The discipline of the first 90 characters. What works in 2026, what doesn't, and the rubric we score against.
Subject lines and preheaders are 90 characters of inbox real estate that determine whether the rest of your email exists for the reader. Most teams over-invest in copy below the fold and under-invest in the 90 characters above it.
This guide is the discipline of those 90 characters.
For email that reaches the inbox, the subject line and preheader determine whether the email is opened. The body's quality only matters for the readers who opened. So the math is:
Total conversions = (Sends × Inbox rate × Open rate × Click rate × Conversion rate)
Open rate is one of five terms — and the only one entirely controlled by the first 90 characters. A 2-percentage-point lift on opens is a meaningful lift on total conversions; the same 2-point lift on click rate matters too, but it's downstream of opens having happened.
For programs where deliverability is solid and the body is reasonable, subject lines are the highest-ROI optimization.
Three properties consistently distinguish strong subjects:
Mobile clients show 30-42 characters before truncation. The subject line should carry its headline payload in the first 30. Anything past that is bonus signal for the reader who decided to engage further.
Compare:
The first version puts its load-bearing content where the reader sees it. The second version puts marketing fluff in the visible portion and the specifics in the truncated portion.
Subject lines try to be witty. Subject lines that try to be witty get filed or skipped. The subjects that earn opens are clearer than clever.
Compare:
The second version tells the recipient what the email is about. The first version asks them to open to find out.
Subjects that reference specific entities (companies, dates, products, numbers) outperform vague subjects. The recipient pattern-recognizes specificity as research.
Compare:
Different stages of the customer relationship reward different subject patterns.
Question-style subjects, specific references, named research anchors. Examples:
The pattern: imply that the email is research-led without saying 'I researched you.' Specific quoted references do the work.
Direct, conversational, action-implicit. Examples:
The pattern: tell the reader what's expected of them without making it sound like a chore. 'Skip the tour' is a generous-feeling opener; 'Complete your onboarding' is not.
Soft service-flavored subjects for the first touch, more urgent ones for later touches.
Question-style, low-pressure, honest framing.
Headline-style. Treat the subject like an article title; spoil the punchline if it earns the open.
Plain, transactional, exactly what they describe.
Don't dress up transactional subjects with marketing voice. The trust premium of transactional sends comes from being honest about what they are.
The preheader is the second line of the inbox preview. Most clients show 40-100 characters of it; mobile clients show less.
Without an explicit preheader, the email client pulls the first line of body text as the preview. Often this is unhelpful — "Having trouble viewing this email?" or "View in browser" or a navigation menu item. The default behavior wastes the preview line entirely.
Treat the preheader as a coordinated second beat to the subject. Examples:
The pattern: subject sets up; preheader pays off. Don't repeat the subject; don't introduce a topic the body doesn't cover.
In HTML email: an invisible <span> at the top of the message body, styled to be hidden but readable by email clients:
<span style="display:none;font-size:1px;color:#fff;line-height:1px;">
Your explicit preheader text here.
</span>Most ESPs offer a preheader field that handles this automatically. If yours does, use it; don't try to engineer it yourself.
When you run a subject line through our predicted-open-rate tool, the rubric evaluates several factors:
Does the subject communicate the topic of the email in 30 characters? Strong: 8/10+. Mediocre: 5-7.
How many specific entities (companies, numbers, named products, dates) appear? Strong: 2+. Mediocre: 0-1.
Does the subject invite engagement or demand it? Open-ended questions invite; commands ('OPEN NOW') demand. Strong subjects invite.
ALL CAPS density, exclamation count, presence of well-known spam trigger phrases. Strong: low across all.
Is the headline payload in the first 30 characters? Strong: yes. Mediocre: payload truncated.
Specific patterns that consistently underperform: clickbait, generic 'just checking in,' subject lines that require opening the email to understand.
The composite score is a calibrated prediction, not a guarantee. For lists under 5,000 subscribers (too small for statistically significant A/B testing), the predicted score is the best signal you'll get without sending blind.
Subject-line A/B testing is the most-tested element in email marketing. Most teams test poorly.
To detect a 1 percentage-point lift in open rate with 95% confidence, you typically need 5,000+ recipients per variant. Smaller tests give directional results, not statistical ones. Don't declare winners on 200 recipients per variant.
If you change two things in the subject (a word and the punctuation), you can't isolate which caused the lift. Single-variable tests are slower but actually informative.
Lists too small for statistical A/B testing benefit from predicted scoring. Our predicted A/B winner tool gives a calibrated prediction; trust it for sub-5,000-subscriber tests where statistical testing isn't viable.
Subject line patterns drift. A subject that worked in 2024 might underperform in 2026. Re-test core patterns once a year; refresh based on results.
Three subject patterns that no longer work and should be cut:
And three to revive:
If you want to improve subject lines fast:
For deeper patterns, our subject line topic pages cover specific contexts with examples. And our subject line tool generates 10 variants from a brief, ranked by predicted open rate, with reasoning per option.
Subject lines reward the senders who take the 90-character constraint seriously. Most don't. The teams that do — running scoring on every send, A/B testing the high-volume sends, refreshing patterns annually — compound an advantage over years.
The 90 characters are small. The leverage is large.
Free anonymous run, no signup.
Try the editor