A/B test a Journey

A split sends each person down one of several paths, so you can compare whole sequences, not just one email. Example: three emails over a week versus one email.

Add a split

1. Set a goal first. A split needs one to measure against.
2. Choose +, then Split here. Everything after that point moves into path A and is copied into path B, 50/50.
3. Change path B: different emails, timing, or fewer steps.
4. Add path for up to 4 paths, or Add control for a path with no emails. A control shows whether emailing helps at all.
5. Set each path’s share (1–99%, adding up to 100%).

Splits can’t be nested, and only go in the main flow. How many paths you can add depends on how many people enter: 2 normally, 3 from about 1,200 a week, 4 from about 3,200.

Choose how to judge it

• Measure success by: goal reached, replied, opened, clicked, or delivered.
• People per path: 400 by default for goal or reply, 200 for the others. Minimum 30.
• Choose a winner: I decide, or Automatically.

Pick the winner

A split is ready when every path has enough people and one leads with at least 95% confidence. The Journey card and Home show it.

Choose Decide this split. The table compares the paths. Promote makes the winning path the only path. Keep running collects more data.

Promoting an active Journey pauses it, so you can check the result before turning it back on. With Automatically, Banger promotes the winner when it’s ready.

A/B test one email instead

To test versions of a single Journey email, open it, add a version, and publish both. Then choose Results to start the test: what varies (subject, design, or sender name), the share of people tested (10–50% or everyone), the metric, and whether Banger picks the winner automatically after a set time. It works like Broadcast A/B tests.

Try it with your AI

Add a split to my Onboarding Journey after the welcome email: path A keeps the current 3 emails, path B sends one summary email. Measure by goal reached and promote the winner automatically.