Plain text vs HTML cold email deliverability test. 60 days, 3000 prospects, clear winner
plain_text_pete · 2026-06-13 · 1,210 views
Spent 60 days testing plain text cold email against HTML-formatted cold email with identical copy, identical list, and identical send timing. Here is what the data says.
Test setup. 3,000 prospects split 50/50. Same copy, same sequence timing. Same sending platform (Instantly) and same infrastructure (PuzzleInbox Google Workspace). The only variable: one group received plain text. The other group received the same email with a logo, branded footer, and minimal two-column HTML layout.
Results. Plain text: 3.8% reply rate, 4.1% spam placement on GlockApps. HTML: 2.1% reply rate, 11.3% spam placement. Plain text won on both metrics by a wide margin.
Why this happens. Email providers look at HTML emails differently from plain text. The code complexity, image tags, and tracking pixels embedded in HTML match the fingerprint of mass marketing automation. Google and Microsoft apply extra scrutiny. More emails land in Promotions or Spam before the prospect ever sees them.
There's a simpler explanation too. Plain text looks like it came from a real person. Because most professional email between humans is plain text. Your prospect's boss does not send them HTML newsletters. When your cold email looks like a marketing blast, it gets treated like one.
What I send now. Plain text only. No logo. No footer image. No unsubscribe link embedded in the body. Clean, minimal, looks like it came from a human inbox. The opt-out mechanism is a simple "Reply with STOP" line at the bottom. That's it.
The one situation where I add minimal formatting. Long-form follow-ups that include specific client results with multiple data points. A few bold headers help readability there. But that's email three or four in a sequence, after the prospect has already engaged. Never in a cold email.
Measure reply rate only. That's the number that reflects whether your email actually reached a human who wanted to respond. Plain text wins every test I've run on this over three years.