An A/B test sends two versions of a step's message, Version A and Version B, to different prospects of the same campaign, so you can keep the one that works best.
Turn on an A/B test
You can test emails, LinkedIn messages and the note of a LinkedIn connection request. On the Sequence tab, click the three dots on the step card and turn on A/B test. Both versions start as a copy of the step's current message.
Write Version A and Version B
Click the step card. Message settings shows a tab for each version. Each version can be written by AI or as your own template, so you can also test an AI-written message against a template. More about writing messages here.
How prospects are split
Prospects reaching the step alternate between Version A and Version B, so each version goes to about half of them. Prospects who passed the step before you turned the test on count for neither. The version each prospect got shows next to their name in the step's prospect lists.
On the Messages tab, a message from a tested step shows which version it comes from. Heavy edits to these messages blur the comparison.
See which version wins
On the Reporting tab, Step performance shows a row for each version once the step has gone out to 40 prospects, so you can compare their opens, clicks, page visits and replies. Winning marks the version with a clearly better reply rate. More about campaign results here.
Turn off an A/B test
Turn off A/B test in the step's menu and choose the version to keep. The other one is deleted, along with the per-version results.
