Signing Off the Small Things: Who Approves Extra Work, and What It Costs
Most project businesses have a formal variation process, and most of them work perfectly well for the changes that arrive by email with a schedule attached. The difficulty lies elsewhere. It sits in the short call where a client asks whether a small extra could be squeezed in, and someone senior enough to be helpful says yes. No form is raised. No price is discussed. The work is done well, the relationship is preserved, and the margin disappears without anyone having decided that it should.
I have seen this pattern across engineering consultancies, digital agencies, fit-out contractors and professional services firms. The mechanism is consistent, and so is the remedy. What follows is a practical structure that a small UK firm can put in place this month without new software or a change programme of its own.
Why the informal addition survives
Three conditions keep it alive. First, the person taking the call usually has the client relationship at heart and no pricing authority, so agreeing feels like service rather than cost. Second, there is often no single place where additions are recorded, which means the cumulative effect is invisible until a project's profitability is reviewed months later. Third, the client is not behaving unreasonably; they are simply asking, and being told yes.
Individually, these are minor. Collectively they represent a meaningful share of turnover on affected projects. In my experience the figure is rarely below three per cent and can be considerably higher on fixed-price work where the original brief left room for interpretation.
The important point is that none of this is a discipline failure to be met with blame. It is a design gap. The business has a process for priced changes and no process for unpriced ones, so unpriced ones go wherever there is no friction.
The one-page ledger
The instrument that closes the gap is deliberately modest: a single sheet, per project, with a fixed set of columns. It should take under two minutes to complete a line, or it will not be completed.
- Date of the request
- Client and project
- What was asked for, in the client's own words
- Who asked and who agreed
- Estimated hours and any disbursements
- Price agreed, or a recorded nil
- Approval band applied
- Date invoiced, or reason not invoiced
Two rules make it work. First, there is one named owner per project, and the ledger is theirs. Second, a line is added when the request is agreed, not when the work is finished. A ledger completed retrospectively becomes a record of losses rather than a control.
Where a firm runs many small projects, a single sheet covering all active jobs is acceptable, provided each line carries the project reference. The objective is visibility, not administration.
Three bands of authority
Thresholds give people a clear answer to the only question that matters in the moment: may I say yes? A three-band structure is usually sufficient, with figures adjusted to the size of the firm.
Band one, up to roughly £500 or half a day. The delivery lead may accept the request, log it, and raise no invoice. To prevent this becoming the default route, apply a monthly cap per client, expressed in hours. Once the cap is reached, band two applies automatically.
Band two, from £500 to around £2,500, or up to two days. The project or account manager may agree the request, but must confirm it in writing within one working day, record the price, and place it on the next invoice run. Agreement and documentation are treated as one action, not two.
Band three, above £2,500, or any request that affects the critical path, a fixed price, or another workstream. Director approval, a formal variation, and a note of the schedule consequence. Time alone is not the test here; a small addition that moves a completion date belongs in band three regardless of value.
Two supporting rules matter as much as the bands. Verbal agreement outside a person's band is not agreement. And any request that cannot be priced on the call is answered with a commitment to revert within a stated period, which is a legitimate response rather than a delay.
The weekly half hour
A fixed slot each week, same attendees, ledger open. Each line receives one of three outcomes: price it and invoice it; absorb it deliberately as goodwill and record the value so the cost is known; or decline and renegotiate the surrounding scope. Nothing is left undecided, because undecided lines are what accumulate.
The routine yields two measures worth watching. The first is the proportion of logged items that convert to priced work; a low figure suggests the bands are being used as a courtesy rather than a control. The second is the concentration of unpriced items by client, which is a useful and unemotional input to the next pricing conversation.
Where the firm has a board pack or monthly management accounts, a single line summarising unpriced additions in hours and value connects the ledger to the numbers that already command attention.
What the ledger tells you after a quarter
Three months of consistent entries produce something more valuable than a control: a diagnosis. Patterns emerge. Requests clustering around ambiguous acceptance criteria point to estimating and contracting weaknesses. Requests clustering after a client's internal reorganisation point to governance on their side. Requests for advice rather than deliverables point to a retainer conversation.
This is the material for the next tender, the next fee review and the next framework agreement. It replaces assertion with evidence.
Handling the client conversation
Clients rarely object to transparency, provided it is offered early and framed as clarity rather than as an invoice arriving by ambush. A monthly summary of what has been delivered, what has been absorbed and what will be charged is usually welcomed, and it removes the awkwardness of raising the subject after the fact.
Scope creep is not a client failing. It is the natural consequence of a relationship in which the supplier is responsive and the process is silent. Give the process a voice, and the margin has a chance of following.