Six rules for every answer
- Concede fast where the answer is no.
- Never improvise on data, permissions, security, dates or money. Data and legal questions go to our team in writing (only Karl Foster gives written assurances). No launch, delivery or roadmap date, ever.
- Never invent social proof. There are no customer firms yet. Never "most firms we talk to", "the pattern we see" or "people tell us".
- Do not name a competitor first. Never repeat a second-hand competitor figure.
- Label every illustration: "Worked illustration, and I'm labelling it as exactly that, because I'm not going to pretend this is a forecast."
- Every answer ends in a question: a calibrated "what" or "how", or "Does it make sense to [next step]?" If not: "What's a good next step then?" The standard first next step: "Thirty minutes. We'll show you Holly drafting for someone in your role, on a demo workspace, and you can judge it for yourself."
Method: for brush-offs (1 to 8) use ledge, disrupt, ask, under 20 seconds, and stop after two noes. For real objections, label only where the buyer is plainly feeling something ("It seems like...", "It sounds like..."), pause, then ask a calibrated question. Run the AI accusation audit once, early: "You're probably thinking this is another AI tool that'll post generic stuff under your name and embarrass you with clients. And that your team will use it for a fortnight and stop." For indecision use JOLT: judge, offer one recommendation, limit the options, take risk off; never add urgency. Each group below has a probe question for when the stated objection does not feel like the real one.
AMBER entries (depend on a pending decision or caveat): 13, 33, 36, 38, 44, 63, 67 (contract sentence), 68 (pilot route), 69 (contract shape), 77, 82, 85, and Appendix A rows marked AMBER.