Cold email still works. But most founders do it wrong.
They write decent emails, hit send, and watch them vanish into spam folders. The average reply rate? Just 8.5%. Meanwhile, founders using AI sales automation are hitting 15-20% reply rates with half the effort.
The difference isn't talent. It's system.
This guide shows you how to use AI for cold email the right way, automated, personalized, and actually landing in inboxes.
Why Most Cold Emails Fail
17% of cold emails never reach the inbox. They're filtered as spam before your prospect even has a chance to ignore them.
Google and Microsoft got aggressive in 2024. New sender requirements, stricter authentication, and AI-powered spam filters mean the old "spray and pray" approach is dead.
The emails that do land? Most sound like they were written by a robot (ironically, the bad kind). Generic templates, zero personalization, desperate subject lines like "Quick question" sent to 10,000 people.
Cold email automation done right fixes both problems: deliverability AND personalization at scale.
Step 1: Warm Up Your Domain First
Skip this step and nothing else matters. Your beautifully crafted emails will rot in spam.
Email warmup is exactly what it sounds like: gradually building your sender reputation before going full volume. New domains have zero trust with Gmail and Outlook. You need to earn it.
The basics:
- Set up SPF, DKIM, and DMARC authentication (non-negotiable)
- Start with 10-20 emails per day
- Increase by 10-20% weekly
- Use a dedicated warmup tool that simulates real engagement, opens, replies, and removing emails from spam
This takes 2-4 weeks. Yes, it's slow. But sending 500 emails from a cold domain is like showing up to a party and immediately asking everyone for money. You'll get blocked.
AI email warmup tools like Smartlead handle this automatically, running warmup campaigns in the background while you focus on your actual business.
Step 2: Write Emails That Sound Human
Personalized emails get 46% higher open rates. But "Hi {First_Name}" isn't personalization. It's a mail merge from 2005.
Real AI cold email personalization means referencing something specific: a recent funding round, a job posting that signals growth, a podcast appearance, or a LinkedIn post they wrote. Something that proves you actually know who they are.
Here's a template that works:
Subject: {Company}'s growth + quick idea
Hi {First_Name},
Saw you're hiring 3 new SDRs, congrats on the growth.
Most founders I talk to are drowning in outbound before those hires even start. We help teams automate the repetitive stuff so your new reps can focus on conversations, not copy-paste.
Worth a quick chat?
{Your name}
Notice what's missing: no "I hope this email finds you well," no walls of text about features, no desperate "just circling back." AI first line generators can help create these personalized openers at scale, but the rest of your email still needs to be tight.
Step 3: Nail Your Subject Lines
47% of recipients decide to open based on subject line alone. And the best B2B cold email subject lines are shorter than you think.
What works:
- 2-4 words
- Questions
- Lowercase
- Specificity
- Personalization (company name or reference)
What kills opens:
- ALL CAPS
- "Quick question"
- "Following up"
- "Exciting opportunity"
- Anything that screams "I'm selling something"
Examples that convert:
- {Company}'s outbound
- idea for {First_Name}
- saw your SDR post
- re: scaling sales
Step 4: Build a Follow-Up Sequence That Converts
Here's the stat that should change how you think about cold email: 70% of responses come from follow-up emails 2-4. Not the first email. The follow-ups.
Most founders send one email, get no response, and give up. They're leaving 70% of their potential replies on the table.
A simple cold email follow up sequence:
- Day 1: Initial email
- Day 3: Quick bump ("Did this land?")
- Day 7: Add new value (case study, insight)
- Day 14: Breakup email ("Should I close your file?")
Automated email outreach tools handle this timing automatically. You set the sequence once, and the system sends follow-ups based on whether (and how) the prospect engaged.
Pro tip: Keep follow-ups short. Your second email shouldn't be another pitch, it should be a gentle nudge that adds one small piece of value or asks a simple question.
The AI-Powered Cold Email Stack
Here's what a modern cold email at scale setup looks like:
- Lead data: Apollo, Clay, or similar for finding prospects that match your ICP
- Email infrastructure: Multiple warmed domains rotating to protect deliverability
- AI personalization: Tools that research prospects and write custom first lines
- Sequence automation: Smartlead or similar for sending, tracking, and follow-ups
This is where an AI SDR starts to make sense. Instead of hiring a junior rep to do manual research and send-copy-paste outreach, you build a system that handles the repetitive work while you focus on the conversations that actually close deals.
The founders winning at outbound in 2025 aren't sending more emails. They're sending better emails, to better prospects, with better timing, all automated.
FAQ
How many cold emails can I send per day?
50-100 per warmed inbox is the safe zone. Go higher and you risk spam flags. Scale by adding more inboxes, not more volume per inbox.
What's a good cold email reply rate?
5-10% is solid for most B2B campaigns. With tight targeting and real personalization, 10-15% is achievable. If you're below 1%, something's broken, usually deliverability or relevance.
Ready to automate your outbound? Starnus builds AI sales agents that handle warmup, personalization, and sequences, so you can focus on closing.