How to write a proposal
Most proposals are rejected on page one, before the price. This walks one real deal through every section, shows the sentences that lose a reader, and says what to cut.
The short version
- Open with their situation in their words, never with your company history.
- Name deliverables, not activities. "A written migration plan by 14 March" beats "planning support".
- Say what is excluded. It is the cheapest dispute prevention there is.
- Give priced options rather than one number, so the question becomes which rather than whether.
- Put a validity period on the price and one clear next step at the end.
- Length follows the decision, not a house standard. Two pages for a repeat client is fine.
One deal runs through this guide: Northwind Traders. A 40-minute discovery call about a customer portal, three phases, roughly $96,000, they mentioned Net 45 and that Sam signs.
-
01Write the summary last and put it first
One page: what they asked for, what you propose, what it costs, what happens next. A decision-maker who was not on the call will read this and possibly nothing else, so it has to stand alone.
On this deal"Northwind asked for a customer portal their clients can self-serve from. We propose three phases over fourteen weeks at $96,000 fixed, starting with a two-week discovery. Sam signs; work starts the Monday after." -
02Restate their situation using their words
This is the section that decides whether the rest gets read, because it is the only proof you were listening. Use the phrases they used on the call, including the ones that were slightly awkward. Do not translate their problem into your vocabulary.
On this dealThey said "our clients ring us to ask where their order is". Write that, not "suboptimal customer visibility". -
03Explain the approach, and why this one
Not a methodology diagram. The reasoning behind the shape of the work: why three phases rather than one, why discovery first, what you would learn in phase one that changes phase two.
On this deal"Discovery first because the order data lives in two systems and we do not yet know whether they reconcile. If they do, phase two is four weeks. If they do not, we will tell you in week two rather than in month three." -
04Name deliverables, not activities
An activity is something you will be doing. A deliverable is something that arrives, in a format, on a date. If a third party could not tell whether it turned up, it is not a deliverable.
On this dealNot "ongoing project management". Instead: "a weekly written status note every Friday, and a signed-off phase report at the end of each phase". -
05Say what is not included
The section most proposals omit and most arguments concern. Exclusions protect the buyer too, because they make the price legible. Three or four lines is enough.
On this deal"Excludes: data migration from the legacy CRM, third-party licence costs, out-of-hours support, and any changes to the ordering system itself." -
06Price it as a choice, not an ultimatum
One number invites yes or no. Three tiers invite a decision about which, and a buyer choosing is further along than a buyer approving. Compute them from a real price list so the middle option is genuinely the right answer rather than a decoy.
On this dealEssential at $72,000, Complete at $96,000, Complete plus twelve months of support at $128,000, with what each includes. -
07Write down what you assumed
Every estimate rests on assumptions. Written down, a wrong one reopens the price cleanly. Unwritten, it becomes an argument about what was reasonable to expect.
On this deal"We have assumed the order API is documented and that one Northwind developer is available two days a week during phase two." -
08End with one action and a deadline
Not "let us know your thoughts". A named action, by a named person, by a date, and a validity period on the price so there is a reason to act.
On this deal"To proceed: Sam signs below. This pricing holds for 30 days. We can start the Monday after signature."
Four sentences, before and after.
Each of these is a line that appears in real documents, and the reason it fails.
We are a full-service digital agency with over a decade of experience delivering transformative solutions.
You said clients ring you to ask where their order is. This proposal is for the portal that stops those calls.
The first sentence is about you and could open anyone's proposal. The second proves you listened and could only open this one.
Pricing: $96,000. Timeline: 14 weeks.
$96,000 fixed across three phases. Discovery (2 weeks, $18,000), build (8 weeks, $62,000), launch support (4 weeks, $16,000). Net 45.
A single figure invites a challenge to the figure. A breakdown invites a conversation about scope, which is a better conversation to be having.
We will provide ongoing support and project management throughout the engagement.
A written status note every Friday, and a named project lead you can call. Support after launch is four weeks, then it stops unless extended.
The first promises something unbounded that you will be held to. The second is specific enough to deliver and specific enough to price.
Please let us know if you have any questions or would like to discuss further.
To proceed, sign below. Pricing holds for 30 days and we can start the Monday after.
The first ends the document without a next step. The second tells the reader exactly what to do and gives them a reason to do it now.
The four common ways this goes wrong.
- It opens with you
- Company history, values, a client logo wall. The buyer already decided you were plausible or the call would not have happened. What they are deciding now is whether you understood the problem.
- It describes activities
- "Strategic planning", "stakeholder alignment", "ongoing optimisation". None of these arrives on a date, so none of them can be judged, so the project cannot be finished.
- It has one number and no options
- A single price is a yes-or-no question, and no is the easier answer. Options change the conversation from whether to which.
- It goes out three days late
- The commonest failure and the least discussed. A proposal that lands while the call is still fresh beats a better proposal that lands on Thursday. This is the whole reason DocuDeal exists.
How long should it be?
As long as the decision requires, which is shorter than most people write. A repeat client who knows how you work needs two pages. A competitive tender against a formal requirements matrix might need twenty and will be scored on compliance rather than prose. The constant is that a decision-maker who was not on the call reads the summary and possibly nothing else, so length is a question about the summary as much as the document.
| Repeat client, known scope | 1–2 pages |
|---|---|
| New client, straightforward project | 4–6 pages |
| New client, complex or multi-phase | 8–12 pages |
| Formal RFP with a requirements matrix | As specified, and answer in their order |
Straight answers.
What should a proposal include?
A summary, their situation in their own words, your approach and why, named deliverables, exclusions, a timeline with dependencies, priced options, your assumptions, who does the work, and one clear next step with a validity period. The order matters as much as the contents.
How long should a proposal be?
Long enough for the decision and no longer. One to two pages for a repeat client, four to six for a straightforward new project, eight to twelve for something complex. A formal RFP is answered in the length and order the RFP specifies.
Should I include pricing in the proposal?
Yes, and as options where the work allows it. Holding price back to force a follow-up call reads as evasive and adds a week. Three tiers computed from a real price list turn a yes-or-no question into a choice.
How quickly should I send it?
The same day if you can. A good proposal on Tuesday beats a better one on Friday, because the conversation is still warm and no competitor has landed theirs. This is the part AI actually changes: hand over the transcript and the notes and the draft comes back written and priced in seconds.