Published vendor benchmarks put an agency proposal at 4–6 hours of staff time, at a volume of 8–12 proposals a month (Pitchsite, 2026). Used as assumptions, not as GradeCircle data, that comes to 32–72 staff hours a month spent preparing proposals.
That range is a fair place to start. It is not your number. Yours depends on who writes your proposals, what they start from, and how many rounds of changes happen before one goes out. It is also the only number you can plan around, so it is worth a few weeks of light record-keeping to find.
The benchmark, in hours you can picture
The arithmetic is short. At the low end, 8 proposals at 4 hours each is 32 hours a month. At the high end, 12 at 6 hours is 72. That is somewhere between most of one person's working week and close to two full weeks, every month.
Who carries those hours matters more than the total. Proposal preparation is senior-heavy. The people who write proposals are usually the people who also sell and run accounts: the owner, the strategy lead, the account director. An hour of theirs spent on assembly is an hour not spent on a client who is already paying you.
It is also repeatable. Every proposal pulls from the same material: discovery notes, your template, service descriptions, and past examples you were happy with. Someone rebuilds that by hand each time.
Where four and a half hours go
An illustrative example, not a client result. The owner of a 20-person digital agency had a good discovery call on Tuesday. On Thursday at 10pm she opens last quarter's proposal for a similar client and starts writing over it.
- Rereading her discovery notes and the prospect's site to rebuild the brief: 45 minutes.
- Rewriting objectives and approach so they are about this client, not the last one: 90 minutes.
- Copying service descriptions and deliverables, then fixing the ones that changed since: 45 minutes.
- Pricing and timeline, checked against the current rate card: 30 minutes.
- Her strategy lead's comments the next morning, and a second pass: 60 minutes.
Four and a half hours, inside the benchmark. Look at how little of it was judgment. Deciding what this client needs and what to charge took a fraction of the evening. The rest was assembly: finding, copying, adapting and rechecking material the agency had already approved once. That split is what you are looking for when you measure your own.
Your number is the one that matters
A benchmark tells you whether to look. Only your own baseline tells you what a change would be worth. Measure three things, the same three any new workflow should later be judged on.
- Preparation time. Hours per completed proposal, from discovery notes to the version that goes out, across everyone who touches it. Not only the writer.
- Corrections. Every change made after the first full draft, and why: scope, pricing, a factual claim, a client detail. Corrections show where drafts go wrong, and they are where saved time quietly comes back.
- Adoption. Who actually uses the template, the approved examples and the rate card, and who starts from a blank page or an old file of their own. A process half the team ignores is not your process.
How to take your baseline
Keep it light. You want a few weeks of honest numbers, not a time-tracking project.
- Pick the next five proposals, or a full month if you send more than that. Include the types you actually send: renewals, expansions, new pitches.
- Name a proposal owner for each one. That person records everyone's hours on it, by role, on the day the work happens. Hours remembered at the end of the month are guesses.
- Save the first full draft. Compare it with the version that went out and count the changes by reason.
- Note what each draft started from. Template, past proposal, or blank page. That is your adoption figure.
- Multiply by your monthly volume. Hours per proposal × proposals a month is your monthly figure. Put it next to 32–72.
Then read it plainly. If proposals take at least 20 staff hours a month, or cause measurable delays, the workflow is big enough to test a change on. That is the same threshold the AI Opportunity Review uses for a recurring workflow. If senior people carry most of the hours, the case is stronger. If corrections cluster around pricing or scope, fix the source material first. A draft can only be as current as the rate card it was built from.
A faster draft is not fewer hours
Your team may already open a general AI tool to start drafts. That is a fair place to begin. It is not a measured process. Proposal software and general AI tools can both produce a draft. What neither comes with is your process: who feeds it, who checks it, and what counts as done. Without a baseline, the hour one person saves on the first draft can come back as someone else's corrections, and nobody sees the trade. AI proposal generator or your own workflow sets out what to compare before you buy.
Two more things to keep straight. Released hours are capacity, not cash: they do not automatically reduce payroll. And faster preparation does not guarantee more wins, because win rates depend on more than drafting speed. The honest case for a faster workflow is time back for your senior people, measured against what you have today.
The Review starts from your baseline
The AI Opportunity Review is built around this measurement. It costs $1,500 and takes 10 business days. At kickoff you name the proposal owner and show us how a proposal is built today, from discovery notes to the version that goes out. We review representative work samples and up to three workflows, interview up to three people, and score each workflow on benefit, effort, data suitability and operational risk.
The business case begins with the same benchmark this article opened with, then replaces it with your own evidence: your volume, your hours per proposal, your people. Capacity is reported as hours, not as payroll savings, in conservative-to-optimistic scenarios. The sample review shows the shape of the arithmetic with an illustrative assumption: 12 proposals a month × an assumed 90 minutes released each ≈ 18 hours of monthly capacity. It also names what remains uncertain: actual time released, rework required, staff adoption, and whether released time can be used productively.
The review ends in a decision meeting. You leave with a recommended first pilot at a fixed price, or a recommendation to defer AI and what would need to change. If your team can get there on its own, the review says so.
A pilot is a separate decision: $5,000 over four weeks, one workflow, one team of up to five. The draft is built from approved inputs only, and a missing budget or timeline is flagged for the owner to resolve, not filled in. Preparation time, corrections and adoption are measured against your baseline, and the pilot ends with expand, revise, or stop. Software and usage fees are extra and disclosed before you commit. For the ten days in detail, see what you get from a $1,500 AI Opportunity Review.
Scoping is free and commits you to nothing. Paid work starts only after you agree the scope and fixed fee.
GradeCircle builds and runs six products of its own: AfroSynergy, CallerBridge, HealthSquire, Heyzl, Lexfora, and LukeAPI, two of them in beta. The assumptions in our models come from operating those systems, not from a framework. See what we run.
If your agency writes proposals every month, scope your proposal workflow: five questions, and you see a preliminary fit summary before sharing contact details.



