Not sure whether an official opener or a more casual tone gets better replies? Now you don't have to guess. A/B testing lets you run multiple versions of a Message action side by side, split your leads between them automatically, and see which version actually performs better.
What this feature does for you
Instead of picking one message and hoping for the best, you can test real variations against each other using your own audience. Each variant gets its own slice of leads and its own analytics, so you can see exactly which wording drives more replies and fewer failures — then double down on what works.
Availability
A/B testing currently supports Message, Send Invite, and Email actions. Support for Sales Navigator InMail is on the way.
This is a beta feature, available to users on the Advanced plan.
Starting a test for Message action
Find a Message action in your campaign.
Click Run A/B test.
In new or draft campaigns, this button is always visible.
In active or paused campaigns, hover over the Message action to reveal it.
This creates a second variant of your message. Once you have two variants, the button changes to Add variant so you can keep testing.
You can run up to 3 variants per action at a time. We capped it at 3 to keep things simple — more variants means fewer leads per version, which makes results harder to trust.
Variants are auto-named alphabetically (A, B, C...) and they cannot be renamed. If you delete one, the remaining variants keep their original letters — they won't shift or renumber.
A/B testing for Send Invite action
A/B testing is available for Send Invite action if you have LinkedIn Premium plan that allows you to send connection requests with a short note.
If you have a free LinkedIn account, A/B testing for Send Invite action is unavailable because of the personalized connection request monthly limit imposed by LinkedIn. The limit is equal to 5 personalized connection requests per month that makes it impossible to test multiple connection request notes.
How leads are split across variants
Every new lead that reaches the action is randomly assigned to one of the currently enabled variants.
Turn a variant's toggle off, and it simply stops receiving new leads — no new assignments, but its existing data stays intact.
At least one variant needs to stay on. If you try to save or launch with everything toggled off, you'll get a blocking notification so you don't accidentally pause the whole action.
Editing, pausing and deleting variants
Use the toggle to turn variants on or off. These changes apply only after you save the sequence.
Editing a variant that's already running follows the same rules as editing any other running action — so your usual editing behavior carries over here too.
If you decide to delete a variant, click on the trash icon that appears on hover.
Deleting a variant doesn't erase its history — its stats stay counted in your campaign's overall analytics.
If you want to delete the whole action — delete all variants one by one.
Reading and evaluating your results
This is where the feature really pays off. Each variant tracks its own analytics — sent, replies, failed leads — displayed as separate tags right on the action, so you can compare variants at a glance without digging through reports.
Give it time before drawing conclusions. Let enough leads flow through each variant before deciding on a "winner" — small sample sizes can be misleading, especially for a variant that's still starting from zero.
Once you've identified a stronger-performing variant, turn off the weaker ones (or delete them) to send all new lead to a better performing variant.