Testing
Cold Email A/B Testing: What to Test First
Most teams test the wrong variable first. Here is the order that actually moves reply rates.
Why Most A/B Tests Waste Your Time
Teams test subject lines first because they are easy. But subject lines move open rates, not reply rates — and in cold email, replies are the only metric that pays rent.
The right order targets the variables with the biggest leverage first: deliverability, then offer clarity, then personalization depth, then length, then subject lines.
1. Test Your Sending Infrastructure First
Before any copy test, check the foundation: SPF, DKIM, DMARC, dedicated sending domain, and warmup state.
If your emails land in promotions or spam, no copy test matters. Fix deliverability before touching a single word.
2. Test Offer Clarity (The One-Liner)
The single highest-leverage copy variable: can a prospect understand what you do and why it matters in under 5 seconds?
Test two versions of your first paragraph:
- Version A: feature-led ("We provide AI-powered cold email software with deliverability scoring")
- Version B: outcome-led ("Write a cold email that gets replies — in 20 seconds, without sounding like a bot")
3. Test Personalization Depth
Real personalization (a company-specific detail in line one) beats merge-tag personalization (first name only). Test one researched detail versus none.
The detail does not need to be impressive. A pricing page observation, a recent hire, a new integration — anything that proves you looked.
4. Test Length
Shorter emails win on mobile and get read faster. Test a 60-word version against a 150-word version.
Rule of thumb: cut every sentence that does not move the reader toward a reply. If a line does not create curiosity or clarity, delete it.
5. Test Subject Lines Last
Subject lines matter for opens, but with a good sender reputation and a clean domain, opens are not the bottleneck. Test subject lines only after the above four are stable.
The Minimum Viable Test Setup
- One variable at a time (changing two things means you cannot attribute the result)
- 50-100 sends per variation for statistical signal
- Same time window, same segment
- Track replies, not opens
What Good Looks Like
After fixing deliverability and testing offer clarity, most teams see reply rates move from 1-2 percent to 3-5 percent. Personalization depth adds another 1-2 points. Length and subject lines are the finishing touches.