Skip to content
OmniLeadDocs
Sign in

A/B test subject lines

Send two subject lines to a split of your leads, compare reply rates and keep the one that works.

Available on Growth2 min readLast updated

An A/B test sends two versions of a subject line to different leads in the same sequence step, then shows you which one earns more replies. A/B testing is included in Growth and Agency.

Add a variant

  1. Open the sequence

    Go to Sequences and open the sequence. You can add a variant before launch or while it's running.

  2. Select an email step

    Click the step you want to test. The first email is usually the best place to start.

  3. Click Add variant

    A second subject field, B, appears under the original, A.

  4. Write the variant

    Change one thing only: the length, a question versus a statement, or a specific detail. Merge tags work in both.

  5. Save

    Click Save. New sends from this step alternate between A and B.

How leads are split

Each lead is assigned A or B when the step is sent to them, alternating so both variants get an even share of leads. The assignment sticks: if the step is a follow-up in the same thread, it keeps the subject that lead already received.

Rotating mailboxes doesn't bias the test. Each mailbox sends both variants.

Read the results

Open the step and click Results. For each variant you'll see:

MetricMeaning
SentEmails sent with this subject
RepliesLeads who replied to this step, excluding out-of-office and unsubscribe replies
Reply rateReplies divided by sent
BouncesUseful to spot a list problem, not a subject problem
OpensOnly shown if you turned on open tracking

OmniLead compares reply rates, not opens. Opens are unreliable because many mail apps load or block images automatically.

When is there a winner?

Small samples lie. OmniLead marks a variant as Leading only once both variants have enough sends for the difference in reply rate to be statistically meaningful. Until then the results show Not enough data yet.

If you're sending to a small, carefully chosen list, you may never reach that point. That's fine: at small scale, the quality of the list and the first line matter far more than the subject.

Pick a winner

  1. Open the results

    Open the step and click Results.

  2. Choose the winner

    Click Keep A or Keep B. The other variant is archived with its results.

From then on every send from that step uses the winning subject. You can start a new test against the winner at any time.

What to test

  • Length: "Quick question" versus "Question about your Q4 hiring plan"
  • Specificity: a generic line versus one that names their company or work
  • Form: a question versus a statement
  • Personalization: {{company}} in the subject versus none

What's next