Blog
7 min read
A client catches you on the way past. "While you're in there, can we move that socket to the other side of the door?" You say yes. Two minutes of work, and refusing over two minutes feels petty. The sparky does it after lunch, nobody writes anything down, and that socket is never mentioned again until someone queries the final account.
A variation agreed on site is a change to the priced scope, asked for in conversation and acted on before anyone has costed it. Knowing how to record variations agreed on site is not a filing problem. It is a timing one. The ninety seconds between the ask and the first cut of the tile is where the money goes.
Any instruction that changes what you priced, given verbally, and acted on before a cost has been agreed. Size is irrelevant. A moved socket qualifies. So does a 100mm wider opening, because somebody has to re-cut a lintel and reorder a door.
These are the asks that arrive in doorways:
None of it is unreasonable. Clients are not trying to get free work out of you. They genuinely do not know that a socket carries a price, and it is your job to tell them before the sparky lifts a chase.
Not into one big miss. Into five or six small ones that nobody could point at individually.
ProBuilders Network reckons the average project loses between £3,000 and £15,000 in unbilled variations, and puts the share of construction firms seeing reduced profit from client changes at over 65%. That is not a rounding error on a domestic extension. It is the margin.
The frequency is the part people underestimate. The Association of Professional Builders puts it bluntly: there is a 97% chance you are already dealing with at least one variation on every build. So this is not an edge case to handle as and when. It is a standing feature of every job you run.
Which earns it a habit rather than an improvisation. The same problem shows up further up the chain in design, where charging for design changes after client sign-off is the same argument in a different room.
Three steps, in that order, all of them finished before the trade moves.
| Step | When | What it sounds like |
|---|---|---|
| Pause | The second the ask lands | "Happy to do that. Give me till this afternoon to price it." |
| Price | Before anyone starts work | "It's about £450 and it adds half a day." |
| Confirm | Same day, in writing | A text listing the change, the price and the effect on the programme |
The order matters more than the polish. Pausing is what buys you the right to charge, because a trade that has already started has priced the job at zero. Pricing on site, even loosely, is what keeps the conversation commercial instead of emotional. Confirming in writing is what turns a conversation into an invoice line.
The Federation of Master Builders sets out the same sequence in its guidance on managing variations without losing work. Confirm what has been added or removed. Give a fixed cost rather than an estimate, then wait for written sign-off before anyone proceeds. Short advice. It is also the only part of change control that most jobs actually need.
A text is enough. Genuinely.
Hi Sarah. Confirming the socket move to the far wall of bedroom two, as discussed this morning. £185, no change to the programme. Reply to confirm and I'll get Dan on it tomorrow.
Three lines. What changed, what it costs, what it does to the programme. Send it from site before you drive off, because the version you write at nine that evening will be vaguer and half of them never get written at all.
Say yes to the work and no to starting it. That is the whole move, and it takes one sentence.
Builders hesitate here because a pause feels like an obstruction. It is not. Say it a few times and it reads as the opposite. A firm that prices every change reads as organised, and a firm that only starts pricing changes in month three reads as though something has gone wrong.
Lines worth having ready:
That last one earns more than it costs. Confirming the free changes alongside the priced ones stops the written note reading as an invoice ambush, because the client has already watched you use it on something that cost them nothing.
No. A site diary and a variation record do different jobs, and running one does not cover you for the other.
The diary is a contemporaneous record: dated documentation of what happened on site. You want it when a job ends in a dispute and someone is reconstructing a timeline months later. That is defensive, and worth doing. But a diary entry saying "client asked to move socket" is not an agreement. It is your word, written by you, with no price on it and no reply from the client.
Recording a variation is a commercial act rather than an evidential one, because it fixes a number and gets a response. The diary proves the conversation happened. Only the confirmation proves what it was worth.
When the change is big enough to move the programme, or when it changes what a later trade has to do.
Formal change-order paperwork is not the enemy. It is just too slow for a five-minute ask in a doorway, and that mismatch is why so many small variations end up recorded nowhere at all. So run two speeds. Anything under a threshold you set gets the text-message treatment. Anything above it gets a proper priced instruction with a revised programme, and the same change request process used on fixed-price work applies.
Pick the threshold yourself and hold to it. A workable shape looks like this:
Whatever number you choose, price the disruption as well as the work. A two-minute job that pulls a trade off sequence and sends someone back to the merchant is not a two-minute job. Start at 10 to 25% on top of the labour element for the interruption, then tune it once you have a few jobs of evidence behind you.
That evidence is the bit worth building. If you already know what your last three kitchens actually cost against what you quoted, a variation price takes thirty seconds instead of a week of avoidance. Our guide to tracking actual costs against a fixed price statement of work covers the same discipline in a different trade, and the mechanism transfers.
Korrel handles the far end of this. It prices a change against the original estimate, so you can see what the change does to revenue and margin before you approve it. The capture problem comes first though. No software fixes a conversation nobody wrote down. That is how a variation agreed on site stops being a favour.
Not a policy. Not a new form. The next ask, on the next job.
When it lands, say the sentence, price it before anyone starts, and send the text before you leave site. Do that for one job from first fix to handover, then compare what you invoiced against what you originally quoted. One job is not proof of anything. It settles the argument in your own head about whether the ninety seconds is worth it.
COMMON QUESTIONS:
RELATED READING:
Unscoped integration work is any interface the signed scope never named. Here is how to count them all before signature, and what to do the day an undocumented one appears.
A fixed price SOW gives you one number and no invoice line to check yourself against. Convert it into a cost ledger, then reconcile actuals every week.
Stage 3 closes, the client approves, and three weeks later they want the roof pitch changed. Here is how to tell a revision from a new instruction, and what to charge.
Korrel turns briefs into structured proposals, then tracks what each job actually costs.
START FREE TRIAL