Not Getting Replies? What to A/B Test First
A prioritized order for A/B testing cold email, starting with the change that moves the numbers the most.
Product & Engineering · July 12, 2026 · 2 min read
When a cold email campaign underperforms, it is tempting to test everything at once. Change five variables in one test, though, and you learn nothing about which one mattered. The fix is to test one thing at a time, in the order most likely to move the number you care about.
Start with subject lines, not the body
Subject line changes affect your open rate, which caps everything downstream. A great email nobody opens cannot earn a reply. It is also the fastest, lowest-risk test: two subject lines, the same body, split the list, and see which gets opened more. Because it isolates a single, easily measured metric, it is the cleanest place to start.
Then the opening line
Once emails are being opened, the next bottleneck is usually whether the first sentence earns the rest of the read. People decide within a line or two whether to keep going. Testing your opening line on its own, with the same subject and the same rest of the email, tells you exactly whether your hook is working.
Then the call to action
How you ask for the next step matters more than most senders assume. A direct ask like "Can we get 15 minutes this week?" and a softer one like "Worth a quick chat, or should I check back later?" can produce very different reply rates depending on your audience and how warm the outreach is. Test this once the subject line and opening are already working, since the ask is what turns an engaged reader into a reply.
Then send time and length
Send time and day of week matter, usually less than the first three, and they are the easiest to test almost for free, since you are already sending on a schedule. Length is similar. Some audiences prefer a short, scannable email, while others want enough context to decide without a back-and-forth. Both are worth testing, just after the higher-leverage variables above.
Why the order matters
Testing in this order means each test tells you something real, and you are always working on the variable most likely to be your actual bottleneck. Testing send time before your subject line is even landing opens, for instance, optimizes something that does not matter yet when the real problem is upstream.
Testing is far easier when you are not starting from a blank page. A solid template gives you a baseline worth iterating on.
Resources in this guide
- Subject line, the single highest-leverage thing to test first
- Opening line, the first sentence of the body
- Call to action, a soft question versus a direct ask
- Send time and day of week
- Length, a short version versus a longer, more detailed one
Continue learning
Start from a proven template
Rather than test from a blank page, start with something that already works.