Ordering Workflow Triggers, Actions, and End Steps

Key Takeaways

  • Start with the trigger and its data, then connect the step that needs that data.

  • Run input-producing actions before dependent comparisons or actions.

  • Connect each branch to a terminal step or to a later common path that reaches one.

  • Use execution inputs, outputs and explicit Failure ends to verify the result of error paths.

Last updated: November 2026

Read dependencies before arranging the canvas

A workflow starts with a trigger and its payload. A trigger establishes why it ran and provides the initial data. A workflow action performs an operation or changes the workflow data. An operator can choose a path, repeat work, validate data, or terminate execution. Each workflow has at least one action and at least one end step. On the visual canvas the directed connection determines which step runs next. The visible position of a box and its descriptive name do not determine order.

Draw the dependencies from each consumer to the step that supplies its inputs. For example, an Identity Attributes Changed event can supply the identity reference and the changed attributes. If the next comparison needs a value absent from that payload, use Get Identity to retrieve it before comparing. Confirm the actual trigger schema and action output before choosing the variable in the Variable Selector. A JSONPath expression can select a value in that data. An expression naming a later action or a prior action on a path that was skipped cannot make the required value available on the current path.

Piece of a scrambled designPlace it by asking
TriggerWhat actual event starts this run and what input does it provide?
Get Identity actionWhich later steps need the returned identity attributes?
Comparison operatorDoes the compared value exist on every path that reaches this step?
Send Email actionWhich condition authorizes this notice and where does its recipient come from?
Success endDid this path perform the required action or intentionally do nothing?
Failure endWhich unrecovered problem should remain visible as an overall failed run?

Order a worked example

A staging workflow should notify a controlled shared mailbox when a qualifying identity attribute changes. In this design, filter the Identity Attributes Changed trigger for the relevant change. Get Identity returns the current identity data. The next Compare Strings step tests a configured department value obtained from that action, and Send Email belongs only on the matching branch. Put the pieces in this order along the matching path: trigger, Get Identity, comparison, Send Email, then Success. Route the other branch to a separate Success when skipping is the expected result. It can also merge into a common downstream path if the remaining actions are valid on both branches. Any eventual path must terminate. Sending the message before the department comparison would notify for both outcomes, even if the workflow later takes the intended no-message path.

If Send Email encounters an action error while error handling is off for that action, the workflow stops as failed. If the design handles errors there, configure the action's success and error branches explicitly. For an unrecovered send error, capture the relevant details and finish the error branch with an End Step - Failure, so that an execution marked complete as a success cannot hide the issue. If policy allows continuation after an anticipated benign error, define which steps still run and which result is expected. Verify what the system received before trying again if an external action may already have sent or updated something.

Verify all paths, including the quiet one

Test a matching identity, a nonmatching identity, and a controlled error. Check the trigger sample shape and use the actual Get Identity output to verify the field mapped into the comparison. Step simulation allows the sequencing and data handling to be checked with controlled values. An enabled action in a test can execute, so use approved sandbox destinations for actions that should run. A missing input can be handled with a type or existence check before the dependent comparison when the business requirement needs a deliberate missing-value path.

On the matching branch, confirm the step took the intended output and produced the expected message at the destination where appropriate. On the nonmatching branch, confirm that Send Email never ran and that the intentional no-op was a success. On the failure branch, read the actual execution details and verify the intended final status. Later changes to the workflow can introduce a variable dependency or branch mistake even if its visible name and purpose still look familiar. Recheck those specific connections when changing a producer or its output.

See SailPoint's workflow concepts, triggers, building and error handling, operators, and workflow actions.

Test Your Knowledge

A workflow must compare a department value available only from Get Identity and email only on a match. Which order meets the requirement on the matching path?

A

Trigger, Send Email, Get Identity, Compare Strings, Success

B

Get Identity, Trigger, Compare Strings, Send Email, Success

C

Trigger, Compare Strings, Get Identity, Success, Send Email

D

Trigger, Get Identity, Compare Strings, Send Email, Success

Test Your Knowledge

Error handling is enabled for an email action. If sending fails and the error remains unrecovered, where should that branch terminate to mark the run as failed?

A

At the original trigger.

B

At End Step - Failure after any required error handling.

C

At Success so that the run stays in its failure state.

D

At an unconnected action with no end.

Sections you finish are checked off in the contents.