Ask a team a simple question: over the last three months, by what routes did someone outside the team cause work to be done? Not the routes they should have used, the routes they actually used. The answer is always longer than the process document. There is the formal change request or variation form, which everyone can describe. Then there is the meeting where the owner says something would be useful and three people write it down as a decision. The email that ends “can you just…”. The phone call from a long-standing customer contact who asks the person doing the work directly, because it is quicker. And the largest channel of all: the team noticing something needed doing and simply doing it.
Only one of those routes produces a record and a price. The others produce neither. Work flows down the cheapest available route, as it does in any system where one door costs something and the others cost nothing, and the formal change process reports faithfully on the only route that anyone pays to use.
This article explains how work enters a business without anyone deciding to take it on, why the cost of considering requests is usually invisible, why goodwill is both a strength and the largest unpriced channel, and how a small business can map its intake routes and make them visible without becoming bureaucratic. It is general information.
Three kinds of unauthorised change
A useful definition from project cost management describes scope creep as unauthorised changes to scope that enter through verbal instructions, email instructions, or written instructions issued without anyone realising how big the change was. Those are three distinct channels:
- Verbal instructions leave no record at all.
- Emails leave a record that nobody classifies as a change.
- Properly issued written instructions whose size nobody recognised are the subtle one. Everyone followed the process, and the change still slipped through, because no one stopped to estimate it.
To these, add two more that formal change registers are not designed to catch:
- Changes to approach rather than to what is delivered: a different method, a different testing approach, a different sequence. Nothing new is added to the deliverable list, yet the effort changes.
- Discretionary effort: the team’s willingness to do whatever it takes, spending time on support and extras nobody foresaw. It is exactly the behaviour every owner says they want, and it is unpriced by definition.
Estimate before you agree, even when the request is small
The third channel, properly issued requests whose size nobody recognised, is closed by one habit: every request gets a quick estimate before anyone agrees to it, however small it seems. The estimate does not need to be precise. “About two hours, no effect on the finish date” or “about two days, and it pushes the final drawings back a week” is enough for the requester to decide whether the change is worth it. Small requests are where this matters most, because they are individually too small to question and collectively large enough to sink a fixed-price job. The variations article covers recording and pricing changes once they are agreed.
A related habit helps with changes to approach. When the team decides to change how it is doing something, such as a different method, extra checking or a new tool, ask the same question: what will this cost in time, and who agreed to it? Changes to approach are often good ideas. They should still be decisions.
Why the formal controls miss most of it
Most businesses have sensible rules for the formal channel: who may request a change, who may approve it, what cannot be changed and who must be told. Every one of those rules governs the front door. None governs the side doors, and none touches the team doing work because it needed doing.
Unmanaged change is expensive for specific reasons. Rework tends to arrive when the job is already under time pressure, so it is done alongside other work rather than in sequence. Planning and doing get mixed up, because the change is planned while it is being built. The budget is hit twice: once for the extra work and again for the extra planning and coordination it needs. And the job simply gets bigger, which increases all the management effort around it.
The cost of considering
Here is a question that exposes the structure: does your business have any time or budget set aside for assessing requests, whether or not they are approved? Almost none do. Assessment is absorbed by the people doing the work, from their own time, at no cost to whoever asked. Three consequences follow:
- Requests become a measure of price, not need. If asking is free, the number of requests tells you how cheap it is to ask, not how important the work is. A change log with two hundred entries and eleven approvals is not evidence of discipline; it is evidence that two hundred requests cost the requesters nothing and consumed someone else’s capacity.
- Saying no becomes expensive for the team and free for the requester. The team learns that the quickest response to a marginal request is to absorb it quietly, because assessing and declining it takes longer than doing it.
- The cost disappears into “running late”. A team spending a sixth of its time assessing requests that were never approved looks like a team that is behind schedule.
A small, named allowance for assessing requests, kept separate from delivery work, changes this. The cost of considering becomes visible, and declining a request becomes affordable because the assessment that supports the refusal is funded.
Silence in the scope cuts both ways
One common contract rule says anything not shown in the agreed scope is out of scope, including implied activities. That is a clean rule, and it is what a contract administrator will enforce. But customers often read silence differently: as “not yet detailed”. Both views exist at once. A business that relies on the first while its customers operate on the second may win the argument and lose the relationship. The scope that says what is out article covers stating exclusions so the two views meet before work starts.
The same pattern inside the business
The pattern is not only about customers. An internal maintenance or engineering team often has a request form, and also a production manager who walks over and asks. If the team’s time is not priced to anyone, nothing in the system expresses a preference between competing requests. The team is chronically late on the work that was formally approved and chronically praised by the people it helped informally. The fix is not more control over the formal channel, which is already the best-behaved one.
A channel audit
The most useful step takes about an hour, done by the team rather than by management:
- List every route by which new work can reach the team: forms, emails, calls, meetings, site conversations, messages, the team’s own initiative.
- For each route, answer three questions. Does it produce a record? Does it produce a price or an estimate? Who is allowed to decide whether the work goes ahead?
- Look at the last ten additions to any job. Which routes did they come through?
Then apply three tests:
- The record test. If more than half of recent additions arrived through routes with no record, the change log is a sample, not a record, and any reporting built on it understates change.
- The free-work test. What share of the week goes to work nobody asked for in writing? Ask the team; no system knows.
- The price test. What does it cost a requester to ask? If the answer is nothing, request volume says nothing about priority.
The third question in step 2, who may decide, often has no answer for the routes that matter most, because closing those routes means telling a senior person or a valued customer that a habit of theirs now goes through a process. That is a decision for the owner. Without it, the audit describes the problem but does not change it.
Make goodwill visible, not forbidden
Discretionary effort is a genuine strength. Teams that notice what needs doing and do it are valuable, and customers notice. The aim is not to stop it, but to make it visible: ask people to record unrequested work briefly, so the business can see how much of its capacity is being allocated by goodwill rather than by decision. Sometimes the answer will be to price it into future quotes; sometimes to keep doing it deliberately as part of the service; sometimes to stop.
A worked example
This is an illustration. A small design and drafting business takes on a fixed-fee job for a manufacturer: a set of production drawings budgeted at 200 hours. The formal variation process exists and is used twice. Meanwhile:
- the customer’s engineer emails the drafter directly most weeks with “one more view” or “a quick revision”;
- a weekly call adds small checks and an extra assembly variant;
- the team, noticing the customer’s own drawings contain errors, quietly fixes them.
None of the requests seems large enough to raise. The job finishes at 260 hours, 30% over budget, and the post-job review records “scope creep” as if it were weather.
For the next job, the business runs the channel audit with the drafting team. It finds five routes: the variation form, direct emails to drafters, the weekly call, phone calls to the owner and the team’s own initiative. Only the first produces a record and a price. The owner makes three changes:
- One route for requests. Customer requests go to the project lead, who logs each one and gives a quick estimate before work starts. Drafters are encouraged to forward direct requests rather than act on them.
- An assessment allowance. About 5% of each job’s hours is set aside for assessing requests, so estimating and declining are funded rather than squeezed.
- Visible goodwill. The team logs unrequested work in a single line per item. On the next job, the log shows that correcting customer drawing errors takes about 15 hours. The business discusses it with the customer, who agrees to pay for a separate checking service.
The next fixed-fee job finishes within 5% of budget, and the customer relationship is better, not worse, because requests are answered with clear estimates rather than silent absorption or late invoices.
How this applies to a small Australian business
- List every route by which work reaches your team.
- Funnel customer requests through one person who logs and estimates them.
- Set aside a small allowance for assessing requests.
- Ask the team to record unrequested work in one line per item.
- Decide what to do with goodwill: price it, keep it deliberately or stop it.
- Agree a route for internal requests too, with someone who can choose between them.
- State exclusions clearly in quotes and contracts.
- Review the last ten additions to a job and where they came from.
Signals worth watching
- Fixed-price jobs regularly overrunning with no recorded variations.
- Customers contacting individual team members directly with requests.
- A long change log with few approvals.
- Teams praised by internal customers but late on approved work.
- Nobody able to say what share of time goes on unrequested work.
- The word “just” in requests: “can you just…”.
Common mistakes
- Tightening the formal process while the informal routes stay open.
- Treating scope creep as bad luck.
- Letting assessment eat delivery time without recognition.
- Discouraging goodwill rather than making it visible.
- Enforcing “silence means excluded” without telling customers what is excluded.
- Leaving no one with authority to choose between competing requests.
Frequently asked questions
Won’t customers dislike being redirected? Usually not, if requests are answered quickly with a clear estimate. What damages relationships is silent absorption followed by an unexpected invoice, or an argument at the end.
How large should the assessment allowance be? Small. A few per cent of a job’s hours is typical. The point is to make the cost visible and give the team room to say no properly.
Should we stop staff helping customers informally? No. Ask them to record it, then decide deliberately whether to price it, keep it or stop it.
What about requests from the owner? They count too. The owner’s casual “it would be good if” can be the biggest unrecorded channel in a small business.
How often should we do the channel audit? Once to start, then whenever jobs repeatedly overrun without recorded changes.
Does this apply to retainer or hourly work? Yes, though the cost lands differently. On hourly work the customer pays, so the risk is surprise invoices and disputes; a quick estimate before each request protects the relationship just as it protects margin on fixed-price work.
Questions to ask
- By what routes does work actually reach our team?
- Which of those routes produce a record and an estimate?
- What does it cost someone to ask us for more?
- How much of our time goes on work nobody asked for in writing?
- Who decides between competing requests?
- What did our last overrun job’s additions have in common?
Bringing it together
Work enters a business through more routes than its processes describe, and it flows down whichever route is free. Map every route, funnel requests through someone who records and estimates them, and fund the cost of considering so that saying no is affordable. Make goodwill visible rather than forbidden, and decide deliberately what to do with it. The aim is not to stop change or helpfulness, but to make sure the business knows what it has agreed to do and what that agreement costs.
Source: KEVOS notes, drawing on A. Khan (2006) and N. Lavingia on scope creep in Cost Engineering, H. Robbers (2009) on scope and impact analysis, and teaching material on scope control. Examples and figures in this article are illustrations. This article is general information.
