Two revisions on the same site
A walk round the works, and a question that sounds administrative: which revision are you working to?
The printouts come out, and they are not the same. Engineering issued a new revision some days ago. The technical office has it. The crew on the wall doesn't, because the copy in their hand was printed before it existed. Sometimes the office has it and the subcontractor has not been sent it. Sometimes everybody has it except the one gang that needed it.
Nobody in that chain did anything wrong. The revision was issued, and the issuing was recorded. What was never confirmed is the thing that actually matters, which is whether the previous version stopped being used.
On most projects this gets found the same way — during a quality check, or on a site walk, when somebody happens to ask. It is not usually found by the system, because the system recorded the sending rather than the withdrawal.
What happens when it is found
The outcome varies more than people expect, and it is worth being honest about that rather than reaching for the worst case.
Often the work is early enough that it is simply stopped and restarted against the current drawing. Sometimes the change doesn't affect what was being built at all, and the work continues untouched. Occasionally it is expensive.
But the cost of any single instance is not really the point. The point is that on a large job you can never be entirely confident that everybody is looking at the same revision, and that uncertainty is permanent unless something is done about the withdrawal rather than the issue.
The half of document control nobody measures
Issuing a revision is straightforward. There is a transmittal, a date, a distribution list, and a record that it went out. Every project can show you that record.
Removing the previous revision from the places it is being used is the other half, and it is the half that is almost never tracked. The old drawing is on a wall in a site cabin, in a folder in a subcontractor's office, in somebody's truck. Sending the new one doesn't remove any of those.
So the useful definition is not the administrative one. Document control has succeeded when there is one valid revision in use and no other, and it has not succeeded merely because a transmittal was sent.
If two revisions are live on the same site, the process is unfinished regardless of what the register says.
The copy that left the system
There is a structural reason this is hard, and it has nothing to do with anybody being careless.
A register controls a file. Work is done from paper. The moment a drawing is printed it leaves the system entirely: it goes into a folder, onto a wall, into a vehicle, and no revision control reaches it. Issuing a new file supersedes the old file. It does nothing at all to the sheet already pinned up in a cabin.
The subcontractor boundary makes it worse. You issue to their office and record that you did. What happens inside that office — who prints it, who gets a copy, whether the previous one comes down — is not visible to you, and the transmittal register will show the distribution as complete either way.
Which is why the check has to be physical. Not more fields in the register. Somebody walking over and looking at what is in a hand.
Current on which date
There is a timing question underneath this that matters for the report specifically.
A revision issued on the thirtieth doesn't make the work done on the twentieth correct or incorrect. What was built was built to whatever was current at the time, and judging it against a drawing that didn't yet exist is unfair and useless.
So the useful question is never simply which revision is current. It is which revision was current when the work was done, and whether the change since then affects it. Answering that needs a dated record, which is the second reason the register has to hold more than one event.
Why this is a reporting problem
It would be easy to file this under quality and move on. It belongs here for a specific reason.
Progress claimed against a superseded drawing is not progress. The work may exist physically, and it may still have to be modified, so the percentage reported that month was describing something that had not been finished in the sense the report implied.
That makes revision status an input to your figures rather than a document management detail. It sits alongside the corroborating records from Week 8: before accepting a quantity, it is fair to ask which drawing it was built to and whether that drawing is current.
It is also one of the six readiness tests from Week 9. An activity is not ready if the current revision has not reached the people who will build from it, and reached means in their hands rather than in the register.
What the register has to record
Most transmittal registers record one event: sent. Two more are needed before the picture is complete.
Acknowledged, meaning the recipient confirmed they have it — not that the email left. And withdrawn, meaning the previous revision has been taken out of the places it was in use, with somebody's name against it.
Three columns instead of one.
The first is administration. The other two are the process.
System design
Most distribution registers record one event. The other two are what tell you whether a single revision is actually in use.
| Record | Produced by | Required quality | Verified against | Feeds |
|---|---|---|---|---|
| Transmittal | Document control | Dated, with the distribution list | The register | Revision status |
| Acknowledgement | Recipient | Confirmed by them, not by the sending system | The reply | Distribution completeness |
| Withdrawal | Named person on site | Old copy out of use, with a name against it | Site spot check | Single valid revision |
| Revision in use | Spot check | What the crew is holding today | Physical check | Readiness, progress validity |
The fourth row is not a system field and can't be. It is a walk, done occasionally on a small sample, and it is the only thing that verifies the other three.
Practical insight
Pick three drawings that were revised in the last month and are being worked to right now. For each one, don't check the register. Go and look, or ask somebody who will go and look.
Ask two questions of whoever is holding the drawing. Which revision is this, and when did you get it. Then compare that against what the register says was issued.
If all three match, your distribution works and you have learned something worth knowing. If one doesn't, you have found the gap on the cheapest possible sample, and you have found it before somebody builds a fortnight of work to it.
Key takeaways
- Revision control usually runs on people telling each other, not on a system, however good the register looks.
- The new revision reaching the office is not the same as it reaching the gang on the wall.
- It is normally found on a site walk or a quality check, because the system recorded the sending rather than the withdrawal.
- Outcomes vary: often the work is early enough to stop, sometimes the change is irrelevant, occasionally it is expensive.
- Document control has succeeded when one revision is in use and no other. A transmittal is not the finish line.
- Progress claimed against a superseded drawing is not progress, which makes revision status a reporting input.
- A register needs three events, not one: issued, acknowledged, and the old copy withdrawn.
Records born here. Transmittal register · revision status list · withdrawal record for superseded copies.
What is coming next
Every source so far has been inside the project. The next one keeps its own books, closes on its own date, and produces records that are correct for a purpose that is not yours.
Next week: commercial — commitments, accruals and invoices, and the delivery note that was never built to allocate cost.
Enjoyed this lesson?
Join with Google to get each new lesson the moment it's published — and help me see which topics matter most to you. No spam, one email a week, unsubscribe anytime.
Already following on LinkedIn works too — this is just for the weekly email.