Fixed! It looks like Microsoft, at some point this year, fixed a bunch of stuff with DocumentIDs.
For the first time ever Microsoft provides a consistent cornerstone of Document Mmgt.
Document identifiers now persist with the document across Office 365.
Search works, with appropriate guidance on using criteria correctly.
Configurable numbering format expanded and automatically generated numbering doesn't (seem) to duplicate anymore.
Direct document Identifier links work same as always - within the site-collection.
If you have need of proper document management in Office 365 and SharePoint Online, use the Document Id service - it will finally work as we would want it to.
To use the Document Id across the whole of SharePoint Online, you will have to enable it in every site-collection to avoid risk of potentially loosing an Id if/when a document moves around.
OneDrive can also be enabled for Document Id service, but it is hidden in the updated modern UI (working, but not visible)
DocumentID persists in synchronisation to desktop, but only in Office document metadata / Windows 10.
In platform Search you cannot use the number's display-format with ‘-‘ (hyphen) spacers as search-criteria.
We still only get one number service, and single configuration for format, per site-collection. If your solutions need other numbers, you will need to design or buy component.
While Document ID service has been available in the SharePoint platform for many releases, there were limitations in the implementation which made it unusable for years:
it could in theory generate duplicate numbers within site-collection;
it did in practice create duplicate IDs in different site-collections, depending on how you configured the pre-fix
i.e. moving a document from one site-collection to another could lose the identifier, or pick-up a new one in the destination.
In the last 6 months (2020) Microsoft seems to have invested some effort in resolving some of the inherent issues which have surfaced in the SharePoint Online.
To confirm the impact we tested if the document's identifier persists across site-collections when Moving a document from location-to-location:
testing for document being moved from:
SPO site-collection with default Id's, to alternative site-collection with default Id's
SPO site-collection with default Id's, to alternative site-collection with configured Id's
SPO site-collection with configured Id's, to alternative site-collection with default Id's
SPO site-collection with configured Id's, to alternative site-collection with configured Id's
testing impact of moving documents from:
SPO site-collection with default Id's to OneDrive with default Id's
SPO site-collection with default Id's to OneDrive with configured Id's
SPO site-collection with configured Id's to OneDrive with default Id's
SPO site-collection with configured Id's to OneDrive with configured Id's
OneDrive with default Id's to SPO site-collection with default Id's
OneDrive with default Id's to SPO site-collection with configured Id's
OneDrive with configured ID to SPO site-collection with default Id's
OneDrive with configured ID to SPO site-collection with configured Id's
In all tests the Document Identifier created persisted during the move!
Word of caution: the document metadata with the identifier is permanently removed if the document moves to a location which does not have the DocumentID service enabled
Modern search experience appeared somewhat broken, but turned out manageable based on testing.
Library search / site search using the document identifier as displayed in document properties panel gave mixed results e.g.
BONGOID001-4-4502 - Fails
DlcDocId:BONGOID001-4-4502 - Success
Global search (Organization), as with Library search - how we use search-term is key to best results e.g.
BONGOID001-4-4502 - Fails, no results
"BONGOID001-4-4502" - Fails, no results
BONGOID001 4 4502 Success
"BONGOID001 4 4502" Success
KEY LEARNING: remove the '-' (hyphen) from the document id number, or use the 'dcldocid:' pre-fix, to get results!
Search done good! Results from anywhere in SharePoint or OneDrive based on the Id.
The reality is that, unless you train your users on:
1. why document identifier is useful for referencing and locating authoritative content, and
2. why targeted search prevents 'groping' around through results
..you needn't bother, because its too easy to lose track of critical content in Office 365 unless you are disciplined or in a well designed (guided) solution.
Microsoft has (finally) provided a useful toolset, all we need now is a revolution in education.