Scope creep is a measurement problem before it's a process problem
Every article about scope creep tells you to implement change control. Most agencies already have change control. They still lose margin to scope creep, because by the time a change request is warranted, the creep has already happened.
Scope creep is usually detected as a process failure and should be detected as a financial one. It shows up first as hours climbing against a fixed fee, months before anyone identifies a specific out-of-scope request. Change control catches the large, nameable additions. It does not catch the accumulation of small ones, which is where most of the margin goes. The fix is measuring hours against the fee assumption continuously, not tightening the approval process.
Why doesn't change control stop scope creep?
Change control catches the requests you can name. Scope creep is mostly made of the ones you cannot.
A formal process works when a client asks for a new deliverable. Someone writes it up, prices it, and it gets approved or dropped. That mechanism is genuinely useful and every agency should have it.
But the typical unprofitable account was not destroyed by a nameable addition. It was destroyed by an extra revision round that became normal, a weekly call that became twice weekly, a stakeholder added to every review, reporting that grew a section, and a Slack channel that became a support desk. None of those is a change request. Each is individually too small to raise. Together they are the margin.
The test: if you asked your team to list the out-of-scope work on your worst account, they would name two or three things. The hours data would show something four times larger. That gap is what change control cannot see.
What it looks like in the numbers
Scope creep has a signature, and it is visible long before anyone complains.
| Quarter | Fee | Hours | Effective rate | Margin |
|---|---|---|---|---|
| Q1 | 36,000 | 204 | $176 | 48% |
| Q2 | 36,000 | 241 | $149 | 39% |
| Q3 | 36,000 | 288 | $125 | 27% |
| Q4 | 36,000 | 331 | $109 | 16% |
The fee never moved. Nobody made an unreasonable request. No single quarter looked alarming compared with the one before it. And the account went from healthy to nearly break-even inside a year.
Two derived numbers make this visible, and neither appears in a standard bookkeeping package. Effective hourly rate — fee divided by hours delivered — trends down as scope creeps, regardless of what you charge. And hours against the fee assumption: when you priced the retainer you assumed a volume of work. If nobody wrote that assumption down, you cannot detect drift from it.
How to catch it early
- Write down the hours assumption when you price. A retainer priced at $12,000 a month assumes some number of hours. Record it. Without it there is no baseline and creep is undetectable by definition.
- Track effective hourly rate per client, monthly. One number, trending. A 15% decline over two quarters is scope creep whether or not anyone can name what caused it.
- Review hours against budget weekly, not at month end. Catching an account at 60% of its monthly hours in week two is actionable. Finding out on the tenth of the following month is a post-mortem.
- Give non-billable and unscoped work a code. If extra revisions are logged to the client's main project code, they are invisible. Logged separately, they are a number you can put in front of the client.
- Report the trend to the account lead, not just to finance. The person who can have the conversation needs to see it, and they usually do not.
The conversation, once you can see it
Hours data changes the nature of the discussion entirely. Without it you are asking for more money, which is a negotiation. With it you are reporting a fact, which is a conversation.
"This engagement was scoped at 68 hours a month and has averaged 110 for the last two quarters" invites a joint decision: pay more, or take less. Most clients pick one without much drama, because the alternative — you quietly absorbing it until the relationship sours — serves neither party.
That conversation is also easier the earlier it happens. An account 15% over is a repricing. An account 80% over has become an expectation, and resetting an expectation is much harder than adjusting a fee. By the time you are considering exiting the client, several cheaper options have already been missed.
Common questions
How do you prevent scope creep at an agency?−
Write down the hours assumption behind every fee, track effective hourly rate per client monthly, review hours against budget weekly rather than at month end, and give unscoped work its own time code so it is visible. Change control is necessary but only catches nameable additions, not accumulated small ones.
How do you measure scope creep?+
Effective hourly rate — fee divided by hours actually delivered — trending over time, and hours delivered against the hours assumed when the fee was set. A sustained decline in effective rate is scope creep whether or not anyone can point to a specific request.
What causes scope creep in agency work?+
Usually accumulation rather than a single event: an extra revision round becoming normal, a call becoming twice weekly, stakeholders added to reviews, reporting growing a section, a Slack channel becoming a support desk. Each is individually too small to raise, which is exactly why change control does not catch them.
How do you have the scope conversation with a client?+
With the hours data rather than a request. "This was scoped at 68 hours a month and has averaged 110 for two quarters" is a fact that invites a joint decision to pay more or take less. Without the data it is a negotiation about money, which is harder and more adversarial.
Is scope creep the client's fault?+
Rarely deliberately. Clients ask for what they need and assume you will say if it is a problem. Agencies that do not measure delivery hours have no basis on which to say so, and silence gets read as agreement. The failure is usually in measurement rather than in either party's behaviour.
Effective rate per client, monthly
Scope creep shows up there first — months before anyone can name it.
Get a free books reviewGeneral information for agency owners and operators, not accounting, tax or legal advice. Figures shown are illustrative unless stated otherwise.