Not sure which subject line or opener will land? Find out inside the Sequence itself. Write a second version of a stage, and Mixmax splits your recipients evenly between the versions. Insights shows how each one performs, and when one pulls ahead, you choose it as the winner.

A stage becomes an A/B test when you add a variant. Your original content becomes Variant A, and the new version becomes Variant B.

Once a stage has a test, it shows an A/B badge with its number of variants in the stage list, and the Sequence shows an A/B badge in your Sequences list.

For the clearest results, change one thing per variant, like the subject line or the opener. Two or three variants usually work best, since each one needs enough recipients to tell them apart. You can edit a variant after the test starts, but its results keep counting, so for a clean comparison, add a new variant instead of rewriting one mid-test.

Only active variants go to new recipients. Each variant has an Active or Inactive switch in the bar at the top of the editor.

Switch the variant to Inactive and click Deactivate variant to confirm. New recipients stop getting it, and recipients still waiting for that stage move to the other active variants. The variant keeps its results, and you can turn it back on anytime. At least one variant has to stay active.

Switch the variant to Active and click Reactivate variant to confirm. New recipients start getting it again. Recipients who moved to other variants while it was off stay where they are.

Open the ⋯ menu, click Delete variant and confirm. This permanently removes the variant and its content, and recipients still waiting for it move to the remaining active variants.

You can only delete a variant before anyone has started receiving it. After that, turn it off instead.

On a stage that's already saved, these changes take effect right away. On a new stage you haven't saved yet, they apply when you click Save or Publish Changes.

By default, each recipient's variant is set to Auto. When you activate recipients, Mixmax takes turns assigning them to each active variant, so the split stays even, even across separate batches. Recipients who were already in the Sequence before you added a variant keep the one they had.

Each recipient keeps their variant for that stage, unless that variant is turned off, deleted or loses to a winner before they reach it. That includes recipients you pinned to a variant.

On the Review & Personalize screen, select a recipient and open a tested stage. Next to This recipient gets, change Auto to the variant you want them to receive. The list shows your active variants, and you can pin recipients on any tested stage. This guarantees the recipient will receive the selected variant.

You can personalize the copy of tested automated emails:

Open the Insights tab. Tested stages have an A/B badge and a row for each variant in the Email Performance, Task Performance and Recipients Status tables they appear in.

In the Email Performance table, a star marks the variant that leads each metric. For opens, clicks, replies, meetings and downloads, that's the highest rate. For bounces and unsubscribes, it's the lowest. A star isn't an overall winner, there's no star when variants tie, and early on a reply or two can earn one. Give the test time and focus on the metric that matters most to you, like replies.

Variants you've turned off keep their rows, marked Inactive. You may also see an Unattributed row for recipients who aren't assigned to a variant on that stage, like the first email of a Sequence added from Gmail.

New numbers can take a few minutes to show up. Reload the page to see the latest.

On the Recipients tab, each tested stage lists its variants in the sidebar with how many recipients have each one. Click a variant to see just those recipients, and use the Variant column to see who got which. If you download a CSV while a variant is selected, you get just that variant's recipients.

You decide when a test is done. When you've seen enough, click Choose as winner in the ⋯ menu at the top of the editor, or in a variant row's ⋯ menu in the Email Performance table. Then click Use this variant. You can choose any variant, including one you've turned off.

The winner becomes the only active variant:

If you've connected Claude or another AI assistant to Mixmax through the Mixmax MCP, you can set up and run tests from chat. Your assistant can:

For example:

To make changes, your assistant needs Create and edit sequences turned on in MCP settings, and A/B testing has to be on for your workspace.

A/B testing works on Sequences you create in the new editor after A/B testing is turned on for your workspace. For an existing Sequence, the A/B test section offers Clone this sequence. The clone is a new draft with the same stages and settings, and you can run your test there.

When you add a recipient to a Sequence while writing an email in Gmail, the first stage loads into your draft using that stage's first active variant, usually Variant A, and you send it yourself. That email isn't part of the test, so Insights shows it in an Unattributed row instead of under a variant. Every stage after that splits between variants as usual.

They keep sending. In the previous editor, Sequences created with A/B testing on are read only, with a note pointing you to the new editor. Turn Sequences Redesign back on to make changes.

The A/B test section is in the Stage settings tab of the side panel, on email, call, to-do and SMS stages in the new Sequence editor. Make sure Sequences Redesign is on in Mixmax Labs, and that you're editing the stage itself rather than personalizing a recipient. If you still don't see it, A/B testing may not yet be available for your workspace.

Once a Sequence is running, you can't change a stage's type. On a draft Sequence, changing a tested stage to a type that holds different content, like a call to an email, removes its variants. Mixmax asks you to confirm first.

You can add up to 8 variants to a stage, including ones you've turned off.