An agency agrees a fortnight's work covering twelve topics. In week one the client asks whether it could also handle bookings. It is a small addition.
In week two they ask about a second location, then about a service that was not on the list, then whether it could send a confirmation email. Each request is small and each one is reasonable.
By week five the project is a third larger than what was quoted, nobody has mentioned money, and the agency is now in the awkward position of either absorbing it or raising it after saying yes four times.
The first failure: the topic list was never written down
Without an itemised list there is no boundary, so nothing is outside scope by definition. This single omission causes most of what follows, as set out in scoping a client assistant project properly.
The second failure: the first small request is free
The response to request number one sets the rule. An agency that absorbs it has taught the client that additions are free, and every subsequent request follows that precedent.
The third failure: the change process exists nowhere
If the proposal does not name a route for additional work, raising one mid project feels confrontational. Agreed in advance, it is simply how the project runs.
The fourth failure: integrations arrive as afterthoughts
Connect it to our booking system is a sentence that costs days. It is asked casually because from outside it sounds like a setting.
The fifth failure: the content grows quietly
Additional topics mean additional answers, which means additional gathering and chasing, which is the dominant cost already, as covered in pricing assistant builds for clients.
The sixth failure: new stakeholders appear
Somebody from another department sees it in week three and has requirements. Nobody scoped their involvement and nobody can reasonably tell them to go away.
The seventh failure: the approval chain lengthens
What was one approver becomes three, each with opinions, and revision rounds nobody estimated for consume the remaining margin.
The eighth failure: the tuning period has no end
Post launch adjustment slides into indefinite free maintenance because the proposal never bounded it.
The ninth failure: the agency fears the conversation
Raising cost feels like it endangers the relationship. In practice, clients respect a clear change process and are unsettled by an agency that absorbs everything silently.
The tenth failure: the requests are never recorded
Without a log of what was added, the agency cannot even show the client what happened, and the overrun looks like poor estimation.
The eleventh failure: the client does not know what they are asking for
Connect it to our system and add another location sound like configuration from outside, and the client has no way of knowing that one is an afternoon and the other is a fortnight. An agency that explains the difference at the moment of the request is educating rather than refusing.
The eleventh failure: the same dynamic appears in every service business
Small reasonable changes accumulating past the quoted scope is universal, which is the ground in teacher matching and change requests.
The twelfth failure: the quote was never detailed enough to defend
A quotation listing a deliverable rather than its contents cannot be defended against a request, which is the discipline in ai assistant service quotes.
The fix: itemise, log and price the additions
A written topic list, a named change route in the proposal, a log of every request and a short quotation for each addition turns the fortnight that became a quarter into a fortnight plus three paid changes. Start at try, see features, and the sector view sits under agencies.