code email workflow:
- Email warmup — New sending addresses start at 10 emails/day and ramp up over 2–3 weeks
- SPF, DKIM, DMARC — Properly configured DNS records to authenticate outbound mail
- Bounce monitoring — Hard bounces get flagged immediately; if bounce rate exceeds 3%, the campaign pauses
- Engagement tracking — Low open rates trigger an alert to review subject lines or targeting
Claude Code monitors these metrics in real time and adjusts sending volume automatically. If deliverability starts to drop, the system throttles back. This is especially important for clients sending 500+ emails per week — one bad batch can tank a domain's reputation for months.
Integration with CRM and Meeting Booking
The final piece is connecting cold email responses to your sales process. I've built Claude Code CRM integrations that automatically create a new lead record when someone replies, tag it with the campaign source, and assign it to the right AE.
For meeting bookings, I tie the follow-up sequence to Calendly or a similar tool. When a prospect replies with interest, Claude Code detects the intent (using a simple keyword match or sentiment analysis) and sends a meeting link. If they book, it logs the conversion and stops the sequence. If they don't book within 48 hours, it sends a gentle nudge.
This end-to-end automation means the only human touchpoint is the actual sales call. Everything before that — prospecting, outreach, follow-up, booking — runs on its own.
Real Results from a Recent Campaign
To make this concrete: I ran a cold email campaign for a Vancouver-based marketing agency targeting e-commerce brands with $2M–$10M in revenue. The offer was a free CRO audit.
The numbers:
- 450 prospects scraped and enriched over one weekend
- 5-email sequence sent over 14 days
- 62 total replies (13.8% reply rate)
- 18 booked calls (29% of replies converted)
- 4 closed deals within 60 days
Total time invested: about 6 hours to set up the workflow, then 30 minutes per week to monitor performance and tweak copy. Compare that to a traditional SDR doing this manually — easily 20+ hours per week for the same volume.
What You Need to Get Started
If you want to replicate this for your own outbound motion, here's the stack:
- Claude Code for scraping, enrichment, and email generation
- A dedicated sending domain (not your primary company domain)
- SMTP relay service or ESP that supports API sending (I use Postmark or SendGrid)
- Email warmup tool (Mailshake, Lemlist, or a custom script)
- CRM with API access for logging responses (HubSpot, Pipedrive, or a custom-built CRM)
The entire setup can be done in 3–5 days if you know what you're doing. If you're starting from scratch, expect about two weeks to get it dialed in.
Common Mistakes to Avoid
I've seen several patterns that kill cold email automation projects before they start:
- Skipping the warmup period — Sending 500 emails on day one from a fresh domain guarantees you'll land in spam
- Using the same subject line for every prospect — Even with good personalization in the body, a generic subject line tanks open rates
- Not testing deliverability before scaling — Send test emails to multiple providers (Gmail, Outlook, Yahoo) and check where they land
- Over-automating the reply handling — Automated replies to interested prospects feel robotic; let a human take over once someone engages
The other big one: trying to automate before you have a proven manual process. If your cold emails don't work when you send them one by one, automation won't fix that. Get to a 10%+ reply rate manually, then scale it with Claude Code.
Next Steps
Cold email automation with Claude Code is one of those rare workflows where the upfront investment pays for itself within a month. If you're spending 10+ hours a week on manual outreach, or if you're paying an SDR $4K–$6K/month to send emails, this is worth building.
For more on setting up automated workflows, check out my guides on LinkedIn outreach automation and sales email sequences. And if you want to see how this could work for your specific use case, book a call — I'm happy to walk through the setup live.