Sales Funnel Prompt: Stage Leaks, Fixes and Follow-Ups
Paste 90 days of stage counts, your offer and capacity: get stage definitions, conversion math, the biggest leak, one fix per stage and a follow-up sequence.
Prompt
Act as a revenue operations consultant who builds sales funnels for small service businesses from their own pipeline numbers. You start from real stage counts, not funnel templates, you show every conversion rate as arithmetic, and you fix the leakiest stage first. You never promise conversion rates; any improved number you use is labeled as a scenario. Inputs: - Offer, price and contract length: [Offer] - Ideal buyer and how they usually find you: [IdealBuyer] - Lead sources with counts if known: [LeadSources] - Stage counts for one period, in order: [StageCounts] - Average time from first contact to signed deal: [SalesCycle] - What happens today after each stage: [CurrentFollowUp] - How many new clients you can take on per month: [Capacity] - Goal for new clients per month: [Goal] - Tools and limits (CRM, email tool, time available): [Constraints] Generate: 1. Stage map: each stage with a one-line definition and the exit criteria that move a lead to the next stage. 2. Conversion math: rate for each stage step and overall, as count divided by count with the percentage. 3. Biggest leak: the stage step with the weakest rate or largest drop, explained using only the inputs, plus the questions you would need answered to confirm the cause. 4. One fix per stage: a concrete asset, offer or process change that fits Constraints. 5. Follow-up sequence for the leakiest stage: day, channel, purpose and a short message draft for each touch. 6. Goal math: how many leads the current rates need to hit Goal, and one labeled scenario showing which rate changes would get there. Compare with Capacity. 7. Weekly metrics: five numbers to track, each with where it comes from in Constraints. Rules: use only the counts and facts in the Inputs. Round percentages to one decimal and keep the math visible. For promotional follow-up email, note the CAN-SPAM basics, which apply to business-to-business email too: accurate sender and subject lines, a valid postal address, a clear way to opt out, and opt-outs honored within 10 business days.
Instructions
Pull the stage counts from your CRM for one clear period, and use the same period for every stage. If your stages differ, list yours in order in StageCounts. Variation: for an online offer, use visitors, sign-ups, trials and paid as stages. Tip: record a lost reason on every closed proposal for one month before you change prices. Quillbrae Bookkeeping is an invented example. Scenario numbers show what would need to change; they are not predictions. HubSpot and Gmail appear only as example tools; PromptDig is not affiliated with HubSpot, Google or OpenAI.
Generated Output
This image was generated using the prompt above.

Examples
Example Input
Offer: monthly bookkeeping for independent restaurants, 650 per month, 12-month agreement IdealBuyer: owner-operators of one to three restaurants; most come from referrals by their accountant or a restaurant supplier LeadSources: referrals 38, website form 22 (last 90 days) StageCounts (last 90 days): inquiries 60; discovery calls booked 30; calls held 24; proposals sent 18; signed 6 SalesCycle: about 21 days from inquiry to signature CurrentFollowUp: inquiry gets a booking link by email; no reminder before calls; proposal sent by email with one follow-up email after a week, then nothing Capacity: can onboard 4 new clients per month Goal: 4 new clients per month Constraints: Quillbrae Bookkeeping (fictional), two people; HubSpot free CRM; Gmail; about 3 hours a week for sales follow-up
Example Output
1. STAGE MAP Inquiry: a restaurant owner contacts you by referral or form. Exit: discovery call booked. Booked: call on the calendar. Exit: call held. Call held: you learned locations, current bookkeeping and pain. Exit: fit confirmed and proposal requested. Proposal sent: price and scope in writing. Exit: signed or a clear no. Signed: 12-month agreement at 650 per month. 2. CONVERSION MATH (last 90 days) Inquiry to booked: 30 / 60 = 50.0% Booked to held: 24 / 30 = 80.0% Held to proposal: 18 / 24 = 75.0% Proposal to signed: 6 / 18 = 33.3% Overall: 6 / 60 = 10.0% 3. BIGGEST LEAK Proposal to signed, 33.3%: 12 of 18 proposals did not close, and follow-up stops after one email at day 7. Inquiry to booked, 50.0%, loses the most people (30), so it is the second fix. To confirm: Did lost proposals say no, or go silent? Was price the reason? Do referral leads close better than form leads? 4. ONE FIX PER STAGE Inquiry: reply within one business day with two call times instead of only a booking link. Booked: reminder email the day before the call. Call held: end the call by agreeing a date to review the proposal together. Proposal: walk through it on a 15-minute call instead of sending it cold. Signed: a short onboarding checklist so the first month starts on time. 5. PROPOSAL FOLLOW-UP SEQUENCE Day 0, call: walk through the proposal. "I'll send this now. Can we pick a time Thursday to answer questions?" Day 2, email: "Here's the proposal we went through. The main question owners ask is how month-end works with us. Happy to show you." Day 5, phone: check in on the decision and who else needs to see it. Day 10, email: "Would it help to start with one location first?" Day 14, email: "Should I close this out for now? If timing is the issue, tell me when to check back." One person, about 20 minutes per open proposal across the sequence. 6. GOAL MATH Goal 4 per month = 12 per 90 days. Now: 6 per 90 days, 2 per month. At the current 10.0% overall, 12 signed needs 12 / 0.10 = 120 inquiries per 90 days, double today's 60. Scenario (not a forecast): inquiry to booked 65%, proposal to signed 50%, other rates unchanged: 60 x 0.65 = 39 booked, x 0.80 = 31.2 held, x 0.75 = 23.4 proposals, x 0.50 = 11.7 signed per 90 days, about 3.9 per month. That is close to Capacity of 4. 7. WEEKLY METRICS (HubSpot free CRM) New inquiries by source; calls booked; calls held; proposals sent; proposals signed or lost with reason.