A winning agency proposal has eight sections: a one-paragraph summary the decision-maker could act on alone, the client's situation in their own words, the objectives, your approach, the deliverables, the scope boundary including exclusions, three priced options, and the next step with a date. Two to six pages is the right length for most agency work. The two sections that decide the outcome are the summary - often the only part a senior approver reads - and the pricing, which loses more deals than any other. Send it within 48 hours of the conversation that produced it, and never send it without a scheduled call to walk through it.
Most agency proposals are written for the wrong reader.
They are written for the person the agency spoke to - the marketing manager who ran the discovery call, understands the context, and is already broadly sold. But that person is rarely the one who approves the spend. The proposal gets forwarded to a director or a finance lead who was not on the call, has forty minutes of context-free reading to do that week, and will form a view from the first paragraph and the price.
A proposal that only works if the reader was in the room is a proposal that dies in a forward.
This guide covers the eight sections that belong in an agency proposal, how long it should be, what to do about pricing, the follow-up process that converts, and the mistakes that lose deals you had already won.
What a proposal is actually for
Worth being precise, because it changes what you write.
A proposal is not a sales document in the persuasive sense. By the time you are writing one, the client has usually decided they want the problem solved and that you are a plausible supplier. Selling harder at this stage rarely helps.
A proposal is a decision document. Its job is to make saying yes easy, and to make saying yes to you specifically the obvious version of yes. That means it has to survive three tests:
- The forward test. Can someone who was not on the call understand it?
- The comparison test. If it is read next to a competitor's, is the difference visible?
- The objection test. Does it answer the two or three concerns the buyer will raise internally, before they have to raise them?
Everything below is in service of those three.
The eight sections
HubSpot's guide to consulting proposals, Consulting Success's template and Projectworks' walkthrough all converge on broadly this structure. The ordering below is the one that survives the forward test best.
1. The summary (half a page, written last)
One paragraph the approver could act on without reading anything else: what the problem is, what you propose, what it costs, and how long it takes.
This is the section that does the most work and gets the least attention. Write it last, when you know what the proposal actually says, and write it as though it is the only page that will be read - because for the person signing, it frequently is.
A workable shape:
Northwind's current site converts 1.2% of demo requests, against an industry benchmark closer to 3%, costing an estimated £180,000 in annual pipeline. We propose a six-week conversion-focused rebuild covering the three highest-traffic journeys, delivered by 14 November for £28,000. Detail follows.
Four sentences. Problem, cost of the problem, proposal, price and date. Everything after it is supporting evidence.
2. Their situation, in their words
Restate the problem as they described it. Where you can, use their phrasing - if they said "our site embarrasses us on sales calls," write that rather than "brand perception challenges."
This section exists to prove you listened, and it is disproportionately persuasive. A buyer reading their own words in a supplier's document concludes, correctly, that the supplier understood them. It also gives the person who was not on the call the context they are missing.
Keep it factual and avoid the temptation to editorialise about how bad things are. You are demonstrating comprehension, not diagnosing incompetence - and the person who built the current site may well be in the approval chain.
3. Objectives
Three to five, stated as outcomes rather than activities, and measurable wherever the client gave you numbers.
- Weak: "Redesign the homepage and key landing pages."
- Strong: "Increase demo request conversion from 1.2% to a target of 2.5% within one quarter of launch."
If the client did not give you numbers, say what will be true rather than what will exist. "Sales can send a prospect to the site without caveating it first" is an objective. "A new site" is a deliverable, which comes later.
4. Approach
How you will work, in phases, with what happens in each. This is where you differentiate, because approach is the thing competitors genuinely differ on and price is not.
Keep it to a paragraph per phase. The buyer does not need your methodology in full; they need to believe you have one and to see where their involvement is required.
Name the client's obligations explicitly here - the reviews, the approvals, the content, the access. Two benefits: it sets expectations before the project starts, and it quietly signals that delays have two possible sources. Our guide to preventing scope creep covers why naming dependencies early is the cheapest insurance available.
5. Deliverables
A concrete list of what they receive. Specific enough to count.
- Weak: "Website redesign."
- Strong: "Designs for 6 page templates, built and responsive across 3 breakpoints, with a CMS the marketing team can update, plus a 90-minute handover session and written documentation."
Specificity here is what makes the price make sense. A buyer comparing £28,000 against £14,000 from another agency needs to see what the difference buys, and this is the section where it becomes visible.
6. The scope boundary
The section most agency proposals omit, and the one that prevents the most pain.
Two parts:
Assumptions. What has to be true for the price and timeline to hold. "Content is provided by 3 October." "One consolidated round of feedback per stage, from a single named approver." "Existing brand assets are available in editable formats."
Exclusions. What is explicitly not included. Not to be defensive - to be clear. "Copywriting for the blog," "ongoing hosting and maintenance," "migration of historical posts," "third-party integrations beyond the CRM."
Agencies avoid exclusions because they fear looking negative or inviting a haggle. In practice they do the opposite: they signal that you have done this before and know where projects go wrong. And they are the mechanism that makes a fixed price safe, because everything unlisted otherwise resolves in the client's favour. Our guide to writing a statement of work covers this in more depth; the proposal version can be shorter, but it must exist.
7. Pricing - three options
Pricing loses more deals than any other section, usually because there is only one number.
A single price is a yes/no decision, and no is always the safer answer for a buyer with a budget to protect. Three options convert the question into "which," which is much easier to answer, and it gives the buyer a way to exercise judgement without saying no.
The structure that works:
- Option A - the reduced scope. Solves the core problem, less of it. Real, not a strawman.
- Option B - the recommendation. What you would do. Mark it as such.
- Option C - the extended version. More scope, more outcome. Some clients take it, and its main job is to make B look measured rather than expensive.
Differentiate on scope, never on quality. "Option A gets 2 rounds of revisions, B gets 4" is a legitimate ladder. "Option A is our junior team" is not - it tells the buyer that some of your work is worse, and they will wonder which they are getting.
State the price plainly, with no apology and no burying. If it is a fixed fee, say what it includes and what triggers a change order. If it is a retainer, state the allocation, the rollover rule and the overage rate - see retainer models.
For deriving the number in the first place, our guide to pricing agency services covers the four models and how to calculate a real rate floor.
8. Next step, with a date
End with one specific action and a deadline.
- Weak: "Let us know if you have any questions."
- Strong: "If Option B works, reply to confirm and we'll send the contract the same day. We're holding capacity for a 21 October start until the 14th."
The deadline must be real - tied to your actual capacity, not manufactured. A genuine constraint is persuasive and fair. A fake one is transparent and damages trust with exactly the sophisticated buyers you want.
How long, and what to leave out
Two to six pages covers almost all agency work. Longer proposals do not win more; they get skimmed more.
Things that almost always belong somewhere other than a proposal:
Long company backstory. They have seen your site. One or two sentences on relevant experience, ideally a comparable project, is enough. If credentials are genuinely decisive, put them in an appendix.
Team bios. Unless a named individual is part of what they are buying, which for some consultancies it is.
Methodology diagrams. Impressive to you, noise to them.
Full terms and conditions. These belong in the contract that follows. A proposal cluttered with legal text invites legal review, and legal review adds three weeks.
The test for any section: does this help them decide? If it helps them evaluate you, it might belong. If it exists because it feels professional to include it, cut it.
The process around the document
The proposal itself is maybe half of what determines the outcome. The process around it is the rest.
Send within 48 hours
Momentum decays fast. A proposal arriving a week after the conversation lands with someone who has half-forgotten the detail and has possibly spoken to two competitors since. Two days is the target; same-day is better for smaller engagements.
This is only realistic if you are not writing every proposal from scratch. Which brings us to templates.
Template the structure, never the substance
The eight sections above should be a template. The content of sections 2, 3 and 6 must be specific to the client every time.
A recognisably generic proposal is worse than a slow one. Buyers can tell immediately - the giveaway is usually a situation section that could describe any company in the sector - and it undoes the impression that you understood their problem.
The efficient version: a template that carries your structure, your standard exclusions, your deliverable descriptions and your boilerplate, leaving the situation, objectives and pricing to be written fresh. That is 80% of the time saving with none of the generic feel.
If you run structured intake, most of section 2 is already written by the client themselves - which is one of the underrated benefits of a proper intake process over an unstructured discovery call.
Never send it cold
Book a call to walk through it, before you send it. Then send it 15 minutes before the call, or present it live.
A proposal read alone is a proposal read for reasons to say no. A proposal walked through is a conversation where objections surface while you are there to answer them. The difference in close rate is large and costs you thirty minutes.
If the buyer will not take a call, that is information. Usually it means you are the third quote for a decision already made.
Follow up on a schedule, then stop
Two follow-ups, then a close-out.
- Day 3: short, useful. Not "checking in" - add something. "One thing I should have included: here's how we handled the same migration for another client."
- Day 7: the direct question. "Is this still live? Happy to adjust the scope if the budget is the issue."
- Day 14: the close-out. "I'll assume the timing isn't right and close this off - do come back if that changes."
The third message is the one people skip and the one that most often gets a reply. It removes the obligation to respond, which is precisely what makes responding easy.
Format: document, deck, or proposal tool
The container matters less than the content, but it is not neutral.
A document (PDF). The default, and correct for most agency work. It survives forwarding, reads well on a phone, and can be skimmed in any order - which is how a busy approver actually reads. The risk is that a badly-formatted document reads as effort-free; give it typographic care.
A deck. Suits proposals that will be presented live to a group, particularly for larger engagements with several stakeholders in the room. The failure mode is that decks read terribly alone, and proposals get forwarded. If you use one, write a document version too, or accept that the deck must carry full sentences rather than bullets.
A proposal tool. Web-based proposals with tracking, e-signature and option selection built in. The genuine advantage is not the signature - it is knowing whether the proposal was opened, by how many people, and which sections they lingered on. That information changes your follow-up from guesswork to targeting. The disadvantage is that some corporate buyers cannot easily forward or print them, and some procurement processes require a PDF anyway.
A practical answer: write in whatever tool you like, export a clean PDF, and send both if you want the tracking.
Measure your win rate, by reason
Almost no agency tracks this, and it is a small amount of work for a large amount of clarity.
For every proposal, record four fields: date sent, value, outcome, and stated reason if lost. Review quarterly.
Three patterns typically emerge, each pointing at a different fix:
Losing on price consistently. Either you are genuinely expensive for the value you communicate, or - far more often - the proposal is not conveying the value, so price is the only variable left to compare. The fix is usually in sections 2 and 3, not in the number.
Losing to "went with someone else" with no price mention. Usually a positioning or trust problem. Look at whether your proposals lead with their problem or with you.
Losing to "we've decided not to proceed." The deal was not qualified. The fix is upstream in intake and qualification, not in the proposal at all - you are writing documents for people who were never going to buy.
A win rate above roughly 60% is worth examining too. It usually means you are underpricing, or only bidding on work you have effectively already won. Neither is bad, exactly, but both are worth knowing deliberately.
The mistakes that lose deals you had won
Leading with yourself. A proposal that opens with three paragraphs about your agency has answered a question nobody asked. Open with their problem.
Pricing by line item. Itemising every component invites the buyer to remove components, and they will remove the ones that look optional - discovery, testing, project management - which are precisely the ones that make the project succeed. Price the outcome, list what is included.
Presenting one option. Covered above, and it is the single highest-impact change most agencies can make.
Hedged language. "We would aim to potentially deliver approximately..." reads as a lack of confidence in your own estimate. If you are uncertain, name the uncertainty explicitly and say how you will resolve it - "the integration scope depends on their API, which we'll confirm in week one" is confident. Vagueness spread evenly across the document is not.
No exclusions. The proposal wins, the project loses money. This is the most expensive mistake on the list because it does not look like a mistake at the time.
Sending a PDF that expires nowhere. A proposal with no validity date can be resurrected nine months later at a price that no longer works. Put a date on it.
When they go quiet
The most common proposal outcome is not rejection. It is silence, and silence is usually organisational rather than personal.
Three things are typically happening. The champion is stuck internally - waiting on a budget holder, a competing priority, or an approval process they did not warn you about. A competitor is in play and they are comparing. Or the priority has genuinely shifted and nobody wants to say so.
All three are best served by the same move: make it easy to tell you the truth.
Completely understand if the timing has moved - that happens. If it's helpful, I can hold the pricing until the end of the month; if it's shelved for now, just say and I'll close it off and check back in the new year.
This works because it removes the social cost of a no. Most people go quiet not because they are avoiding you but because "we've decided not to proceed" feels like a difficult message to write. Offering to close it off yourself is a kindness, and it converts a large number of dead proposals into either a clear no - which is valuable - or an honest explanation of the real blocker, which is frequently something you can help with.
What does not work is repeated cheerful check-ins with no new information. Three of those and you have trained them to ignore your name in the inbox.
A reusable skeleton
If you want to build a template this week:
- Summary - four sentences. Problem, cost, proposal, price and date.
- Your situation - their words, factual, no editorialising.
- Objectives - three to five outcomes, measurable where possible.
- Approach - phases, a paragraph each, client obligations named.
- Deliverables - specific enough to count.
- Assumptions and exclusions - what must be true, what is not included.
- Options - three, differing by scope, recommendation marked.
- Next step - one action, one real date.
Two to six pages. Sent within 48 hours. Walked through on a call. Followed up twice and then closed off.
None of that is clever. It is just the version that survives being forwarded to someone who was not in the room - which is where most proposals are actually decided.