The handover readiness review, run at sixty per cent complete

Late on in a job there is always a meeting where somebody has printed the asset register, or the thing being called the asset register, and the project manager is working down the blanks asking which can be filled in by Friday. The mechanical subcontractor left months ago. The commissioning engineer is on another site and has stopped answering. Several serial numbers were typed off photographs of nameplates, and one of those photographs is of the wrong unit.

That meeting is where a project discovers handover was supposed to have been assembling itself all along, and was not.

Handover gets treated as a completion event. The pack is put together when the building is nearly finished, the trades are leaving and the remaining labour is compilation — the worst possible moment to find out whether the record exists. Run the same review at sixty per cent complete and it becomes a gate. It asks whether the four objects operations will inherit — asset tags, inspection and test plans, O and M drafts, and the issued set — already exist in a form you can defend. A handover meeting at ninety-eight per cent is a surprise.

How to run it

Pick one live system or package that is being installed this week: a plant room, a floor of terminal units, a fire-alarm loop, a lift shaft. Do not start with the archive. Do not “clean” the common data environment the night before. The point is the live set, as it actually is.

Name the information standard, asset information requirements, or quality plan you will test against — including the status-code list and the tagging convention those documents actually publish. If there is no list, write that down. It is already a finding.

Walk the four objects in order. Asset tags first, because an ITP, an O and M draft or a drawing that cannot name the plant is storage. Then ITPs, because a tag without a contemporaneous inspection is a label. Then O and M drafts, because a vendor pack you can still refuse is cheaper than a ZIP assembled after demobilisation. Then the issued set, because the other three cite titles, and two live-looking revisions of one title poison every later check.

Write findings as tag numbers, ITP identities, vendor titles and sheet numbers. “Mostly on track” is not a finding. A tidy dashboard that nobody walked is not readiness.

Why sixty per cent and not later

Sixty per cent complete is mid-build. Structure is up, services are going in, plant is arriving, finishes have started in some places and not others. The people who know the facts — the installer, the tester, the vendor, the document controller — are still contracted, and revisions are still moving. That is the last cheap moment to ask whether handover is assembling itself.

Practical completion is when the project wants to call the building finished. The same four objects are still the objects, but the labour that could have closed the holes has gone. A review starting then is archaeology: serials typed from nameplate photographs, ITPs reconstructed from a folder named Close-out, O and M PDFs dumped into a ZIP, an issued set that cannot say which revision the tester used.

The gate does not ask whether handover is finished. It asks whether the objects exist, whether the remaining work is named, and whether the people who can still repair the record are on the job. Fail it at sixty per cent and you have time. Fail it at practical completion and you have a meeting.

Sixty per cent is a proportion of the works, not a date in a calendar. Plenty of jobs have no honest per cent-complete; on those, use the same test against one system already being installed. The information problem does not wait for a programme to call itself sixty.

Asset tags — the identity operations will inherit

Operations will not query a procurement code. They will query a tag, a location, and a system. Those identities have to exist before commissioning writes a row against them, and before an O and M draft can point at anything.

For every asset already on the chosen system — installed, or delivered and sitting in the store — the review must be able to show four facts from the system of record, not from memory:

the tag as installed, or the reserved identity for plant that has not yet gone in
the location an operator will use, not only the procurement zone
the system the asset belongs to
the drawing identity the tag is supposed to match

A tag on the unit that is not in a register row is folklore. A row that is not on the unit is a spreadsheet. A tag that does not match the title in the issued set gives you two identities for one machine, and that is how an O and M becomes fiction.

Uninstalled but procured plant still needs a reserved identity. If the review counts only what is bolted down, the rest of the register will be invented in the last fortnight. Write the reserved tags and write the holes. A blank looks like a future that will take care of itself, and it never is.

If you cannot list the tags on the live system without opening a photograph folder or emailing a vendor, you have a promise of a register rather than a register.

ITPs — quality records already attached to those tags

An inspection and test plan that is not tied to an asset tag is a generic checklist. At sixty per cent, some hold points have already passed. Those records are either contemporaneous and attached, or they are folklore already.

For the live system, reconstruct five facts from the quality record, not from a progress meeting:

which ITP, and which hold or witness point
which asset tag it was run against — not “AHUs,” not “the package”
date, result, and any outstanding item still open
tester and witness as names that will still mean something after demobilisation
the title and revision used, and whether that revision is the one in the issued set

A passed hold point with no tag is a story about a room. A failed first pass that vanished once the eventual pass arrived is a hole operations will rediscover, usually at the worst moment. A sheet citing “see drawing 123” with no revision is the same failure as a health and safety file that cannot follow a supersede.

Future hold points must already name the tag they will attach to. If the remaining ITP is still written against a trade and a floor, the last stretch of inspections will not assemble into a register. It will assemble into a pile.

Quality records are handover records, and the readiness review tests that while the inspector is still on site. The longer argument sits in quality records are handover records. The question here is narrower: does the ITP already meet a tag?

O and M drafts — refuseable while vendors are still contracted

O and M is not a completion deliverable that arrives as four hundred PDFs. At sixty per cent, plant has been selected and some of it is installed, so first vendor drafts should exist in a form you can refuse.

For each tagged asset on the live system that already has a manufacturer and a model, ask:

Is there a draft — not a promise of one, not a vendor email saying “at handover”?
Does it point at the same tag and location the register uses?
Can someone who was never on the job find the identity, the maintenance action, and the related title without opening a ZIP?
If the draft is wrong, can you still refuse it under the purchase order, or has the vendor already demobilised?

A draft you cannot refuse is not a draft. It is the eventual pack arriving early enough to look industrious and too late to correct. Completeness of a folder named O+M_DRAFT is not readiness. Readiness is a queryable row, or a structured pack, already using the project’s tags.

Gaps are findings. “Vendor will supply at practical completion” is a finding. Write the title you are missing and the order it sits on. Labour spent chasing a draft at sixty per cent is cheap; labour spent reconstructing a manual after the vendor’s account is closed is a variation wearing a filename.

Nobody is asking you to finish the O and M at this stage. The question is whether the object has started, on the same identities as the tags and the ITPs.

The issued set — the titles the other three will cite

The issued set is the bounded list of current, authorised containers for the package and the purpose — not “everything in a folder named Published,” but the list a competent person could hand over today and still stand behind next month.

Asset tags, ITPs and O and M drafts all cite drawings. If two revisions of the same title look current, every later check is poisoned. If the tester used C01 and the register shows C02, the quality record is already a dispute.

For the live package, each title must carry five facts:

identity — number plus revision, not a filename hint
state — published for the stated use
suitability — what the recipient may do with it, on the agreed list
supersede — which previous identity is dead
date — when this identity became the one the site may rely on

Confirm there is exactly one current revision per title. Confirm you can produce the set as it stood last month, not only as it stands this morning. Confirm the ITP and the O and M draft cite that current identity, or a dated older identity that has been archived, rather than a sheet number with no revision.

The unofficial set in the trailer is the alternative issued set. If the readiness review cannot trust the official one, operations will inherit both, and the two will disagree.

The score sheet

Print this. One row per object. The document controller and the quality lead fill it in together, and the project manager does not edit it afterwards. Forty-five to ninety minutes, on the live system, every time the works can honestly be called about sixty per cent — or, where the programme will not say so, once on each major system as it comes out of the ground.

Object	This review’s sample	Pass
Asset tags	Every asset installed or delivered on the live system	Tag, location, system, and matching drawing identity; reserved tags for plant not yet in
ITPs	Hold and witness points already run, plus the next points still open	Each point tied to a tag; date, result, people, and the revision used
O and M drafts	Each tagged asset that already has a manufacturer and model	A refuseable draft pointing at the same tag; gaps named against an order
Issued set	Every title the tags, ITPs and drafts cite	One current revision; previous marked; set reconstructable on a named date last month

A pass is boring. A fail is a tag, an ITP identity, a vendor title or a sheet number. Do not average the four rows into a percentage; that invents a precision the hour does not have. Carry fails forward until they close, because a review that resets when somebody reprints the pack is theatre.

Giving the review an owner

The mechanics are easy. What is hard is that the readiness review has no natural owner. The document controller has no authority over vendors, the quality lead has none over the common data environment, and the project manager is measured against a programme with no row for any of this. Every hour spent on it looks, from a distance, like an hour not spent building.

So put it in the programme. A named gate, on a date, run by the document controller and the quality lead together, with the project manager obliged to accept the findings rather than soften them. Half a day per major system, repeated as systems come out of the ground, and the fails carried on a list that survives between reviews.

The client’s side is the part that actually bites, and it has to happen at tender rather than on site. Ask for the four objects as appointment requirements, not closeout deliverables: a tagging convention agreed before the first plant order, ITPs written against tags rather than trades, first vendor drafts due while the purchase order still has teeth, and a published status-code list from day one. Some of those will be priced by the contractor, and it is more honest to accept that than to pretend the record is free. It still costs less than one manual rebuilt after the vendor’s account is closed, and far less than an operations team running a building on identities nobody can trace.

If you have inherited a job already past sixty per cent, run the review anyway, on the next system to come out of the ground. A late gate is still worth more than the meeting it would otherwise replace.