The problem
Every proposal was assembled by hand. Price the deal in a calculator, rebuild a deck for the client, assemble a contract packet page by page depending on the services sold, then place signature fields manually. The same numbers were retyped at each step into documents a client would sign, and a prospect who said yes waited days for paperwork to catch up.
What I built
One page, top to bottom. Configure the quote with tiers, terms, per-item discounts, custom line items, promotions, and referral credits while the breakdown recalculates live. One click produces a branded proposal from a 22-page template with unsold service pages removed and the money pages rendered from engine output. A second click builds the contract packet from a 20-page template, keeping only the terms pages for the exact products sold and generating the payment authorization language from the quote math. A third click sends it for signature with fields already placed and routed.
The browser and the server share one pricing module. The server recomputes every quote before generating a document and instructs the generator to recompute and reformat nothing. Only one piece of code is permitted to do arithmetic, so a figure on screen, in the proposal, and on the contract cannot diverge.
Decisions worth noting
Deterministic layout over generative layout: page selection, pagination, and content packing are computed server-side against a measured fit budget and handed to the generator as instructions. Removal of image-only template pages is anchored to background filenames after title-based removal silently failed. Signature fields sit on image-only pages, so I rendered real packets with a percentage grid overlaid, read the positions off the render, and encoded them as a maintainable coordinate map.
DocuSign is integrated through JWT service auth with sequential signing and automatic internal distribution. A self-hosted Documenso instance carries the equivalent integration so the agency is not locked into per-seat signature pricing. Everything generated is a draft; nothing goes to a client automatically.
Result
The presales-to-sales handoff went from days to the same conversation. Proposals take about two minutes unattended, contract packets assemble themselves correctly, and 44 quotes have run through production across eight team accounts. The pipeline lives inside the authenticated team portal I built alongside it, which replaced shared passwords with scrypt hashing, httpOnly sessions, rate-limited logins, and expiring setup links.
In brief
- What is Quote-to-Signature Pipeline?
- A sales rep configures a deal on one page and gets back a client-ready proposal PDF and a signature-ready contract packet, with every figure computed once by one shared engine.
- What did Brian Powell build for Quote-to-Signature Pipeline?
- One page, top to bottom. Configure the quote with tiers, terms, per-item discounts, custom line items, promotions, and referral credits while the breakdown recalculates live. One click produces a branded proposal from a 22-page template with unsold service pages removed and the money pages rendered from engine output. A second click builds the contract packet from a 20-page template, keeping only the terms pages for the exact products sold and generating the payment authorization language from the quote math. A third click sends it for signature with fields already placed and routed.
- What was the result of Quote-to-Signature Pipeline?
- The presales-to-sales handoff went from days to the same conversation. Proposals take about two minutes unattended, contract packets assemble themselves correctly, and 44 quotes have run through production across eight team accounts. The pipeline lives inside the authenticated team portal I built alongside it, which replaced shared passwords with scrypt hashing, httpOnly sessions, rate-limited logins, and expiring setup links.