Mule-Dev-202 · Domain 8 of 8

Mule-Dev-202 Domain 8: Use Salesforce Flow Orchestrator to build parallel, multi-user, multi-step workstreams

Flow Orchestration coordinates long-running processes across people and automation. This section covers stages and steps, custom entry and exit conditions, assigning interactive work to groups or queues, and debugging and managing orchestrations.

Exam weight estimate
8%
Official objectives estimate
4
Estimated questions estimate
~5
Suggested study time estimate
5–7 hrs

About 12 min read Reviewed October 2026

Overview

Flow Orchestration (built in Flow Builder and often called Flow Orchestrator) combines existing flows into one coordinated process. An orchestration is made of stages that run in sequence, and each stage contains steps that can run in parallel. Background steps run autolaunched flows without user involvement. Interactive steps run screen flows and are assigned to a user, a public group or a queue; the assignee sees the work item on the record page through the Orchestration Work Guide component or in their list of work items.

By default, a stage completes when all its steps complete, and a step starts when its stage starts. You can customize these with evaluation flows: autolaunched flows that return a Boolean telling the orchestration whether a step may start or a stage may finish. Orchestrations can be record-triggered or autolaunched. Administrators monitor runs and work items, debug orchestrations, deploy them like other flow metadata, and can pause, resume or cancel runs. In hyperautomation, orchestrations often coordinate human approvals around automated steps that call Composer, APIs or RPA bots.

How it's tested

Questions describe a multi-step business process with several roles, parallel tasks and conditions for moving forward, then ask how to structure stages and steps, how to assign work, how to customize when something starts or ends, or how to troubleshoot a run. The best answer reuses existing flows as steps and uses evaluation flows for custom conditions.

  • Structure the process

    A process with sequential phases and parallel tasks must be modelled as stages and steps.

  • Control the timing

    A step should start only when a condition is true, or a stage should end early, and you configure an evaluation flow.

  • Route the work

    Human tasks must go to a team rather than a named person, and you assign interactive steps to a group or queue.

Reasoning it through

A new-hire onboarding process needs IT, facilities and payroll to complete tasks at the same time once an offer is accepted, then a manager review. The parallel tasks belong as three steps in one stage, and the review is a step in the next stage. Each team's task is an interactive step assigned to that team's queue so any member can pick it up, and the manager review starts when the first stage completes.

Objectives 4

Official objectives from the Salesforce outline, in outline order.

8.1 Combine multiple automated workflows into a coordinated process.Stages, steps and reuse of existing flows.
What it means
An orchestration sequences stages, and each stage holds one or more steps that run in parallel by default. Each step runs an existing flow: an autolaunched flow for a background step or a screen flow for an interactive step. Orchestrations can be record-triggered or autolaunched and pass data into steps through input variables.
What to understand
Put work that must finish before the next phase in separate stages; put work that can happen at the same time in the same stage. Reuse existing flows as steps rather than rebuilding logic. Background steps can call Composer flows, APIs or other automation, letting the orchestration coordinate people and systems in one process.
How it shows up
A process description lists sequential phases and parallel tasks, and you choose how to arrange stages and steps.
Common confusion
Steps in the same stage run in parallel by default; stages, not steps, provide the sequence.
8.2 Customize entry and exit conditions for evaluation flows.Using evaluation flows to control when steps start and stages finish.
What it means
By default a step starts when its stage starts, and a stage completes when all its steps complete. You can override this with an evaluation flow, an autolaunched flow of the evaluation type that returns a Boolean output telling the orchestration whether the condition is met. Step entry conditions can also depend on another step completing.
What to understand
Use step entry conditions to start a step only when a record meets criteria or another step has finished. Use stage exit conditions when a stage should finish before all steps complete, for example when one approval is enough. The orchestration re-evaluates the condition when related records change, so evaluation flows should be lightweight.
How it shows up
A step must wait for a field value or a stage should end after any one approval, and you configure an evaluation flow.
Common confusion
An evaluation flow decides whether a condition is met; it does not perform the step's work.
8.3 Assign interactive steps to groups or queues.Routing human work items to teams.
What it means
Each interactive step has an assignee: a specific user, a user stored in a resource or variable, a public group or a queue. Assigned users see work items in the Orchestration Work Guide on the related record page and can complete the screen flow there.
What to understand
Assigning to a group or queue lets any member pick up work, which avoids bottlenecks when people are absent. Dynamic assignment from a variable lets the orchestration route to, for example, the record owner's manager. Assignees need access to the record and the screen flow for the work item to be usable.
How it shows up
Work should go to whichever support specialist is available, and you assign the interactive step to a queue.
Common confusion
Assigning to a named user creates a single point of failure; groups and queues share the work.
8.4 Debug, deploy, and manage Flow Orchestration.Testing runs, deploying metadata and handling in-flight orchestrations.
What it means
Orchestrations can be debugged from Flow Builder, and their runs and work items can be reviewed to see which stage and step a process is in and where it stopped. Orchestrations are flow metadata, so they deploy with change sets, the Metadata API or DevOps Center, and administrators can pause, resume or cancel runs.
What to understand
When a run stalls, check whether an interactive step is waiting on an assignee, whether an entry or exit condition is never satisfied, or whether a background flow failed. Deploy referenced flows and evaluation flows together with the orchestration. New orchestration versions apply to new runs; in-flight runs continue on the version they started with.
How it shows up
A run has been stuck for days, and you choose where to look and what to check first.
Common confusion
Activating a new version does not move in-flight runs onto it.

High-yield concepts

The terms, rules and formulas this domain turns into questions most often.

  • Orchestration

    A Flow Builder process that coordinates flows across stages, steps and people.

    Remember
    Built for multi-user, multi-step, long-running work.
    Tested as
    Tool-selection and structure questions.
    Trap
    Using one long screen flow instead.
  • Stage

    A sequential phase containing one or more steps.

    Remember
    Stages run one after another.
    Tested as
    Sequencing questions.
    Trap
    Putting parallel work in separate stages.
  • Step

    A unit of work inside a stage that runs a flow.

    Remember
    Steps in a stage run in parallel by default.
    Tested as
    Parallelism questions.
    Trap
    Assuming steps run in order.
  • Background step

    Runs an autolaunched flow without a person.

    Remember
    Use it for automation and integration calls.
    Tested as
    Mixing systems and people.
    Trap
    Using a screen flow for unattended work.
  • Interactive step

    Runs a screen flow assigned to a user, group or queue.

    Remember
    Work appears in the Orchestration Work Guide.
    Tested as
    Human task questions.
    Trap
    Forgetting record access for assignees.
  • Evaluation flow

    Autolaunched flow that returns whether an entry or exit condition is met.

    Remember
    Controls timing; keeps logic separate from the work.
    Tested as
    Custom condition questions.
    Trap
    Putting business actions inside it.
  • Stage exit condition

    Rule for when a stage completes, by default when all steps finish.

    Remember
    Customize to finish early, such as after one approval.
    Tested as
    Early completion scenarios.
    Trap
    Waiting for every step when one suffices.
  • Step entry condition

    Rule for when a step starts, by default when its stage starts.

    Remember
    Can depend on another step or a record condition.
    Tested as
    Dependency scenarios.
    Trap
    Creating a new stage just to delay one step.
  • Orchestration Work Guide

    Lightning component that shows assigned work items on a record page.

    Remember
    Add it to the record page so assignees can act.
    Tested as
    Usability and troubleshooting.
    Trap
    Assigning work with no place to see it.
  • Run management

    Reviewing, pausing, resuming and cancelling orchestration runs.

    Remember
    In-flight runs stay on their starting version.
    Tested as
    Manage and debug objective.
    Trap
    Expecting a new version to fix running instances.

Side by side

Concepts candidates mix up, compared on the point that decides the answer.

Stage vs. step

ConceptMeaningKey differenceCommon trap
Stage Sequential phase Next stage waits for the exit condition Using stages for parallel work
Step Unit of work within a stage Parallel with sibling steps by default Expecting steps to run in order

Background vs. interactive step

ConceptMeaningKey differenceCommon trap
Background step Autolaunched flow, no user Automation and integrations Assigning it to a person
Interactive step Screen flow for an assignee Human work via work items Assigning to one named user only

Flow Orchestration vs. single flow

ConceptMeaningKey differenceCommon trap
Flow Orchestration Coordinates many flows and people over time Stages, parallel steps, work items Skipping it for multi-team processes
Single flow One automation path Best for one user or one transaction Stretching it across teams with pauses

Exam traps

Each trap is an answer that looks right. Learn the clue that should change your mind.

  1. One giant screen flow

    What candidates pick
    Build one screen flow that pauses and waits for each team.
    Why it looks right
    It keeps everything in one place.
    Why it's wrong
    Screen flows serve one user at a time; multi-user, parallel, long-running work is what orchestration is designed for.

    Clue to watch for Several teams act on the same record over days.

  2. A stage per task

    What candidates pick
    Create a stage for each department's task.
    Why it looks right
    It looks organized.
    Why it's wrong
    Stages run sequentially, so independent tasks would wait on each other instead of running in parallel.

    Clue to watch for The tasks can happen at the same time.

  3. Named user assignment

    What candidates pick
    Assign the interactive step to the team lead.
    Why it looks right
    Someone is clearly responsible.
    Why it's wrong
    Work stalls when that person is unavailable; a group or queue lets any member complete it.

    Clue to watch for Any member of the team can do the work.

  4. Wait for every approver

    What candidates pick
    Keep the default stage exit condition when one approval is enough.
    Why it looks right
    Defaults need no extra work.
    Why it's wrong
    The default waits for all steps; an evaluation flow for the stage exit condition lets it finish after the first approval.

    Clue to watch for The scenario says any one approver can approve.

  5. Fix in-flight runs by reactivating

    What candidates pick
    Activate a corrected version to unblock stuck runs.
    Why it looks right
    New versions usually fix bugs.
    Why it's wrong
    Running instances stay on the version they started with; they must be handled through run management.

    Clue to watch for Runs that started before the fix are still stuck.

Answer framework

Draw the process as phases and parallel tasks before choosing orchestration settings.

  1. Confirm the process needs orchestration: several people or teams, parallel tasks, or work that spans days.
  2. Group work that must finish before the next phase into stages and work that can happen together into steps in one stage.
  3. Make unattended work background steps with autolaunched flows and human work interactive steps with screen flows.
  4. Assign interactive steps to groups or queues where any member can act, and add the Orchestration Work Guide to the record page.
  5. Add evaluation flows only where default start and finish rules do not fit, then debug runs in Flow Builder and deploy the orchestration together with every flow it references.

Practice question

A loan process needs credit, compliance and legal teams to review an application at the same time. Any member of each team can complete its review, and underwriting must start only after all three reviews finish. How should the orchestration be designed?

Choose one answer

Want more like this? The Mule-Dev-202 practice test has questions for every domain.

Checklist

0 of 10

Tick each one you can do without notes. Your ticks stay in this browser.

Study strategy

Build one orchestration in a Developer Edition org; the hands-on version sticks far better than reading. Create a record-triggered orchestration on a custom object with two stages: the first with two interactive steps assigned to different queues and one background step, the second with a single review step. Add the Orchestration Work Guide to the record page and complete the work as different users.

High-priority topics are stage versus step behaviour, interactive step assignment and evaluation flows. Underestimated areas are run management and versioning, because questions about stuck runs often hinge on whether the work item has an assignee who can see it or whether an evaluation flow ever returns true.

Next, add an evaluation flow so that one step starts only when a field is set, and another that ends a stage after the first of two approvals. Break something deliberately, such as removing record access for an assignee, and see how the run behaves. In practice questions, sketch the stages and steps on paper before looking at the options.

FAQ

How much of the Mule-Dev-202 exam is Use Salesforce Flow Orchestrator to build parallel, multi-user, multi-step workstreams?

The Flow Orchestration section is 8% of the exam, about five of 60 questions as an editorial estimate.

What is the difference between a stage and a step?

Stages run in sequence and represent phases of the process. Steps sit inside a stage and run in parallel by default, each running a flow.

What is an evaluation flow?

An autolaunched flow that returns whether a condition is met, used to customize when a step starts or when a stage completes.

Can interactive steps be assigned to a team?

Yes. An interactive step can be assigned to a user, a public group, a queue, or a user identified by a variable at run time.

How does orchestration fit into hyperautomation?

It coordinates human and automated work in one process, with background steps that can call Composer flows, APIs or other automation and interactive steps for people.

How long should I study this section?

Plan about 5 to 7 hours as an editorial estimate, including building at least one orchestration end to end.

Other CRM & Business Applications certifications

Where the same subject shows up in other exams. Cards marked "Same subject" open the matching domain.

Test your Domain 8 knowledge

You have covered stages and steps, evaluation flows, interactive step assignment and orchestration run management. Check whether you can structure a multi-team process quickly.

Keep going