A/B testing posts

A/B testing lets you publish two (or more) versions of a post to a portion of your audience, then automatically promote the winning variant to the remainder. No manual analysis required.

Creating a test

1. Open any draft post in the editor.
2. Click A/B test in the top-right toolbar.
3. Add a second variant — either write it manually or click AI variant to generate an alternative.
4. Set the test split (default: 50/50) and the winning metric (Engagement rate, Click-through rate, or Reach).
5. Set the evaluation window (minimum 4 hours, maximum 48 hours).
6. Approve the test — both variants are scheduled simultaneously.

How the winner is chosen

After the evaluation window, Anthyx compares the metric you chose. The variant with the statistically better result (minimum 90% confidence) is declared the winner. If neither reaches the confidence threshold, the test is declared a draw and both results are reported without auto-promotion.

Viewing results

Go to Posts → [Post name] → A/B results. You'll see side-by-side performance for each variant, the declared winner, and the auto-promotion status.

Plan limits

A/B testing is available on Starter and above. The number of concurrent active tests depends on your plan tier. Check Settings → Plan for your current limit.