Flow & Process Automation

Flow types, before-save vs after-save, building common automations and choosing Flow or Apex.

Admin 2 min readFlowAutomationApproval process

Flow is Salesforce's main declarative automation tool. Workflow Rules and Process Builder are retired for new automation — migrate them to Flow.

Flow types

Type	Starts when	Example
Record-triggered	A record is created/updated/deleted	Default a field, create a task
Screen	A user launches it (button, page, action)	Guided wizard
Schedule-triggered	On a schedule	Nightly clean-up
Autolaunched	Called by Apex, another flow, REST	Reusable logic
Platform event–triggered	An event is published	React to integrations

	Fast Field Updates (before save)	Actions and Related Records (after save)
Updates the triggering record	✅ very fast, no extra DML	possible but costs a DML + re-save
Create/update other records	❌	✅
Send emails, call subflows/actions	❌	✅
Record Id available on create	❌	✅

Object: Account · Trigger: created · Optimize for: Fast Field Updates

Decision: AnnualRevenue >= 1,000,000 → Hot; >= 100,000 → Warm; default → Cold.
Assignment: {!$Record.Rating} = the chosen value.

No Update Records element is needed — changes to $Record are saved automatically.

Object: Opportunity · Trigger: updated · Entry condition: StageName = Closed Won · Only when a record is updated to meet the condition.

Update Records: Account where Id = {!$Record.AccountId} → Type = Customer.

The "only when updated to meet the condition" option is the declarative version of comparing Trigger.old and Trigger.new.

Keep one or few record-triggered flows per object and use entry conditions and trigger order.
Never put Get/Update/Create Records inside a Loop — collect into a collection variable and update once (same bulkification rule as Apex).
Add fault paths to handle errors.
Use subflows for reusable logic.

Choose Flow when	Choose Apex when
Logic is simple to moderate	Complex logic, heavy data volumes
Admins will maintain it	Callouts with complex parsing, advanced error handling
Field updates, records, emails, approvals	Precise control of transactions, unit tests, reuse in APIs

Multi-step sign-off (e.g. discounts > 20%) with entry criteria, approvers (manager, queue, user), approval/rejection actions and record locking.

Before-save vs after-save flow?
How do you avoid hitting limits in a flow loop? → No DML/queries inside loops; use collections.
How do you make a flow run only when a field changes to a value? → "Only when a record is updated to meet the condition requirements".
When would you pick Apex over Flow?