Classify these webshop situations as regular, potentially irregular or irregular: (a) an IdP user without company and no CRM customer; (b) an IdP user with company and no CRM customer; (c) a CRM customer created five minutes ago whose IdP user has no company yet.
(a) is regular, a private customer who registered in the self-service; (b) is irregular, a company can only come from a CRM customer, so it is reported as an error; (c) is potentially irregular, most likely the sync has not run yet, so it is reported as a warning.
The reasoning in each case is "could the model have produced this state legitimately?"
- (a) User without company, no CRM entry. Exactly the result of case 1.3 (self-registration). Nothing to report.
- (b) User with company, no CRM entry. The only way the model puts a company on a user is via a CRM customer (cases 1.1, 1.2, 2.2), and deleting that customer deletes the user completely. So this state has no legitimate origin: perhaps a deletion was not propagated, someone edited the IdP by hand, or an e-mail change created a duplicate. Error.
- (c) Fresh CRM customer, IdP not yet updated. This is a legitimate intermediate state between "customer saved" and "sync completed". Whether it is a problem depends on time, which the data alone cannot tell. Warning, and if it persists past the sync interval it deserves investigation.
Tip: The same data can change category with time. A state that is "potentially irregular" for five minutes becomes a real failure after five days, which is why warnings need someone who follows them up.