Testing email subject lines and preview text with AI

Most email marketers generate one or two subject line options, choose the one that sounds better, and send. AI changes that dynamic — variants are cheap, so the decision moves from writing to choosing. But choosing between ten subject line options without evaluation criteria is just opinion at larger scale.

What makes a subject line work (criteria, not instinct)

A subject line works when it does two things: gets the email opened and sets accurate expectations for what's inside. Open rate measures the first; click-through and unsubscribe rate measure the second. A subject line that inflates open rate by promising something the email doesn't deliver degrades list quality over time.

The criteria differ by audience and goal. Cold outreach favors specificity — the subject line's connection to the recipient's actual situation matters more than clever wordsmithing. An engaged list responds to familiarity and brand voice; a re-engagement sequence needs directness about the gap rather than an oblique tease.

Generating variants: how many is enough

Five to eight subject line variants is the productive range for most campaigns. Fewer than five is not enough variation to surface a real difference; more than eight produces diminishing returns — the additional variants tend to be slight permutations of ones already on the list.

The brief for generating variants should specify the evaluation criteria before requesting options. If you don't know how you'll evaluate the variants, you don't yet know what you're looking for. Include the email's goal, the audience's likely current state, and any key constraint. Clarity and brevity are defaults; override them when the audience and goal call for something different.

Preview text: the second subject line most teams ignore

Preview text is the secondary copy that appears in the inbox alongside the subject line — in most email clients, it's the 40–80 characters immediately after the subject. Most teams leave it unset or auto-populated from the first line of the email body, which wastes a second opportunity to earn the open.

The preview text should complement the subject line rather than repeat it. If the subject line establishes the topic, the preview text establishes the relevance: why this email matters for this reader now. If the subject line creates curiosity, the preview text is where to add a hint of specificity that makes opening feel worth the time.

Evaluating options before sending to the full list

Evaluation criteria applied before sending are more useful than intuition applied after. Three questions cover most subject line decisions.

First: does this subject line reflect what's actually in the email? Subject lines that overpromise inflate open rates temporarily and degrade list health over time.

Second: does this subject line match the audience's current context? A subject line written for someone who has never heard of your product performs differently than one written for someone mid-consideration.

Third: do the subject line and preview text combination earn the open? Read them together, not separately — the inbox shows both.

Copper Sun's email modules include built-in evaluation criteria matched to your audience and campaign goal, so each subject line decision has a framework rather than a blank checklist. See how it works.

What A/B test results actually tell you

A/B test results tell you which subject line performed better with that specific audience on that specific day. They don't tell you why it performed better, or whether the winner would hold up in a different context. The insight is the margin, not just the result — a 2-point open rate difference is noise; a 12-point difference is a signal worth understanding.

What to log after each test: the winning variant, which evaluation criteria predicted the winner, and whether they held up. Patterns across ten to twenty sends reveal what your audience actually responds to. That pattern-level insight is what separates a list you understand from a list you just happen to be sending to.

For building the sequence these subject lines will head up: writing a nurture sequence with AI. For the broader email marketing guide: using AI for email marketing. For the newsletter workflow where subject lines repeat weekly: newsletter production with AI.

Frequently Asked Questions

Can AI write better subject lines than I can?

AI generates more variants faster and without the cognitive overhead of starting from a blank page. Whether those variants are better depends on the evaluation criteria — AI doesn't know your audience's specific preferences unless you brief them in. The advantage is volume and variation; the judgment about which variant fits the campaign still belongs to the marketer.

How do I test email subject lines without a big list?

Send to a segment rather than a full split. Even a 200-person test group produces directional signal on open rate. What you're looking for isn't statistical significance in the academic sense — it's a consistent pattern across several sends. If the same type of subject line outperforms across five consecutive sends, that's a real preference, not a one-off result.

How many subject line variants should I generate?

Five to eight is the productive range. Fewer gives you too little variation to surface a real difference; more produces diminishing returns — most additional variants are slight permutations of ones already on the list. Generate variants against a clear brief, evaluate against criteria, and decide from that smaller set.

What makes a subject line work for B2B email?

Specificity and relevance to the recipient's actual situation — more so than cleverness or curiosity gaps. B2B readers tend to open emails that signal direct relevance to a problem they're actively working on. Vague curiosity hooks can inflate open rates while reducing click-through if the email doesn't deliver on the implied promise. Test directness before testing creativity.