A/B Subject Line Comparer — Which Line Wins, Check by Check

Who this is for: For the two-drafts-and-a-deadline moment: the argument you're having about line A vs line B, settled by checklist instead of seniority.

Two lines side by side — scores, check-by-check differences, and the verdict on which to send.

Quick answer: Paste two lines: each gets the ten-check score, the comparer names every check that differs and who wins it, then gives a verdict. 'Your Q4 plan: 2 spots left' (81) beats 'URGENT!!! 100% FREE miracle offer' (26) on nine of ten checks.

Line A

Score A

81/100

Strong

Line B

Score B

26/100

Poor

Verdict: Line A wins by 55 points.

4 checks differ — the reasons are below, and you can overrule any of them with list knowledge the checklist doesn't have.

Check by check

37 chars — the sweet spot (mobile shows ~35, desktop ~70).Length43 chars — the sweet spot (mobile shows ~35, desktop ~70).
8 words — scannable at a glance.Word count6 words — scannable at a glance.
A number — specifics read as credible and get opened.Concrete numberA number — specifics read as credible and get opened.
Statement only — a question frame adds pull.Curiosity frameStatement only — a question frame adds pull.
Speaks to the reader directly — the strongest single lever.Second personNo 'you/your' — make it about the reader.
No emotional trigger words — one curiosity/urgency word helps.Power wordsNo emotional trigger words — one curiosity/urgency word helps.
No emoji — fine for B2B; one can lift B2C opens.EmojiNo emoji — fine for B2B; one can lift B2C opens.
Sentence case — reads like a person, not a promo.Capitalization2/5 words in caps — caps for ONE word max.
No exclamation marks — calm reads confident.Exclamations6 exclamation marks — one max, zero is calmer.
Clean against the 175-word trigger list.Spam wordsHigh-risk: 100% free, miracle, urgent; medium: offer (−27).

Pre-screen rule: both arms should clear 70 before they earn a real A/B test — and lists under ~1,000 per arm get more value from this checklist than from a split test.

Core facts

Ten checks	Length · words · number · curiosity · you/your · power · emoji · caps · exclamations · spam
Verdict	Score gap + itemized check deltas
Pre-screen rule	Both arms ≥70 before spending an A/B test
Worked gap	"Your Q4 plan: 2 spots left this month" 81 vs "URGENT!!! 100% FREE miracle offer inside!!!" 26
Tie rule	Within ~2 pts: pick the voice that sounds like you
Status	Rules follow public email-marketing benchmark practice (2026): ~35–70 visible characters by client, 175-word spam trigger list, sentence case, one clear idea per line. Client truncation figures are engineering approximations — devices, fonts and zoom shift them.

How should I compare two subject lines?

Paste both. The comparer scores each line on the same ten checks as the tester, then diffs them check by check: where they differ, it names the winner and why — 'A wins on length: 38 chars vs 74, B truncates on mobile'. The verdict is the score gap plus the itemized reasons, so you can overrule it knowingly (you know your list; the checklist doesn't). Two honest uses: before an A/B test, to only test lines that both clear 70 (testing a 40 against an 80 wastes a send); and instead of one, when your list is too small for statistically meaningful splits — a 12-point score gap is a better decision than a 50-open sample.

Common uses

Settling the two-drafts argument with named, itemized reasons
Pre-screening A/B test candidates so both arms are sendable
Reviewing copy changes: did the rewrite actually score better?
Coaching: show a junior the two lines and the checks that split them

Frequently Asked Questions

How do I A/B test subject lines properly?

Test one variable, on a real send, with enough volume: split 10–20% of the list, wait the client's window (Gmail: 4 hours max), and let the winner go to the rest. Test big differences first (offer vs curiosity), not punctuation. The comparer is the pre-screen: both candidates should clear 70 before they earn a test slot.

How big should my list be for an A/B test?

Rule of thumb: each arm needs ~1,000 recipients to detect the 20–30% relative open-rate gaps subject lines typically produce. Below that, differences drown in noise — which is exactly when the comparer's checklist verdict is the better decision tool.

Can the comparer be wrong about my audience?

Yes — it compares against best-practice checks, not your subscribers. If your list demonstrably loves ALL-CAPS, the comparer will keep deducting for it and your open data will keep disagreeing. When the checklist and your data split, your data wins.

What differences matter most between two lines?

In order: spam-word hits (up to −33 combined), truncation (mobile shows ~35 chars — a cut payoff is a lost payoff), missing second person (−8 of upside), and length band. The delta list shows exactly which of these fired for your pair.

Should I send the higher-scoring line every time?

Send it when the gap is 5+ points or the deltas name a fixable difference. When scores sit within a couple of points, the checklist is indifferent — pick the one that sounds more like you, and save the A/B machinery for real questions.

More guides

Catchy Email Subject Lines

What actually makes a subject line catchy — the five mechanisms, with scored examples for each and the cheesy line to avoid.

Email Subject Line Examples

A working library of scored subject line examples by scenario — welcome, promo, event, re-engagement, announcement — with the check each one passes.

Email Spam Words to Avoid

The 175-word spam trigger list, explained: the 107 that should never appear, the 68 that are fine alone — and why words matter less than they used to.

More Deliverability checkers

Spam Word Checker

Scan your subject line against 175 spam trigger words — high-risk and medium-risk flagged, with fixes.

Subject Line Length Analyzer

See exactly where Gmail, Outlook, Yahoo and Apple Mail cut your line — and the character band that fits.

Inbox Preview Tool

Subject + preheader rendered as Gmail desktop, Gmail mobile and Outlook will show them — before you send.