How to define deliverables so a client cannot expand them
7 min read
Open the last statement of work you sent. The deliverables are nouns.
A requirements document. A discovery workshop. A target operating model. Each one reads like a finished object, and not one of them says how big it is.
That silence is where the fee goes. Learning how to define deliverables so a client cannot expand them is not a matter of tougher legal wording. It is a matter of arithmetic. A deliverable becomes finite the moment your proposal names how many of something it contains. And what has to be true before it counts as delivered.
Consultancies feel this more sharply than most. You sell days. A fixed-scope engagement that quietly absorbs two extra stakeholder interviews and a third draft pays for them out of margin. Never out of a change order.
Why does a deliverable written as a noun invite expansion?
Because a noun has no edges.
"A requirements document" does not say fourteen pages or ninety. It does not say one round of review or five. It says nothing about whether the six people you planned to interview become the sixteen that the client's programme director would like included. Each of those blanks gets filled in later, by whoever asks first.
The guidance that ranks for scoping repeats the pattern. ProjectManager's guide to the project scope statement tells you to list the deliverables your team needs to produce. Its examples are manuals, marketing materials and a prototype. All nouns. Acceptance criteria get a mention, but no method for writing them.
The result is well documented. PMI's 2018 Pulse of the Profession found that 52% of projects experienced scope creep in the preceding twelve months, up from 43% five years earlier. The direction of travel is the wrong one.
Blame lands on the client. Wrongly. Systemology's account of scope creep in service businesses puts the cause plainly: you hold one picture of what you are delivering, and the client holds a different one. The noun permitted both readings. Nobody noticed until week four.
Run the test on your own last engagement letter. Take the largest item on it and ask three questions. How many of these are you producing, who decides the item is finished, and how long do they get to decide. If the letter answers none of the three, the scope was never fixed. It was only described.
How do you define deliverables so a client cannot expand them?
Give every deliverable a quantity and a test. The quantity caps how much of it exists. The test says when it is finished.
Those two additions turn a noun into a bounded unit of work. "Stakeholder interviews" becomes "twelve stakeholder interviews, scheduled by the client, completed within four weeks of kick-off". The thirteenth interview is now visible. Not a favour, and not a misunderstanding. A priced addition that both sides can see before anyone books a diary slot.
Nor is the count a guess. It comes from the last three engagements of the same shape, which is the only evidence a client will not argue with. If your operating model reviews have taken eleven interviews, then thirteen, then twelve, the number twelve is defensible. If nobody has ever counted, the first bounded proposal you write will be wrong. It will still beat a noun.
Tight scoping is a margin instrument before it is a legal one. The same logic sits underneath the margin targets in designing profitable projects for creative agencies. It holds for a consultancy selling days as firmly as it does for a studio selling weeks.
Which numbers belong in the deliverable line?
The ones a client would otherwise pick for you. Count of sessions. Count of attendees. Count of drafts.
Then two more. Count of reviewers, because a document read by nine people generates nine sets of comments and a reconciliation job nobody quoted. And the length of the review window, because an open-ended comment period is a deliverable that never closes.
Five numbers. Here is the substitution, across the deliverables a consultancy sells week in, week out.
| Open noun | What is left unbounded | Bounded version |
|---|---|---|
| Discovery workshops | How many, how long, who attends | Three half-day workshops, eight attendees each |
| Stakeholder interviews | How many people, how many passes | Twelve interviews, one pass, scheduled by the client |
| Requirements document | Length, depth, revisions | One document, two review cycles, one named owner |
| Target operating model | Number of options, level of detail | Two options at level 3 process detail, one recommendation |
| Board readout | How many audiences, how many decks | One readout, one audience, one deck |
Read the right-hand column aloud. None of it is aggressive. It is specific, which is a different thing. Specificity is what the client's own finance team wants from you anyway.
What has to be true before a deliverable counts as done?
Acceptance criteria answer that, in one line per deliverable, written so a reviewer can mark pass or fail without argument.
NMS Consulting's statement of work template puts it directly: a deliverables list is only useful if each item has acceptance criteria that can be checked quickly and consistently. The template asks for a named file format, a named approver and a stated review window. Those three fields close the gaps a noun leaves open.
A workable acceptance line has four parts:
- Form. A 30-slide deck in the shared folder, not "documentation".
- Approver. One named person, not "the client".
- Window. Five business days to comment, after which the item stands accepted.
- Basis. Built from the twelve interviews listed above, not from anything raised afterwards.
None of this needs a lawyer. It needs a number and a name.
Of the four, the window earns its place hardest. Without one, a deliverable sits in review for six weeks while the client's own priorities move, and your team holds capacity for a job that has quietly stalled. Five days to comment, after which the item stands accepted, is not a hostile clause. It is the clause that lets you plan next month.
The three places a consultancy leaves the door open
Start with the review round. A round is itself a deliverable, so it deserves the same treatment as anything else you have quoted. The companion piece on how to charge for extra revision rounds handles the pricing question. The definition question has the same shape. State how many rounds are included, who consolidates the comments into one set, and what formally closes a round, so the fourth set of comments arrives as a quote rather than as an assumption.
Next comes the attendee count. A workshop without a cap becomes a room that expands every fortnight. Inviting one more person costs the client nothing, and costs you a re-run of the session you already delivered. Two extra directors change what the session is, as well. A facilitated decision becomes a briefing.
Third is the walkthrough: somebody asks you to talk the board through the findings, and the request arrives dressed as a courtesy rather than as a sixth session nobody quoted. Define the number of readouts in the deliverable itself. One audience, one deck, one date. Anything beyond that is quoted separately, at the rate you already published in the proposal.
All three are proposal problems rather than delivery problems, which is the premise behind the other posts on proposals here. By the time a job is running, the argument you are having was decided months earlier, in a sentence somebody wrote quickly.
Does bounding a deliverable make a consultancy look inflexible?
No. It reads as competence, as long as the bound arrives with a price for going past it.
The version that annoys clients is the bound with no exit: twelve interviews, full stop. The version that works names the rate for the thirteenth. You are not refusing the extra work. You are telling the client what it costs before they commit to it, which is the same courtesy they extend to their own board.
Say it out loud in the scoping meeting, before it reaches paper. Three workshops, twelve interviews, two review cycles, and here is the day rate if you want more of any of them. That is an easy conversation. The hard one is an invoice for work the client assumed was included.
There is an operational half to this too. A bound you cannot see being breached is decoration. Korrel builds proposals around deliverables, risk flags and milestone billing, then tracks time and cost against those estimates as the job runs.
The point of that pairing is timing. A fourth draft shows up as a number in week six, while you can still price it. Not as a surprise at the final account.
Rewrite one deliverable this week
Take the engagement letter sitting in your drafts folder. Find the vaguest noun in it. Add a count, a named approver and a review window, then send it as you normally would.
What happens next tells you something either way. The bound does the same job for the client that it does for you. It says what has been bought, in numbers both sides can check.
COMMON QUESTIONS:
- What is an acceptance criterion for a consulting deliverable?
- An acceptance criterion is a one-line test a reviewer can mark pass or fail without debate. In practice it names four things: the form the item takes, the person who approves it, the window they have to comment in, and the inputs it was built from. Anything vaguer than that gets argued over at sign-off, usually by someone who was not in the scoping conversation.
- How many stakeholder interviews should a fixed-scope engagement include?
- Whatever number you can evidence from the brief, stated as a hard count in the proposal. The figure matters far less than the fact that it is written down. Name the rate for interviews beyond the count at the same time, so the thirteenth is a purchase rather than a dispute.
- Does adding quantity limits to a proposal put clients off?
- A specific count rarely puts clients off, because it tells them exactly what they are buying. That is easier to approve than a vague promise. What does put clients off is a limit with no route past it, so pair every count with the price of exceeding it. A bound plus a rate reads as a menu. A bound on its own reads as a refusal.
- What is the difference between a deliverable and a milestone?
- A deliverable is the thing you hand over. A milestone is a point in the programme, usually tied to payment, that a deliverable triggers once it is accepted. Bounding the deliverable matters more, because a milestone inherits whatever ambiguity the deliverable carries and simply delays the argument until the final account.
RELATED READING:
How to charge for extra revision rounds
A revision round becomes billable the moment it stops matching the proposal. Define a round properly, then price the extra ones from what delivering one costs.
Charging for design changes after client sign-off on a RIBA stage
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.
A change request process for fixed price implementation projects
Enterprise change control was built for internal IT operations. A delivery partner working to a signed statement of work needs something lighter: raise it, size it, price it, book it.
Stop guessing. Start profiting.
Korrel turns briefs into structured proposals, then tracks what each job actually costs.
START FREE TRIAL