A statement of work, on the same engine as the deck.

What we propose, in one page.

Open with the recommendation, not the reasoning — a reader who stops after this page should still know what is being bought, what it costs, and when it lands. Everything after this is evidence for a decision already stated here.

Replace this paragraph with the shape of the engagement in the client's own words: the outcome they asked for, the constraint that makes it urgent, and the one thing you will do differently from the obvious approach. Two paragraphs at most — the detail has its own pages.

What we heard, said back plainly.

Prove you listened before you propose anything. Everything on this page should be traceable to something the client said, wrote, or published — no invented context, and nothing you can't point at a source for.

• The situation today — what exists, who owns it, and where it stops working. One sentence per fact.
• What triggered the ask — a deadline, a launch, an audit, a hire. Name it; urgency is the reason a proposal gets read.
• The constraint that shapes the work — a budget ceiling, a compliance regime, a system you must not touch.
• How success gets judged — in the client's measure, not yours. If it isn't measurable, say so here rather than implying it later.

Anything on this page you can't source — a number, a date, a name — wears the TBD marker until the client confirms it: TBD confirm at kickoff.

Deliverables, one card each.

A deliverable is something the client can receive, review, and accept. If a line here can't be handed over, it belongs in the approach on the next page, not in scope.

Each acceptance criterion is repeated in the schedule on page 6 and in the acceptance table on page 10 — three places, one wording. If they ever disagree, this page governs.

Three phases, each one payable.

Phases exist so both sides can stop cleanly. Each one ends in something the client accepts and pays for, which is also what makes the schedule on the next page enforceable.

Phase durations are working weeks and stay TBD until a start date is agreed. The milestone dates on page 6 are derived from them — change one and the other follows.

Milestones and who owns each one.

Every row names an owner. A schedule where the client owns nothing is a schedule that slips quietly — the dependencies are the honest part of the table.

Dates are working days from signature and assume the access listed on page 9 is granted within five days of kickoff. Every date above is a TBD until the start date is fixed.

The fee, broken down by phase.

Price the phases, not the hours — it is the only breakdown a client can act on. The highlighted row is the total; it is the number the reader is looking for, so let it be found.

Fees are quoted in TBD currency, exclusive of tax, and valid for 30 days from the issue date on the cover. Payment terms: TBD days from invoice.

Who does the work, by name.

Name the people who will actually be on the engagement. A roster of roles with no names reads as a roster of people not yet hired.

What this price depends on.

Assumptions are the cheapest insurance in a proposal: every one you write down is a change request you don't argue about later. Exclusions are the same courtesy in reverse.

Sign here, and phase one starts.

The signature block is the one component a proposal needs that a slide deck never does — ruled lines, a name, a role, a date. It lives in shared/paper.css as .sign.

• A countersigned copy comes back to you within one working day.
• Kickoff is scheduled inside five working days, against the milestone table on page 6.
• The first invoice is issued on signature, payable in TBD days.

Signed by both parties, this document and its assumptions are the whole agreement for the work described. Anything agreed later is added in writing and priced before it starts.

Your wordmark here

The margin note that never ships.

This page carries data-internal on its <section> tag, so node build.mjs starter-a4 --public drops it from the output entirely — and the build refuses to write the file if any internal page survives the strip.