Controlled copies and the uncontrolled copy problem
A released drawing is worth very little if the people building to it have a different version. Distribution is where most document control quietly stops working.
What a controlled copy is
A controlled copy is a rendition of a released document that knows what it came from. It carries a watermark, the revision it renders, the moment it was produced, and the identity of the location it was published to.
That last part is what distinguishes it from a PDF. Manufacturing PLM records that this revision was published *to this destination*, which means the set of places a document has reached is a queryable fact rather than an oral history.
The essential property follows from it: when the source revises, every controlled copy of the previous revision is marked superseded automatically, wherever it was sent. A supplier portal shows it struck through. A SharePoint or Drive copy is stamped. A printed copy cannot be reached, which is why printed copies get a different treatment.
Printed copies, which are the hard case
Software cannot supersede paper. Any document control system that claims otherwise is describing a policy rather than a mechanism, and the shop floor knows it.
So printed controlled copies are handled by making the print an event. Printing produces a numbered copy recorded against a holder and a location, and the register of outstanding printed copies is a report somebody can act on — which is the only honest answer available.
When the source revises, the register tells you exactly which printed copies are now wrong and who holds them. The system cannot retrieve them; it can make retrieving them a specific, finite task rather than a walk around the building hoping.
Copies printed without going through that path are uncontrolled by definition, and every rendition carries a watermark saying so when printed outside the controlled route. It is a small thing that changes behaviour, because an operator holding a sheet marked *uncontrolled copy* asks a question they would not otherwise ask.
The failure it prevents
A machined part is produced to a drawing pinned beside a CNC machine. The drawing is correct, clean and two revisions old. The tolerance on one feature tightened in revision C after a fit problem.
Four hundred parts are produced to revision A. They are all within revision A's tolerance and all wrong. The cost is not the scrap — it is the two weeks of investigation before anyone thinks to check which revision the printed sheet shows, because it never occurred to anyone that the paper on the wall could be stale.
A copy nobody can find is a copy nobody can supersede. The register exists so the answer to *where are the outdated copies* is a list rather than a search.
How it meets the rest of the product
Publication is a release-time action rather than a separate task, so a document reaching release publishes to its configured destinations without anyone remembering to do it. The destinations are configured on the document type once, which is where the reliability comes from.
The supplier portal is the destination that removes the most work: suppliers see the current controlled copy of what they are building, superseded copies visibly struck through, and no email attachment anywhere in the flow.
Every publication and every supersession is an audit event with an attributed actor and a timestamp. At an audit, the question *how do you control distribution* has a demonstrable answer — a register, a supersession history, and a report of outstanding printed copies — rather than a description of intent.
How it meets your ERP
The common anti-pattern is attaching drawings to purchase orders in the ERP. It feels like the right place because the PO is what the supplier receives, and it produces a snapshot: the drawing as it was on the day that PO was raised, frozen, with no relationship to what has happened since.
The alternative is that the ERP references the controlled copy rather than holding one. A reference supersedes itself; an attachment does not. Where your ERP supports a document URL on the item or the PO line, that is where it should point.
Manufacturing PLM does not read documents out of your ERP or treat its document store as a source. Where drawings currently live there, migration brings them here and the ERP's copies become references — which is a one-time piece of work with a permanent payoff.
Where the boundary is
Manufacturing PLM cannot supersede a copy it does not know about. A drawing emailed as an attachment, saved to a laptop, or forwarded to a supplier outside the portal is outside the mechanism, and no feature can reach it. The mechanism works by making the controlled route the easy one.
It also does not render every file format. Renditions are produced for the formats configured at deployment; where a rendition cannot be produced, the native file is published with its watermark applied as metadata rather than being silently passed through as though it were controlled.
Facts
| A controlled copy | Knows its source revision, its destination and its moment |
| Essential property | Automatic supersession when the source revises |
| Digital destinations | Supplier portal · SharePoint · Drive · ERP reference |
| Printed copies | Numbered, registered to a holder and a location |
| On revision | The register names which printed copies are now wrong, and who holds them |
| Outside the route | Watermarked uncontrolled copy when printed |
| ERP | Should reference the copy, not attach a snapshot to a PO |
| Cannot reach | Emailed attachments · laptop saves · forwarded files |
Frequently asked
What makes a copy controlled rather than just a PDF?
It knows what it came from. A controlled copy carries a watermark, the revision it renders, the moment it was produced and the destination it went to — which makes the set of places a document has reached a queryable fact rather than an oral history.
How do printed copies get superseded?
They do not, and no software can supersede paper. Printing is recorded as an event producing a numbered copy against a holder and a location, so when the source revises the register names exactly which printed copies are wrong and who holds them.
What about copies printed outside that route?
They are uncontrolled by definition, and every rendition carries a watermark saying so when printed outside the controlled path. It is a small thing that changes behaviour, because an operator holding a sheet marked uncontrolled asks a question they otherwise would not.
Why not attach drawings to purchase orders?
Because an attachment is a snapshot of the day the PO was raised, with no relationship to what has happened since. A reference to the controlled copy supersedes itself; the attachment sits in the ERP looking authoritative for as long as the PO exists.
Does publication need to be remembered?
No. Destinations are configured once on the document type, and publication happens at release rather than as a separate task somebody actions afterwards. That is where the reliability comes from — the step nobody has to remember is the step that never gets skipped.
What does this give us at an audit?
A demonstrable answer instead of a description of intent: a register of what was published where, a supersession history with attributed actors and timestamps, and a report of outstanding printed copies. Distribution control is the part most organisations cannot evidence.
What if a file format cannot be rendered?
The native file is published with its watermark applied as metadata rather than being silently passed through as though it were a controlled rendition. Renditions are produced for the formats configured at deployment, and anything outside that set is handled explicitly.