I ran 44 cold email subject line split tests over six months. Here is what the data actually shows about what gets replies in 2026
abtester_vince · 2026-08-15 · 440 views
Not open rates. I do not care about open rates. Apple MPP and security bots made that metric useless two years ago. I measured reply rate. 44 split tests over six months across four industries and seven ICPs. Here is what actually emerged from the data.
Pattern 1: Lowercase subject lines outperform title case by an average of 0.6 percentage points in reply rate. Not huge but consistent. 11 of my 14 lowercase-vs-title-case tests showed the lowercase variant ahead. The hypothesis is that lowercase reads more like a colleague email and less like a mass send.
Pattern 2: Subject lines under 5 words beat longer ones in 8 of 11 tests. My highest performing subject line across the entire test set was three words. The longer subject lines that beat shorter ones all shared one trait: they included a specific number or a company name. Specificity earns length. Vague phrases do not.
Pattern 3: Questions underperform statements. I expected questions to win because they create curiosity. In my data, direct statements like "idea for [company]" outperformed question formats like "question about [company]" in 7 of 10 matched tests. I think questions feel slightly more like a template because so many senders use them.
Pattern 4: The prospect's company name in the subject line lifts reply rate an average of 0.9 points versus no company name. Most consistent finding in the whole data set. 17 of 20 tests confirmed this. Personalization at the subject line level works even for people who know what a cold email looks like.
Pattern 5: Re: subject lines are dead. I tested five re: formats. Four lost to straightforward alternatives. Prospects know the fake reply trick. It reads as manipulative now and kills credibility before the first sentence.
All campaigns ran through PuzzleInbox Google Workspace inboxes in Smartlead. Deliverability was stable throughout so the variation is attributable to subject line, not infrastructure noise.
The single change with the most consistent impact: drop re: prefixes, shorten to under 5 words, and put the company name in. That combination alone lifted reply rate by 1.4 percentage points average in my data set.