How do you prove which revision the operator actually used?
An auditor is rarely asking whether your current revision is correct. They are asking what was physically in front of the person building this unit, on a Tuesday eight months ago. Those are different questions. Your document control system answers the first one well and cannot answer the second one at all.
By Javier Ortiz
The gap is not sloppiness. It is structural.
Here is the part that makes this genuinely hard rather than merely annoying. Correct document control requires that when a revision is superseded, the obsolete copies are removed from the point of use. That is the right rule. It exists so nobody builds to an old drawing.
It also means the physical evidence of what was at the station last March is destroyed on purpose, as a consequence of the process working properly. You cannot go back and look, because doing the thing correctly is what removed the thing you would look at.
So a paper system can prove what should have been at the station on any given date. The master list says revision C was effective from February. What it cannot produce is evidence that revision C was the sheet in the binder, that revision B came out when C went in, or that the operator on second shift was not working from a photocopy someone made in December and kept in a drawer because the new one was harder to read.
Everyone knows this. It is why the answer to the auditor is usually a sentence about the process rather than a document, and why the finding, when it comes, is written against the point of use rather than against document control.
Four things plants do, and where each stops
All four are worth doing. None of them closes the gap, and it is useful to know exactly where each one runs out.
Revision number in the page header
Proves the sheet you are holding today is current. Says nothing about what was in the binder in March, because the old sheet was thrown away when it was replaced, which is exactly what a controlled document process instructs you to do.
A controlled-copy stamp and a distribution list
Proves a copy was issued to that station. It does not prove it arrived, that it replaced the old one, or that the old one was not still in the drawer. The list records an intention, not an outcome.
An operator sign-off sheet for the revision
Closer, because now a person has attested to something. But the signature says the operator acknowledged a revision on a date, not that this specific unit was built against it. It links a person to a document, and what an auditor wants is a link between a document and a part.
Periodic floor audits of posted documents
Genuinely useful and worth doing. It samples. A quarterly walk finds the wrong sheet at a station on the day you looked, which tells you the process failed, but it cannot reconstruct which units were affected or for how long. It detects the problem without bounding it.
What actually closes it
The question binds three things: a specific unit, a specific revision, and a specific moment. Any answer that binds only two of them will fail under a follow-up question.
That means the record has to capture the revision that was displayed, at the moment it was displayed, stamped onto the record for that unit. Not the revision that is current when someone runs the report later. The distinction sounds pedantic and it is the whole thing, because the two produce identical answers today and different answers a year from now, and a year from now is the only time anybody asks.
This is one of the few genuine structural advantages of an electronic instruction over a good paper system. Not that it is faster to update, though it is. It is that the display event itself is a recordable fact. The screen knows what it showed. A binder does not know what was in it.
Worth testing rather than assuming, including with us: ask to see a record from a unit built before the most recent revision change, and check which revision it names. If it names the current one, the system is looking up rather than recording, and you have the same gap with better lighting.
Common questions
What is the most common work instruction audit finding?
Not a missing document. It is an obsolete or uncontrolled document in use at the point of work, and the related inability to demonstrate which revision was actually in front of the operator when a given unit was built. The document control system is usually fine. The gap is between the controlled master and what physically sat at the station.
Why can paper not prove revision currency?
Because the sheet in the binder can be replaced without leaving a trace, and correct document control requires the superseded copy to be removed. That combination means the physical evidence of what was at the station on a past date is deliberately destroyed as part of the process working properly. A paper system can prove what should have been there. It cannot prove what was.
What does an auditor actually ask for?
Typically: show me the current revision of this instruction, show me it is the one in use at the station, and then show me the records for this unit built on this date and the revision that applied to it. The first two are usually answered easily. The third is where plants reach for a binder, and the reason it is hard is that the answer has to bind a specific unit to a specific revision at a specific moment.
Does an electronic system solve this automatically?
Only if it records the revision that was displayed rather than looking up the revision that is current. Those produce identical answers today and different answers a year from now, which is the only time the question ever gets asked. It is worth asking any vendor to show you a record from a unit built before the most recent revision change and confirm which revision it names.
We passed our last audit on paper. Why change?
Plenty of plants pass on paper, and passing is a legitimate outcome. The exposure shows up somewhere else: a customer escape where you cannot bound how many units were affected, so you contain everything shipped in the window rather than the four hundred that were actually built wrong. The audit tests whether your process is defensible. A recall tests whether your records are precise, and those are not the same test.
Doing document control correctly is what destroys the evidence. That is not a process failure, it is the reason the point of use needs its own record.