Keep a Failed Automation from Losing a Lesson Inquiry
A recovery checklist for independent coaches: preserve the inquiry, distinguish temporary failures, and prevent duplicate follow-ups.
A trial-lesson inquiry can move through several systems before anyone answers it. A form receives it, a spreadsheet records it, and an email service sends the acknowledgement. A failure in one step should not make the original inquiry disappear. This guide proposes a small recovery design you can adapt; it is not a tested integration or a promise that every failure can recover without intervention.
Start with a durable inquiry record
Give each inquiry an identifier at the point where you first receive it. Keep that identifier unchanged as the inquiry moves between systems. A planning record might contain inquiry_id, received_at, contact_email, lesson_type, status, delivery_state and last_error. Use invented records while developing the workflow.
Keep booking status separate from delivery state. For example, someone may already be booked even while an acknowledgement is awaiting delivery. Treating these as one field makes it too easy for a delayed retry to put a booked customer back into a follow-up sequence.
Separate a retryable problem from a missing decision
A brief connection failure may disappear on a later attempt. A missing email address will not become valid just because the workflow repeats. Record the error category and choose the next action accordingly.
For a temporary failure, use a bounded retry policy appropriate to the connected service. For incomplete input, preserve the record and mark it as held. Do not invent a customer address or replace a required fact with a plausible-looking value merely to produce a successful run. A held inquiry is visible; a fabricated successful record can conceal the problem.
What Make documents
Make's documentation says incomplete executions are disabled by default and must be enabled in scenario settings. When enabled, unfinished runs can be stored for recovery. Some supported errors can retry automatically. This is a recovery feature, not a guarantee of unattended resolution.
Its error-handling overview distinguishes handlers such as Retry, Skip and Resume. These choices have different consequences. Skipping an error may allow the overall run to appear successful even when an individual item did not finish. Consult the current documentation for the modules and error types in your own workflow before choosing a handler.
Prevent a second message after a retry
Before each send attempt, look up the inquiry identifier and the specific message identifier. A suggested message identifier could combine the inquiry identifier with a purpose such as trial-acknowledgement. If a confirmed delivery record already exists, do not send again.
The awkward case is a timeout after the email provider accepted a message but before your workflow recorded the result. Automatically resending may create a duplicate. Prefer a provider-supported idempotency mechanism or a delivery-status lookup when available. If neither is available, hold ambiguous cases for reconciliation rather than claiming exactly-once delivery. This is an architectural recommendation, not a statement that every email integration supports those features.
Recheck customer state at the moment of sending
An inquiry may turn into a booking while a failed workflow waits. Re-read the latest booking and communication preferences immediately before sending a follow-up. A message that was appropriate yesterday may no longer be appropriate today. Keep any necessary consent and opt-out handling separate from technical delivery success.
Test with five fictional inquiries
Use one ordinary inquiry, one missing an email address, one repeated submission with the same identifier, one booked customer and one simulated connection failure. For each, write the expected record state and expected message count before running the test. Compare the stored results with those expectations.
Add a sixth test for an uncertain delivery result. The expected outcome should be an explicit unresolved state unless your connected provider offers a reliable way to determine whether the message was accepted.
Define what success means
Count completed inquiry records, held records and duplicate-prevention events separately. A green scenario result is not the same as every inquiry being answered. Your recovery design should make unfinished work visible without losing the original request.
Start with one form and one destination. Add more services only after the recovery path is understandable and the sample cases behave as expected. Product plans, module behavior and limits can change; verify the linked official documentation when implementing this design.