1
00:00:01,000 --> 00:00:04,935
A workflow is useful only when someone understands what to do

2
00:00:04,960 --> 00:00:06,570
when it fails.

3
00:00:06,600 --> 00:00:09,930
A fictional training company receives a registration.

4
00:00:09,960 --> 00:00:13,930
Its workflow checks the fields, creates a contact and prepares a

5
00:00:14,240 --> 00:00:15,210
confirmation.

6
00:00:15,240 --> 00:00:18,850
Then the connection to the customer system becomes unavailable.

7
00:00:19,240 --> 00:00:23,690
Define the information each stage receives and what confirms success.

8
00:00:24,360 --> 00:00:28,730
A received registration, a confirmed contact and a sent message are

9
00:00:28,840 --> 00:00:30,970
different events, with different evidence.

10
00:00:31,480 --> 00:00:35,050
Keep a recognisable reference for the same registration.

11
00:00:35,400 --> 00:00:38,810
Before repeating a step, check what has already happened.

12
00:00:39,000 --> 00:00:43,850
Otherwise a temporary failure can produce another contact or a duplicate message.

13
00:00:44,680 --> 00:00:48,490
No response from the customer system does not necessarily mean nothing

14
00:00:48,520 --> 00:00:49,450
was saved.

15
00:00:49,640 --> 00:00:52,210
Investigate the destination first.

16
00:00:52,600 --> 00:00:57,250
Give an unknown outcome a visible status rather than assuming it failed.

17
00:00:57,480 --> 00:01:01,450
An administrator needs to see the failed stage, the relevant input

18
00:01:01,640 --> 00:01:03,370
and the permitted recovery action.

19
00:01:04,120 --> 00:01:09,730
Keep access restricted; credentials should not appear as plain text in the workflow.

20
00:01:10,120 --> 00:01:14,170
Test a repeated registration, missing fields and a failure after

21
00:01:14,280 --> 00:01:14,810
saving.

22
00:01:15,320 --> 00:01:19,930
Then check both systems: recovery should leave one correct outcome,

23
00:01:20,000 --> 00:01:22,810
while unresolved work remains visible to its owner.

24
00:01:23,640 --> 00:01:26,290
Develop one bounded workflow first.

25
00:01:26,520 --> 00:01:30,250
Have its future administrator locate a failed run and follow the

26
00:01:30,320 --> 00:01:33,690
agreed recovery process before connecting more stages.

27
00:01:34,520 --> 00:01:39,610
Read OXIAE’s n8n consulting guide for the connection between process

28
00:01:39,640 --> 00:01:42,330
design, exceptions and maintenance.