Relayby Switch Labs
Back to User Guide

Automation Best Practices

Build effective automations without creating chaos.

Automation Philosophy

The Goal

Automations should:

  • Remove repetitive tasks
  • Speed up processes
  • Ensure consistency
  • Let humans focus on humans

The Risk

Bad automations can:

  • Annoy customers
  • Create chaos
  • Be hard to debug
  • Damage trust

What to Automate

Good Candidates

Routing:

  • Assign based on tags
  • Route to specialists
  • Balance workload

Status Management:

  • Set pending after reply
  • Reopen on customer response
  • Close stale tickets

Notifications:

  • SLA warnings
  • Escalation alerts
  • Important updates

Tagging:

  • Auto-categorize by content
  • Apply workflow tags
  • Mark for review

Bad Candidates

Customer Responses:

  • Don't auto-reply with "solutions"
  • Don't close without human review
  • Don't send mass updates without thought

Complex Decisions:

  • Refunds and exceptions
  • Escalation judgment
  • Priority overrides

Building Automations

Start Simple

First automation should be:

  • One trigger
  • Clear condition
  • Single action
  • Easy to test

Example:

Trigger: Ticket created
Condition: Tag = "billing"
Action: Assign to billing team

Test Thoroughly

Before enabling:

  1. Create test ticket
  2. Trigger the automation
  3. Verify result
  4. Test edge cases

Use Test Mode

Enable test mode first:

  • Automation runs
  • Actions logged but not executed
  • Review what would happen
  • Enable when confident

Common Automation Patterns

Assignment Routing

Name: Route by tag
Trigger: Tag added
Condition: Tag = "technical"
Action: Assign to technical-team

SLA Warning

Name: SLA warning
Trigger: 1 hour since created
Conditions:
  - Priority = Urgent
  - Status = Open
  - No agent reply
Actions:
  - Notify assigned agent
  - Add tag "sla-warning"

Auto-Pending

Name: Pending after reply
Trigger: Agent replied
Condition: Status = Open
Action: Set status = Pending

Reopen on Reply

Name: Reopen on customer reply
Trigger: Customer replied
Condition: Status = Resolved OR Closed
Action: Set status = Open

Auto-Close Stale

Name: Close stale tickets
Trigger: 7 days since updated
Condition: Status = Pending
Action: Set status = Closed

Avoiding Problems

Infinite Loops

Problem: Automation triggers itself

❌ Bad:
Trigger: Status changed
Action: Change status
→ Triggers itself forever

Solution: Add exclusion conditions

✓ Good:
Trigger: Status changed to Open
Condition: Previous status was Resolved
Action: Notify agent

Conflicting Automations

Problem: Two automations fight

❌ Bad:
Automation 1: If urgent → assign to Alice
Automation 2: If billing → assign to Bob
Ticket: Urgent billing ticket → who wins?

Solution: Clear priority and conditions

✓ Good:
Automation 1: If urgent AND billing → assign to Alice (billing lead)
Automation 2: If billing AND NOT urgent → assign to Bob

Over-Notification

Problem: Too many alerts

Solution:

  • Consolidate notifications
  • Use appropriate channels
  • Don't notify for everything

Customer Spam

Problem: Automated messages annoy customers

Solution:

  • Limit automated customer emails
  • No auto-replies to auto-replies
  • Add rate limits

Organizing Automations

Naming Convention

Use clear, descriptive names:

  • ✅ "Route billing tickets to billing team"
  • ❌ "Automation 3"

Grouping

Organize by purpose:

  • Routing automations
  • Status automations
  • Notification automations
  • SLA automations

Documentation

For each automation, document:

  • What it does
  • Why it exists
  • Who to contact
  • Last reviewed date

Testing Changes

Before Changing Production

  1. Understand current behavior
  2. Document what you're changing
  3. Test in isolation
  4. Consider edge cases
  5. Have rollback plan

Making Changes Safely

  1. Disable automation
  2. Make changes
  3. Test with test tickets
  4. Re-enable
  5. Monitor for issues

Rollback Plan

Always know how to undo:

  • Keep previous configuration
  • Know how to disable quickly
  • Have support contact ready

Monitoring

Regular Reviews

Monthly:

  • Review automation logs
  • Check for errors
  • Verify still needed
  • Update as needed

What to Watch

  • Automation firing count
  • Error rates
  • Unexpected behavior
  • Team feedback

Audit Trail

Keep record of:

  • What changed
  • When it changed
  • Who changed it
  • Why it was changed

Scaling Automations

As Volume Grows

  • More automations may be needed
  • But complexity increases
  • Document everything
  • Review regularly

Team Knowledge

  • Multiple people should understand
  • Don't let one person own all automations
  • Cross-train
  • Document thoroughly

Red Flags

Signs of Problems

  • "I don't know why this automation exists"
  • Customers complaining about spam
  • Tickets bouncing between queues
  • Automations failing silently
  • Nobody knows what triggers what

How to Fix

  1. Audit all automations
  2. Delete unused ones
  3. Simplify complex ones
  4. Document remaining ones
  5. Assign ownership

Checklist for New Automations

Before creating:

  • Is this actually needed?
  • Could this be simpler?
  • What could go wrong?
  • How will I test it?
  • Who needs to know about it?

Before enabling:

  • Tested with test tickets
  • Tested edge cases
  • Team informed
  • Documentation updated
  • Monitoring plan in place

After enabling:

  • Verified working correctly
  • No unexpected behavior
  • Scheduled review date

← Team Collaboration | Knowledge Base Tips →