When the Client Wants One Number
You give a range because the spread genuinely widened. The client says: I understand, but I need a number for the budget.
They are not being difficult. They have a finance process, or a boss, or a decision to make between you and someone else, and a range does not fit into any of those. Refusing to give a figure is not integrity; it is passing your problem to them. When this estimation problem needs an operational counterpart, this practical guide offers a useful reference point.
There is a way to give one honestly.
Why they want it
Worth understanding, because it decides which answer helps.
To decide whether to proceed. They need to know the order of magnitude, not the exact figure.
To get approval. Somebody else signs, and that person wants one line. PMI provides a broader project-management reference for planning and uncertainty See Project Management Institute.
To compare quotes. Your range against someone's confident single number loses every time, even when your range is the honest document.
To plan cash. They need a worst case they can survive.
Only the last genuinely needs precision, and it needs the top of your range rather than the middle.
The four honest ways to give one number
A capped price. It will be no more than X. You quote near the top of your range, carry the variance, and the client gets certainty. The cleanest answer when you can size the risk. If your range is 20–35, quote 35 and mean it.
A number with a named trigger. X, and here is the one thing that would change it: if the existing auth turns out to be undocumented, add Y. Clients accept this far more readily than a range, because it is a number plus a specific condition rather than vagueness.
A staged number. Phase one is X, fixed. After it we will both know enough to price phase two, and my current expectation is around Y. Best where the uncertainty is genuinely front-loaded, which is most projects.
A number for a reduced scope. At X I can do these three things. The fourth is where the uncertainty is; let us do it separately. This is often the answer the client actually wanted.
All four give a figure. None requires pretending to a precision you do not have.
What not to do
Do not quote the middle of your range. The middle is where you lose money — you have taken on the downside without pricing it. If you are giving one number, it lives near the top or it is capped.
Do not give a number you cannot defend. If the honest answer is that you cannot size this yet, say so and propose a paid discovery step. Small, fixed, with a deliverable.
Do not compete on the number against someone who is guessing. You will win the job and lose the money. If a competitor's confident figure is far below your range, one of you has misunderstood the work, and it is worth finding out which before you agree to match it.
Do not hide the assumptions. A number with three assumptions written under it is still one number, and it is the version that survives contact with reality.
The sentence that does most of the work
I can give you a firm number if we agree the scope precisely — or a range if you want flexibility on scope. Which is more useful to you?
It reframes the question accurately: certainty about price and flexibility about scope are the two things you cannot have together, and the choice belongs to the client. Most will pick certainty, which means they will help you nail down the scope, which is what you needed anyway.
When to refuse
Rare, and it exists.
If the work is genuinely unsizeable — an unfamiliar legacy system, an integration with something undocumented, a problem nobody has diagnosed — a firm number is not honest and a capped one may be ruinous.
Then: I can't price this responsibly yet, and a number I invent now would be wrong in a way that costs one of us. Let me spend two days finding out, fixed price, and then I can quote properly.
Clients accept this from people who otherwise give firm numbers. It is only unconvincing from someone who hedges everything, which is the argument for being able to size most work in the first place.
The short version
- Refusing to give a figure passes your problem to the client, who has a process to feed
- Four honest single numbers: a cap, a number with a named trigger, a staged number, or a number for reduced scope
- Never quote the middle of your range — that is where you take the downside without pricing it
- A number with written assumptions is still one number, and it is the one that survives
- Offer the real trade: certainty on price or flexibility on scope, their choice
- Refuse only when the work is genuinely unsizeable, and propose a paid, fixed discovery step instead