GoHighLevel workflows now have a dedicated A/B Test action. It sends contacts evenly down 2 to 10 different paths, tracks which path converts best against a goal you choose, and can then route every new contact to the winning path automatically. Before this, the closest option was the Split action, which divides traffic but never tells you which branch actually won.
If you run nurture sequences, reactivation campaigns or booking follow-ups in HighLevel, this turns guesswork ("is the SMS-first version better?") into a measured answer inside the workflow itself. Below: what the action does, how to set it up step by step, how it differs from Split, and the mistakes to avoid.
Quick version: open a workflow under Automation → Workflows, click + to add an action, choose A/B Test, add 2–10 paths, set Winning Criteria (a goal event or a custom condition), choose a Test length, pick what happens After the test ends, save, then build the actions you want to compare under each path.
What the workflow A/B Test action does
HighLevel introduced the A/B Test action as part of its LevelUp Day 2026 releases, and the official help article was updated in late October 2026. HighLevel describes it as separate from the existing Split action. In short, it adds three things Split never had:
- Conversion tracking. Every path is measured against the same goal, so you can see how many contacts entered each path and how many converted.
- Winner selection. The path with the highest conversion percentage wins, either when the test ends on its own or when you declare a winner manually.
- Post-test routing. Once a winner is known, the workflow can send new contacts only to that path, without you editing anything.
Each path can contain whatever actions you want to compare: a different email, an SMS instead of an email, a shorter or longer wait, a different offer. Contacts are distributed evenly across the paths you create.
A/B Test vs. the Split action
Both actions branch contacts at random, so it's easy to confuse them. The differences matter when you choose which one to use:
| Split action | A/B Test action | |
|---|---|---|
| Number of paths | Up to 5 | 2 to 10 |
| How traffic is divided | Random, by percentages you set (must total 100%) | Evenly across all paths |
| Conversion goal | None | Goal event or custom condition |
| Picks a winner | No | Yes, highest conversion % |
| Ends automatically | No | By contacts entered, goal completions, or a date and time |
| After the test | Keeps splitting | Route new contacts to the winner, or keep splitting and show the result |
Use Split when you want an uneven, permanent distribution (for example, 80% to your proven sequence and 20% to something new) or simple load-balancing. Use A/B Test when the question is "which version converts better?" If you need to branch contacts by who they are (a tag, a source, a field value) rather than at random, neither is right: HighLevel points you to the If/Else action for that.
Split
- Up to 5 paths
- Percentages you set
- No conversion goal
- No winner selection
A/B Test
- 2 to 10 paths
- Even split across paths
- Goal event or custom condition
- Routes new contacts to winner
How to set up an A/B Test in a GoHighLevel workflow
Step 1: Add the action
Go to Automation → Workflows and open the workflow you want to test, or create a new one. In the builder, click the + icon where the test should start and select A/B Test. Give the action a clear name, such as "Day 1 follow-up test – SMS vs email", so the statistics make sense months later.
Step 2: Create your paths
Use Add path to create between 2 and 10 paths and rename each one to describe what it tests. Two or three paths is usually the practical choice for a small business: every extra path divides your traffic further and makes it slower to reach a clear result.
Step 3: Set the winning criteria
Under Winning Criteria, choose one of two options:
- Goal event — conversion is a tracked event. HighLevel's help article lists email activity, clicking a trigger link, appointment status, a form submission, a payment received and an invoice paid as examples.
- Custom condition — conversion is defined by contact or custom field values, such as a lead qualification status you set elsewhere in your automations.
All paths are judged on the same criteria. The criteria only measure success; they don't decide which path a contact enters.
Step 4: Choose the test length
Under Test length, decide when the test ends:
- Total contacts entered — after a set number of contacts have entered the action.
- Goal completions — after a set number of conversions.
- Date and time — at a specific moment, using the account's time zone.
Step 5: Decide what happens after the test
Under After the test ends, pick either Send new contacts to the winning path or Keep splitting and show the result. With the first option, the winning path gets a Winner badge on the canvas and the other paths are muted for new contacts. Contacts already partway through a losing path still finish the steps they were on.
Step 6: Build each path, then publish
Click Save action, add the actions you want to compare under each path, and save or publish the workflow. Change only one thing between paths where you can, so you know what caused the difference.
Reading the results
Open the action's Statistics tab to see, for each path: contacts entered, contacts converted, conversion percentage (converted ÷ entered × 100) and the winner. Two controls live here:
- Declare winner ends the test early when you already have enough data.
- Reset test starts a fresh comparison. Saving after a reset clears the earlier results, so export or note them first.
If two or more paths tie on the highest rate, HighLevel treats the first of them as the winner. Put your current, proven version in the first path so a tie defaults to what already works.
Practical tests for agencies and small businesses
- Speed-to-lead channel: immediate SMS vs. immediate email for new form leads, judged on appointment status.
- Follow-up timing: a 1-hour wait vs. a 1-day wait before the second touch.
- Offer wording: two versions of a reactivation message to old customers, judged on payment received or a trigger link click.
- Invoice reminders: a friendly reminder vs. a firmer one, judged on invoice paid.
- Booking nudges: a calendar link vs. a "reply YES to book" message, judged on appointment status.
Tests like these are where automation stops being "set and forget" and starts improving month over month. If you'd rather have someone design the test plan and wire up the workflows, our GHL workflow automation service covers exactly that.
- Only one change between paths
- Proven version placed in path one
- Goal tied closely to the paths
- Enough contacts entered each path
- Re-entry settings checked
- Account time zone confirmed
- Results noted before any reset
Common mistakes and gotchas
- Calling a winner too early. HighLevel notes that path counts can be uneven when few contacts have entered. A 3-vs-1 result on a handful of leads is noise, not a winner. Pick a test length that gives each path a meaningful number of contacts before routing traffic.
- Testing too many things at once. If path B changes the channel, the copy and the timing, you won't know which change mattered.
- A goal that can fire outside the test. If the goal is "appointment booked" and contacts can also book from an ad or your website, every path gets credit for bookings it didn't cause. Choose goals that are tied as closely as possible to what the paths actually send.
- Forgetting re-entry. After a reset, returning contacts may land in a different path than last time. Check your workflow's re-entry settings if contacts can enter more than once.
- Using A/B Test for uneven splits. The A/B Test action always divides evenly. If you only want 10% of traffic on an experiment, use Split instead.
- Time zone surprises. Date-based test endings follow the account time zone, which matters if you manage clients across US time zones from one agency account.
What this means for agencies
For agencies, the A/B Test action is a simple way to show clients measurable improvement from the automations you build. Instead of reporting that a nurture sequence "is running", you can report that the SMS-first version booked more appointments and is now the default for every new lead. It also makes snapshots more valuable: a sequence that has already won tests across clients is a stronger starting point than an untested template.
The catch is volume. Smaller sub-accounts may need weeks to collect enough data, so set expectations with clients and favor high-traffic steps like new-lead follow-up for your first tests. If you want help choosing what to test, setting goals that measure the right thing, or cleaning up existing workflows first, you can hire a GHL expert to plan and run the tests for you.
Frequently asked questions
Does GoHighLevel have A/B testing in workflows?
Yes. Workflows have an A/B Test action that divides contacts evenly across 2 to 10 paths, measures each path against a goal event or custom condition, and picks the path with the highest conversion percentage as the winner.
What is the difference between the Split action and the A/B Test action?
Split sends contacts down up to 5 paths using percentages you set and has no goal or winner. A/B Test splits evenly across 2 to 10 paths, tracks conversions, ends on a condition you choose and can route new contacts to the winning path.
What goals can a workflow A/B test measure?
You can use a goal event, such as email activity, a trigger link click, appointment status, a form submission, a payment received or an invoice paid, or a custom condition based on contact or custom field values.
When does a GoHighLevel A/B test end?
When the test length you chose is reached: a total number of contacts entered, a number of goal completions, or a specific date and time in the account time zone. You can also end it early with Declare winner in the Statistics tab.
What happens to contacts already in a losing path?
If you choose to send new contacts to the winning path, only new contacts are routed to the winner. Contacts already in other paths finish the steps of the path they entered.
Want this set up in your GoHighLevel account?
Get hands-on help from a GHL expert — scoped per project, no retainer. Start with a free 30-minute call, or see how we handle GHL workflow automation.
Book a Free Call →Sources
- HighLevel Support: Workflow Action – A/B Test
- HighLevel Changelog: A/B Test Action – Compare Workflow Paths and Route Contacts to the Winner
- HighLevel Support: Workflow Action – Split
- HighLevel Blog: LevelUp Day 2026 features and enhancements