Turn discovery notes into a proposal draft and ROI proof

Create one working folder or Google Doc for the deal before drafting begins. Pull the Sybill or Gong call summary, transcript snippets, HubSpot deal fields, buyer emails, pricing guidance, implementation constraints, proposal template, and approved proof points. Add a short source index that labels what came from the buyer, what came from CRM, what came from internal teams, and what is still missing. QA check: do not proceed until pricing, legal terms, timeline commitments, and delivery assumptions are either approved or clearly marked as open. For each input, keep the source, collection date, owner, and confidence level so later AI summaries can be checked instead of trusted blindly.

Use Claude to turn the source packet into a facts, assumptions, and open questions table. Require evidence for every confirmed fact, such as transcript language, CRM notes, buyer emails, or internal pricing guidance. Mark reasonable assumptions separately from unsupported assumptions so the proposal does not accidentally present guesses as truth. QA check: anything tied to price, timeline, integration effort, legal terms, or customer outcomes must be validated before it appears in buyer-facing copy. Give the AI concrete source material, required fields, and a review checklist so the output can be judged by a human operator. Use the prompt template in this step only after pasting the relevant source material; require the answer to return assumptions, missing information, and a clear next action.

Use the confirmed buyer evidence to write a short problem narrative before writing any proposal sections. Include current state, pain, business impact, cost of inaction, desired future state, and success criteria. Keep the buyer’s wording where it is useful, but avoid over-quoting or turning transcript fragments into a dramatic story. QA check: the rep should be able to read the narrative and say, “Yes, this is what the buyer actually told us.”. Ask for a working draft with named sections, assumptions, gaps, and review flags rather than treating the first AI answer as final. Use the prompt template in this step only after pasting the relevant source material; require the answer to return assumptions, missing information, and a clear next action.

Create a pain-to-scope table in Google Docs before drafting the proposal. For each proposed service, capability, feature, or phase, show the buyer pain it addresses, the evidence source, expected impact, proof point, dependency, and open question. Remove or move to an appendix any scope item that does not connect to buyer evidence or a required implementation step. QA check: no major scope item should survive unless it has a buyer problem, an internal rationale, and an owner for validation. Write down the decision criteria, owner, required inputs, and what is deliberately out of scope so the rest of the workflow does not drift. Use the prompt template in this step only after pasting the relevant source material; require the answer to return assumptions, missing information, and a clear next action.

Create a simple Google Sheets model with value drivers such as time saved, revenue lift, cost avoided, risk reduction, adoption improvement, or cycle-time reduction. Use conservative, base, and upside ranges instead of a single exact ROI claim. Add columns for assumption source, required buyer input, confidence level, and validation owner. QA check: if the buyer has not validated an assumption, label it clearly in both the sheet and the proposal. Give the AI concrete source material, required fields, and a review checklist so the output can be judged by a human operator. Use the prompt template in this step only after pasting the relevant source material; require the answer to return assumptions, missing information, and a clear next action.

Review case studies, customer quotes, screenshots, product metrics, implementation examples, and security or integration proof. Choose only the proof points that support the buyer’s specific pain, industry, technical concern, or success criteria. For each proof point, note where it should appear: executive summary, scope section, ROI section, implementation plan, or appendix. QA check: remove generic logo slides unless they support the decision this buyer needs to make. Write down the decision criteria, owner, required inputs, and what is deliberately out of scope so the rest of the workflow does not drift. Use the prompt template in this step only after pasting the relevant source material; require the answer to return assumptions, missing information, and a clear next action.

Use Claude to create a proposal outline from the narrative, scope map, ROI assumptions, and proof list. The outline should include executive summary, buyer context, recommended approach, scope, assumptions, ROI model, proof, timeline, roles and responsibilities, open questions, and next steps. Add a note beside each section showing which source material supports it. QA check: get the rep or sales engineer to approve the outline before generating polished copy. Ask for a working draft with named sections, assumptions, gaps, and review flags rather than treating the first AI answer as final. Use the prompt template in this step only after pasting the relevant source material; require the answer to return assumptions, missing information, and a clear next action.

Use Claude to write the first proposal draft in Google Docs from the approved outline. Keep assumptions, open questions, buyer inputs needed, and validation items visible instead of hiding them in polished language. Write in a consultative tone that connects each recommendation back to the buyer’s stated pain. QA check: scan the draft for unsupported numbers, invented commitments, vague benefits, and anything delivery or legal has not approved. Ask for a working draft with named sections, assumptions, gaps, and review flags rather than treating the first AI answer as final. Use the prompt template in this step only after pasting the relevant source material; require the answer to return assumptions, missing information, and a clear next action.

Use Claude and Canva to create a one-page executive summary that the buyer champion can forward internally. Include the buyer problem, why now, recommended path, expected impact range, proof point, decision needed, and next step. Keep the page skimmable and avoid cramming the full proposal into a visual summary. QA check: ask whether the page still makes sense if it is forwarded without your email or sales call narration. Ask for a working draft with named sections, assumptions, gaps, and review flags rather than treating the first AI answer as final. Use the prompt template in this step only after pasting the relevant source material; require the answer to return assumptions, missing information, and a clear next action.

Route the draft to the account owner, sales manager, sales engineer, delivery owner, and anyone responsible for pricing, security, or legal terms. Ask reviewers to comment directly in Google Docs using a checklist for buyer accuracy, scope feasibility, unsupported claims, pricing assumptions, delivery risk, legal risk, and missing next steps. Resolve comments before sending instead of relying on the rep to explain issues live. QA check: the final version should have no unresolved comments on scope, price, implementation timeline, or commitments. Sample real examples, mark pass and fail reasons, and send ambiguous cases back to the owner before anything is published, synced, or used in front of a buyer. Use the prompt template in this step only after pasting the relevant source material; require the answer to return assumptions, missing information, and a clear next action.

Draft the proposal send email after the proposal is reviewed, not before. Reference the discovery conversation, state what the proposal is designed to solve, mention any assumptions that need validation, and recommend a live walkthrough. Keep the email short enough that the buyer understands the purpose without reading the full proposal first. QA check: the email should make it clear what feedback you want and what meeting should happen next. Ask for a working draft with named sections, assumptions, gaps, and review flags rather than treating the first AI answer as final. Use the prompt template in this step only after pasting the relevant source material; require the answer to return assumptions, missing information, and a clear next action.

After the buyer reviews the proposal, log questions, objections, requested edits, decision-process changes, and which sections created confusion. Update HubSpot with the buyer’s response and save the lessons in the proposal template or proof library. Use Claude to summarize what should change in future proposals, including better discovery questions, stronger proof, clearer ROI inputs, or sections to remove. QA check: every proposal should leave behind at least one improvement to the proposal system, even if the deal does not close. For each input, keep the source, collection date, owner, and confidence level so later AI summaries can be checked instead of trusted blindly. Use the prompt template in this step only after pasting the relevant source material; require the answer to return assumptions, missing information, and a clear next action.