Variation Disputes: The Evidence That Wins Valuations

The valuation meeting is twenty minutes old when the main contractor's quantity surveyor reaches your variation account and says the sentence everyone in the room has heard before: "I can't see the instruction for that." The work is built. Your team remembers the Tuesday it was ordered, the supervisor who took the call, the two days it cost. What nobody can put on the table, in the meeting, is the record.

What's Actually Being Asked

The question is not whether you did the work. It is narrower and harder than that: can you show what changed, who authorised it, when you told them, and what it actually cost you. Everything else in a variation dispute is argument about those four facts.

The standard UK forms, and the bespoke subcontracts cut from them, route change differently. JCT contracts work through instructions and variations valued under the contract's own rules. NEC contracts work through compensation events, notified and quoted for. What they share is a structure: change is valued on the basis of what was notified and recorded at the time, within periods the contract sets, and some of those periods are applied strictly. Underneath sits the payment regime of the Housing Grants, Construction and Regeneration Act 1996, as amended: for the contracts it catches, payment notices, pay less notices, and a statutory right to refer a dispute arising under the contract to adjudication at any time. The application itself is a creature of your contract, not the Act.

Adjudication is where the evidence question becomes sharp. It runs to a short timetable, with a decision usually due within 28 days of referral, and it is decided largely on documents. There is no disclosure exercise, no months of witness preparation, no time to go and find the thing you should have kept. What you can put in the referral is, in practice, what you have.

It is worth being clear-eyed about the asymmetry. The party you are arguing with is usually the one running the document control, chairing the meetings and issuing the minutes. Their version of events is written down by default. Yours is written down only if you decided, months earlier and on a busy day, that it was worth the effort.

The Evidence That Answers It

A variation that survives scrutiny is usually not one record. It is a short chain of them, each pointing at the same event:

• The instruction. The written instruction, or the written confirmation of a verbal one. Standard forms differ sharply here: some give you a route for confirming an instruction given other than in writing, within a short period; others require every instruction to be in a recordable form in the first place, which means a verbal one is not an instruction at all. Either way, the confirmation sent the same week is worth more than the best recollection offered a year later.
• The notice. Evidence that you notified the change and its likely effect on time and cost, in the form and within the period the contract requires. Dated, sent, and provably received.
• The as-found condition. Dated, located photographs of what you encountered before the change, and of the work in progress. This is the record that answers "it was always like that".
• The site record. Who was on site, on which day, on what activity, for how long. Daily records, allocation sheets and labour returns are the backbone of a disrupted-works argument, and the easiest thing to let go thin when a job gets busy.
• The resource cost. Labour, plant, materials and sublet cost attributed to the change, not smeared across the package. If the cost cannot be separated from business as usual, it will be valued as business as usual.
• The programme effect. A baseline you can still produce, and progress records against it, so the displacement can be shown rather than asserted.
• The approvals and inspections. Who accepted the changed work, when, and which QA or ITP record closed it out. This is what converts "we say it is done" into "it is signed off".

Why the Spreadsheet Version Fails

Variations are not usually lost through carelessness on the day. They are lost because the record was assembled after the dispute started, from sources that were never built to be evidence. The instruction is in a message thread on a supervisor's personal phone. The as-found photographs are in a folder named by date, with nothing tying them to a level or a grid reference. The daily diary is diligent to week twelve and sparse after that, which is exactly when the job went wrong. The cost is a column in a spreadsheet that somebody rebuilt from memory in March for work done in October. None of it is dishonest. All of it is retrospective, and retrospective is the weakness the other side will work on: they will not try to prove the account wrong, only to show it was reconstructed. A contemporaneous record and a reconstructed one read very differently, even when they say the same thing.

Keeping It Provable

Contemporaneous is the word doing the work. A record made as the work happens carries weight that the identical fact, typed into a claim document eight months later, does not. Three properties turn a pile of records into an account somebody else can follow.

It is made at the time, and timestamped in a way nobody can quietly adjust afterwards. It is anchored to the building, to an elevation, a level, a grid line or a room, so it can be found by the thing it describes rather than by remembering who filed it. And it is linked, so the instruction, the photograph, the labour and the inspection all reference the same change, and pulling one pulls the rest.

The test of whether you have this is not how organised the files look. It is whether somebody who was not there, working only from the records, can reach the same conclusion your team reached at the time, and see when they reached it. That is a way of working before it is a system. The system only decides how much effort it costs to keep it up when the job is busy, which is the only time it matters.

The Monday Checklist

1. Take your largest open variation and try to assemble it from records alone, with nobody allowed to fill gaps from memory. Give yourself thirty minutes. What is missing is your real exposure.
2. Find where verbal instructions actually arrive on your jobs, whether that is a phone call, a site conversation or a messaging app, and put one written confirmation route in front of it.
3. Read the notice periods in the contract you are on now, not the one you remember, and put the dates somewhere you will see them.
4. Audit last month's site photographs: can each one be tied to a date and a location without a person standing there explaining it?
5. Check that daily records name the activity and the change reference, not just the headcount.
6. Agree with your commercial team who closes the loop between an instruction being issued and the record that proves it was carried out.
7. Confirm you can still produce the baseline programme you are measuring displacement against.

Where BrieXO Fits

The part of BrieXO that is live today is the FIELD bundle, and it covers the site end of this chain: daily records, labour and allocation data, photo evidence and QA checks captured as the work happens and anchored to the part of the building they describe, with decisions and approvals written into a tamper-evident ledger that exports with its history intact. The commercial modules, cost and change and payment, are on the roadmap rather than shipping. It supports your record-keeping duties under the contract; it doesn't discharge them, and nothing in it decides how a variation is valued. The export side is set out on our tamper-evident audit ledger page.

Related reading:the as-built record that survives scrutiny, what a "show me" request looks like, and where programme overruns actually start.

George Sfica is a senior design manager at a specialist façade contractor, with 23 years in manufacturing and construction, eleven of them in façades and external envelopes, spanning fabrication, installation and design management, an MSc in Façade Engineering and an IFE Level 2 Certificate in Passive Fire Protection. He is the founder of BrieXO.