Run in order. Prompts 1-4 produce the outreach artefacts before you send a single message — most founders do the reverse, send 30 random DMs in a week, get three replies and conclude that “cold doesn’t work”. Cold didn’t fail. The mechanic was never built. Don’t send a single outreach message until prompt 4 has produced a ranked list, a second-touch rule, and an honest count of how many real conversations the list can produce.

1. The warm-circle list, ranked by likelihood and segment fit. “My segment from M2 is [paste]. My offer from M5/M6/M7 is [paste]. List the 30-100 people I already have some relationship with who fit the segment — past colleagues, ex-clients, peer founders in the segment, people I’ve helped without asking for anything back, anyone who has already told me ’tell me when you launch’. For each: name, relationship strength (1-5), segment fit (1-5), likelihood-to-act (1-5), channel I’d reach them on, the last time we were in contact. Sort by (segment fit × likelihood-to-act). The top 20 are the warm-circle outreach list. The rest sit in a follow-up queue for week 3-4. Output as a table. If the top 20 are mostly relationship-strength-5 friends with segment-fit-2, the warm circle isn’t a warm circle for this offer — it’s a friend group. Flag it.”
2. The first outreach message — one specific person at a time. “For each of the top 5 names from prompt 1, draft the one outreach message — personal, specific, not a template. Voice: from the M6 rulebook, peer-to-peer, not salesperson-to-prospect. Structure for each: one sentence that earns the right to be in their inbox (a specific shared context, a recent thing they did, a problem I know they have), one sentence that names what I’ve built in plain English, one sentence that asks the one specific thing (a 15-minute call, a look at the page, an honest reaction, a referral to one named person). Under 120 words. Reject any sentence that could be sent unchanged to any of the other four. Output the 5 messages, each clearly tied to the named person. If any message could be copy-pasted to a stranger, rewrite — the warm circle’s value is that you already know them, use it.”
3. The second-touch protocol — what you send when they don’t reply. “Most warm-circle outreach fails not at message 1 but at the silence after message 1. Draft the second-touch protocol: how long I wait before following up (rule of thumb: 5-7 days for a peer, 10-14 days for a senior contact), what the second message looks like (shorter than the first, adds one new piece of information or social proof, makes the ask easier — a yes/no question instead of a request for time), the third-touch rule (one only, then they go in the follow-up queue for next quarter — no fourth message), and the disqualifier rule (if they reply with anything that isn’t a clear no but isn’t a clear yes, what’s my next move). Output as a 4-step protocol with the templated lines for touches 2 and 3. The rule: never let a non-reply die in the inbox — but never become the founder who sent five messages to a peer who has already decided not to engage.”
4. The cold-expansion plan — only after the warm circle has produced data. “Don’t go cold until the warm circle has produced its first data — at least 10 real conversations and ideally one paying customer. Once it has: based on the segment from M2, the funnel from M7 and the warm-circle conversion data I now have, write the cold-expansion plan. Three channels max — pick the ones where my segment is most concentrated and most reachable (likely candidates: LinkedIn outbound, segment-specific Slack/Discord/community, one targeted Meta or X campaign pointed at my MVS). For each channel: the audience, the daily/weekly volume cap (e.g. 10 personalised LinkedIn messages per day, not 100 templated), the message variant I’m testing, the conversion target from contact to call, the kill criteria (what level of underperformance makes me stop this channel). Output as a 3-row table. Cold without data from warm is guessing — the warm-circle results tell me which variant works before I burn cold reach on the wrong message.”
5. The qualifying question that ends the call fast. “On a 15-30 minute discovery call or DM exchange with a warm or cold lead, the most expensive thing I can do is treat every conversation as a sales conversation. Most should end inside 5 minutes — either with a clear next step or a clear no. Draft the one qualifying question I open every conversation with — designed to surface in 60 seconds whether this person is in segment, has the budget, has the urgency, and can decide for themselves. Voice: peer, not interrogator. Examples to start from (rewrite for my offer): ’before we go further, two quick questions — are you already working on [problem the offer solves], and if I showed you the right thing in the next 20 minutes, are you in a position to act on it this quarter?’ Output the one question for my specific offer, plus three variants for the three highest-leverage sub-segments. The point is not to push — the point is to release the conversations that aren’t real, fast, so the real ones get more of me.”
6. The three objections, handled without flinching. “For a $497 / $2,997 / $5,997 offer to a corporate professional in my segment, the three most common objections are almost always: (1) price — ’that’s more than I planned to spend right now’, (2) timing — ’I’m interested but not until [Q3 / after the project ends / next year]’, (3) scepticism — ’how is this different from the dozens of other AI / coaching / cohort offers I see every week’. For each objection, draft (a) the wrong response (the apologetic, the discount, the over-explanation, the defensive), (b) the right response (one sentence that names the real concern under the surface objection, one sentence that earns trust with a specific number or named partner from the credibility stack, one sentence that puts the decision back in their court without pressure), (c) the line I say only if they push past that (a clear option, not a discount — e.g. ’the next cohort opens in October, I’d rather have you in that one than rush this one’). Output as a 3-row table. None of the right responses contain a discount. Discounts trained into the conversation in the first month of selling are discounts I’ll be giving for the next five years.”
7. The closing line I can say out loud without flinching. “The hardest sentence in M8 isn’t the pitch — it’s the close. Most founders pitch beautifully, then ask for the sale in a hedged, apologetic, conditional sentence that signals to the buyer that even the founder doesn’t believe in the price. Draft the one closing line — direct, specific, calm — that I can say at the end of a call where the buyer has indicated interest. Structure: one sentence that restates what they get, one sentence that names the price clearly, one sentence that asks the binary question (yes / not yet, never ’what do you think’). Voice: from M6 rulebook, peer-to-peer. Examples to rewrite: ’so what we just talked through is [outcome], the investment is $2,997, and the next cohort starts [date] — do you want me to send the enrolment link tonight, or would you rather sit with it for 48 hours and reply by Friday?’ Output the closing line for each tier — Toolkit, Accelerator, Founder Lab. Read each one out loud. If any one of them makes me wince, rewrite until I can say it without changing my tone.”
8. The case-study capture, the day they pay. “The five paying strangers aren’t five sales — they’re five case studies in waiting, and the time to capture the raw material is the week of purchase, not six months later when the memory has decayed. For each of the first five buyers, design the case-study capture: a 5-question form sent in the first-purchase confirmation email or first onboarding call. Questions: (1) what specifically were you trying to solve when you found this, (2) what alternatives did you consider and reject, (3) what almost stopped you from buying, (4) what was the one sentence on the page or call that closed it for you, (5) who else in your circle is dealing with this same problem. Output the 5 questions in the wording that fits the M6 voice. The output of this form across the first 5 buyers becomes — directly — the testimonial bank, the objection-handling library, the headline rewrites and the referral list. The case study is a by-product of the sale, captured at the right moment or lost forever.”
9. The referral ask that doesn’t feel transactional. “A satisfied first-five buyer is the highest-quality acquisition channel in M8 — but only if I ask, and only if the ask doesn’t feel like the moment the relationship turned commercial. Draft the one referral ask, timed for the moment the buyer has had the first specific win with the offer (not on day 1, usually around day 14-30 depending on the offer). Structure: one sentence that names their specific win in their words, one sentence that asks them to think of one specific person in their circle who is dealing with the same problem they were when they bought, one sentence that gives them the soft tool to make the intro easy (a forward-friendly paragraph they can paste, not a link with my company name on it). Voice: peer-to-peer. Reject any ’refer a friend and get $50’ mechanic — the segment hates it and it cheapens the offer. Output the referral ask, plus the forward-friendly paragraph they can use unchanged.”
10. The first-five debrief, before you move to scale. “Once five strangers have paid, stop selling for 48 hours and run the debrief. For each of the five buyers, capture in a single table: name, sub-segment, channel they came from (warm circle / referral / cold channel), days from first contact to purchase, the specific objection they raised, the specific line that closed them, the one friction in the buy-flow they hit, the one signal they gave that I should over-index on for buyer 6-20. Then answer the four questions that decide what M9 looks like: (a) which channel produced the highest-quality buyer, not just the most volume, (b) which objection killed the most almost-buyers, (c) which line in the offer was the single biggest converter across all five, (d) what’s the one thing in the funnel I rebuild before scaling to the next 20. Output as the debrief table plus the four answers. The first-five debrief is the bridge from M8 to M9 — without it, scaling means scaling whatever was accidentally working, including the parts that won’t survive 5x volume.”