Establish what already happened
Illustrative scenario: Make receives inquiry FORM-101 and creates card JOB-01. Saving the card’s link back to the register fails. Restarting the entire workflow could create a second card even though the first is usable.
An incomplete execution is a retained workflow run that stopped at an error. This guide covers manual recovery of one such run. It is not a tutorial for building the entire integration or a guarantee that any existing scenario prevents duplicates.
You need access to Make’s history, the destination, and the original event identifier. If the scenario does not preserve an identifier, use the retained original and inspect the relationship manually. Do not automatically retry an ambiguous case.
1. Prevent two people from repairing the same case
Assign one person to the repair and ask others not to resend that input. Check whether Make has already scheduled an automatic attempt. If a retry is pending or in progress, have the maintainer establish its status before adding a manual attempt.
Download the recovery record. Enter the event identifier, execution link, responsible person, and findings. Retain the original arrival time: today is the recovery time, not a new inquiry’s arrival.
2. Identify the failed step
Open the scenario in Make, choose Incomplete executions, and select Details for the relevant run. Click the marked failed module. Record whether the failure concerns creating the card, saving its link, access, or missing data. These interface steps follow Make’s management documentation.
If no retained execution exists, inspect the history and original input. Incomplete-execution storage is a separate setting. Enabling it now does not recover previously unretained inputs. Continue with manual investigation rather than an assumed replay.
3. Open the destination app
Search for the original event identifier if the card stores it. Otherwise compare the client, request, time, and retained source. Matching a name alone is insufficient.
| Finding | Next step |
|---|---|
| The correct card exists; saving its link failed | Record its actual link; do not create it again |
| The card demonstrably does not exist; failure preceded the write | Repair the cause and assess one attempt from the failed step |
| The result cannot be established | Keep the case under review and assign the maintainer |
| The client has since canceled | Do not continue the original action; record the cancellation |
Microsoft explains how a lost response can cause an already completed operation to be repeated. An error or timeout alone does not establish a failed write.
4. Repair the cause and repeat only the safe part
An authorized administrator repairs revoked access. For invalid mapping, correct the relevant input or setting. Recheck that the requested outcome is still valid and has not been created by another run.
Make retries an incomplete execution from its failed module. When settings need changing, open the incomplete-execution detail, fix the module, save, and use Run once in that detail as documented. Do not confuse this with restarting the entire scenario. Ordinary Retry uses the original settings. If you cannot distinguish the routes, or the scenario may immediately message a client, hand recovery to the maintainer.
Make requires an active scenario for Retry. Activation may also resume other intake, so check the source and other waiting cases first. An automatic attempt does not replace destination inspection.
5. Close the case against the destination result
After recovery, open the card and its register link. There should be one correct card for the action, a usable link, and a clear next step. Record the verified result and whether the original incomplete execution remains open. Manually repairing the destination may not change its Make status; the maintainer must prevent a later repeat of the remaining work.
Practice on a harmless test case first. Documentation reviewed October 1, 2026. FORM-101 is illustrative, not a live CRM test. Fix recurring problems through design changes and pre-launch tests, rather than daily manual retries.
For your own records, use the blank tracker.