Skip to main content

A/B test subjects and templates

An A/B test sends different versions of your campaign to a small test cohort first, waits, then automatically sends the winning version to everyone else. You can test subject lines, whole template designs, or both at once.

Turn the test on​

In the composer, the A/B test section has a toggle labeled Test subjects or templates. Switch it on and the variant editor appears.

  • Variant A is always your main campaign: the subject you typed above and the template you picked. You do not configure it separately.
  • Click Add variant to add up to three more variants (B, C, and D).

For each extra variant you can set:

  • A subject line. Leave it empty and the variant inherits A's subject.
  • A template. The dropdown defaults to Same as variant A; pick another of the brand's saved designs to test a different layout or content.

A variant counts as valid if it has its own subject, its own template, or both. To send, you need at least one valid variant besides A; the hint next to the send button reminds you if it is missing.

Configure the test​

Three controls shape the test:

  • Test size: the percentage of your audience that takes part in the test (default 25%). Each variant goes to an even slice of that cohort. The remaining recipients are held back for the winner.
  • Decide after: how long to wait before picking the winner, in hours (default 4, up to a week).
  • Winner by: Open rate or Click rate.

A summary line under the controls restates the whole plan in plain words so you can sanity-check it before sending.

tip

If any variant uses a different template, the winner is always judged by click rate and the Open rate option is disabled. Opens do not change with the email body, so only clicks can tell two designs apart.

How the winner is decided​

  1. When you send, the test cohort ships immediately, split evenly across the variants. Everyone else is parked as the holdout.
  2. After the Decide after window has passed and the whole cohort has finished sending, the platform compares the variants by your chosen metric. Ties break toward the variant with the larger sample.
  3. The winning variant's subject (and template, if variants differed) is applied to the campaign, and the holdout is released automatically. No action is needed from you.

Send-time optimization is skipped during an A/B test: the cohort has to ship promptly so it can be measured inside the window. If you had Predictive send time selected, enabling the test switches the campaign to Send all at once.

Read the results​

Once the campaign is sending or sent, the report shows an A/B results panel:

  • Each variant with its subject, how many it was sent to, and its open and click rates. Variants that used a different design are tagged.
  • While the test runs, the panel notes how many recipients are held back for the winner.
  • After the decision, the winning variant is highlighted and the panel confirms the winner was sent to the remaining recipients.

The rest of the report works the same as any campaign; see Track, pause, and cancel a campaign.