ICP Validation in 7 Days

Produce a testable buyer hypothesis and a seven-day field test. Treat a polished persona as a hypothesis until real buyers support it. This sprint lays the groundwork for a 30-day first-meeting goal; research conversations do not automatically count as sales meetings.

Find a buyer the user can serve

Use their delivery skills, past paid work, access to people, preferred channel, constraints, and any interview notes. If several markets are plausible, compare at most three on evidence of pain, funding, reachability, and delivery fit. Select a provisional test with reasons; let the user override the choice. Do not confuse being an AI freelancer with selling to AI freelancers.

Describe the person who pays or authorizes the work, their operating situation, painful workflow, desired result, and likely purchase trigger. For an internal operator, identify the sponsor and budget or resource approval path.

Write: I help [specific buyer] get [concrete outcome] without [unwanted effort or risk].

Make five observable disqualifiers relevant to this buyer. Start from inability to fund, wrong problem, no viable authority path, unreachable channel, and service expectations that do not match budget. Do not reject every multi-person buying process: distinguish a real approval path from an inaccessible buyer. Label a guessed budget threshold as provisional.

Run ten conversations in seven days

Start warm on days 1-2, approach matching buyers on days 3-5, and complete missing conversations and score evidence on days 6-7. Aim for ten distinct people, including five who fit the hypothesis beyond the closest warm circle. Make the calendar fit available access. Ten invitations are not ten conversations. If fewer people respond, report the actual sample and leave the verdict provisional.

Draft a warm invitation and a cold-ish observation plus one question. These are learning conversations. Do not disguise a sales pitch as research.

Useful questions from the deck:

• What triggered the last time you bought help with this?
• What almost made you say no?
• How are you handling the problem today?
• What have you already tried?
• If this gets worse over the next 90 days, what breaks?
• Who else should I speak with?

Ask about how purchases or internal approvals happen when it fits the conversation. Capture what the buyer actually says rather than supplying the desired answer.

Capture and score

Create a blank ten-row log, or fill it only from real supplied evidence:

Person ID | Date | Role/situation | Warm or cold-ish | Exact pain quote | Current workaround | Trigger | Funding/authority evidence | Reachable channel | Contradiction | Next action

Use anonymized IDs in shareable output. Preserve exact quotes separately from paraphrases. Count a repeated theme by distinct people, not by the number of times one person repeats it. For each theme, show the person IDs supporting it. Distinguish an explicit budget or previous purchase from a proxy such as company size.

Apply the deck's decision rules after ten talks:

• KEEP: the same pain appears in at least four distinct conversations, credible funding evidence exists, and matching buyers can be reached weekly. This supports a next sales test, not proof of product-market fit.
• REWRITE: pain is real but the buyer title, company size, trigger, or boundaries were wrong. Change the sentence and rerun the test.
• KILL: people like the idea but cannot fund it, or the only workable channel is unavailable or unacceptable to the user.

If the evidence does not distinguish those paths, say INCOMPLETE and name the missing evidence. Never manufacture a KEEP to fit a deadline. Recommend the verdict, explain uncertainty, and ask the user to choose whether to adopt it. Revisit any disqualifier contradicted by actual buyers.

Deliver

• ICP sentence, buyer situation, and five disqualifiers.
• Evidence versus assumptions, including the most important uncertainty.
• Seven-day plan, two invitation drafts, and interview questions.
• Blank or evidence-filled conversation log.
• When notes exist: theme counts, counterevidence, verdict, revised sentence, and one quote that changed the recommendation.
• Next sales test: who to approach and what would justify a qualified meeting.

Working agreement

Use context already supplied. Ask only for missing facts that change the next decision, preferably in one short batch, then make a useful first draft. Label assumptions and distinguish buyer evidence from your interpretation. Never invent interviews, prospect facts, results, testimonials, budget, or meetings. A target date is a planning goal, not a promised commercial outcome. Use plain sentences and normal hyphens in generated copy.

Create drafts and plans within the requested scope. Sending messages, booking meetings, publishing, changing a CRM, or committing the user to commercial terms requires authorization for that action. A request for advice alone does not authorize execution. Work from pasted notes if tools are unavailable; identify what remains unverified.

Source

Agent J Sales 101 week1-icp-deck.pdf, pages 3-10 and 12; curriculum-deck.pdf, page 4. The deck credits Josh Braun for the without formula and illumination questions. These are practical adaptations by John Peslar AI, not endorsements by the referenced educators. The skill is self-contained; the original decks are not required at runtime.