make home / Playbook
Automation / practical steps
Design automations that can recover
An independent Make workflow planning guide focused on inputs, duplicate events, errors and human approval.
STEP 01
Write the trigger and outcome
Describe what starts the workflow, which system supplies the record and what should be different at the end. Use a small sample with ordinary, incomplete and repeated inputs. Decide which actions need human approval before configuring an integration.
STEP 02
Map and validate fields
Record required fields, formats and destination identifiers. Check blank values, dates, currency and time zones explicitly. Use a stable source record ID to find the destination item instead of matching only on a person’s name or a changing display label.
STEP 03
Make retries safe
A repeated event should not create a second invoice, message or order. Plan a lookup or other duplicate-control method appropriate to the destination. Test the exact point where a timeout can happen after an action has already succeeded.
STEP 04
Route failures to an owner
Decide what is retried, what is held for review and what notification includes enough context to investigate. Keep secrets out of logs and alerts. Use the official Make documentation for current error handling, scheduling and plan limits before putting the scenario into service.
Choose the right next step
- Read-only summary: validate completeness and source timestamps.
- Record creation: define duplicate protection before enabling retries.
- Customer-facing action: include an approval or exception path appropriate to the consequence.
Worked example
A form submission creates a CRM task. The design stores the form response ID with the task. When a delivery retry repeats the event, the workflow finds the existing task instead of creating another. A missing customer email routes the record to review rather than sending an empty message.