Each Workflow can have many Versions, but there is only one Version live in production at any given time. This is the Version that will be called when the Workflow is triggered by API.

Editing an existing workflow is easy, and has three basic steps:

1) Edit logic or add new rules

You can edit any of the existing rules by clicking on them, and then editing the If or Then fields in a simple rule, or the rows and actions in a decision table rule.

You see the available fields by typing Customer and pressing Tab for autocomplete. We also support comparisons (<, =, >=), logical operators (and, or, parentheses), and Excel-style formulas (beta)


You can always click Check Syntax on the right hand info panel to validate the new logic you've created, before clicking Save.

2) Test impact

There are three ways to test impact of your potential changes.

First, with the edited rule open, click on the Testing tab of the right hand info panel to display the backtesting panel. Here you can see what the impact would have been on the past 100 customers, using your proposed change to this rule.


Second, you can run individual test customers through your complete Workflow, including multiple changes on multiple rules. Click anywhere on the gray workflow graph to display the workflow versions pane, and click Test. The Test Workflow Modal will appear.


You can edit the fields of a sample customer in the left-hand panel. Clicking Test will then show the rules triggered by running this user through the current version of the workflow.

<aside> 💡 The Advanced Options tab lets you adjust the fields returned by data sources as part of the test run of your workflow. Be careful turning off the Mock API Run toggle, as this will make your test runs hit live data – this can incur costs for your company depending on the data source.


When you're done, click Close.

Finally, the Backtest button lets you test samples of 20, 50, or 100 past customers through your new workflow version, to see statistics on the impact of your changes. No live calls to data sources are made during backtesting.