A/B testing campaigns

What you can test

Subject line, preview text, sender name, body creative, call-to-action, and send time. One variable per test — Orbit360 will warn you if two variants differ in more than one dimension.

Sizing the test

Choose a holdout percentage per variant. The default splits 10% / 10%, measures for a set period, then sends the winner to the remaining 80%. For small lists, Orbit360 recommends a full 50/50 split instead, because a 10% slice of 4,000 contacts cannot resolve a realistic difference.

Reading results

The results panel shows lift, the confidence interval, and a plain-language verdict. A test that has not reached significance says so; it does not declare a winner because one bar is taller. Peeking at an unfinished test does not change its outcome, but acting on it is how teams fool themselves.

Applying learnings

Promote a winning variant into your campaign templates so the next campaign starts from the better version.