A/B Test Node

What is an A/B test node?

The A/B Test node is a traffic splitter that automatically routes users across several scenario variants in a set proportion. It lets you test hypotheses — which welcome message, price, or sales funnel performs better — on real users, with real numbers, instead of guessing.

Use cases:

Compare two versions of a welcome message and see which one drives more sign-ups.
Test different prices for the same product.
Compare a short funnel against a long one.
Check which call-to-action button gets clicked more often.

Once a user lands in one of the A/B test's branches, they will always follow that same branch going forward — even if they restart the bot. This guarantees every person sees a consistent, coherent scenario from start to finish.

Adding an A/B test

Click the flask icon on the canvas to add an A/B test node.
In the settings panel, set what percentage of traffic each branch (A, B, …) should receive. Weights must add up to 100%.
For each branch, select (or create with one click) the command the user will be routed to.
Add more branches with the "Add test" button if needed (up to 10 variants).
Publish the scenario.

Reading the results

On the node itself (in statistics mode), you can see how many users landed in each branch.
Next to each outgoing connection, the traffic share and current conversion rate for that branch are shown.
The chart icon button in the node's header opens detailed analytics: a table per branch (traffic, conversions, win rate) and a daily trend chart — so you can confirm a branch's lead is a stable trend, not a fluke.
In the analytics modal, you can switch between goals (conversions) if your scenario tracks more than one.
You can reset a test's statistics if you've changed your hypothesis and want to start comparing again.

Tips

For a fair comparison, don't change the branch weights after the test has started — it will skew the results.
Start with a simple two-branch (A/B) test before moving to multivariate tests.
Pair it with a "Conversion" node to automatically see which branch drives more of your target actions (leads, payments, sign-ups).
Let the test accumulate enough traffic before drawing conclusions — small samples can be misleading.