The prompt I had planned for a client's mock last month was burned before we started. He told me at the top of the call he had run it with another coach the night before. I want to walk through what happens around a mock with me, because a fair amount of it is AI now and I would rather say exactly where than let you guess. The format first. The client gets a prompt they have never seen, a few minutes to think, then they answer as if it were the real thing. I do not model answers. If they stall I ask one question and go quiet, and I score each step of the answer by how many questions it took to get them through it, zero to four. Zero means they found it alone. Before the call, a system I built in Claude reads the prompt I plan to use and checks it against everything that client has done with me, because a prompt they have already worked anywhere, with anyone or any tool, is burned. That check is what failed last month. It only knows what I know, and clients self-drill with other coaches and with AI more than they mention. So the burned question is now a sentence I say out loud in my opening script, and I always carry a spare prompt. The system also finds the trap in the prompt. Every product sense prompt has a neighbor, an easier question the obvious answer answers instead. "Design the purchase experience" invites pricing tiers. That neighbor is the second of the three places I score, so the system has to find it before I can score fairly. During the call it does nothing. If the client stalls, I count to thirty before I say a word. After the call it reads the transcript, scores each of the three places by how many questions it took, and drafts the feedback under a rule I will not relax. Every claim traces to something the client said on the recording. Then I edit it, because it has never sat in the room with them. Two of my questions got rewritten in August because the transcripts showed them misfiring. "What's the noun in the prompt?" turned into a grammar debate. It became "Who is this prompt about, and what do they go through?"

I want to share something with this group before I talk about it anywhere else. If you have ever gotten "be more strategic" in a review, you know how useless it is as feedback. I heard it handed to clients for years and I could never tell them what to actually do with it, because it is an adjective, and you cannot practice an adjective. So I started keeping track of what the PMs with real pull in their orgs were doing differently. It eventually settled into five behaviors. They propose a direction before anyone hands them one, even when they might be wrong. They change how the room sees the problem instead of just answering the question as asked. They say what they are choosing against out loud, before someone makes them. They hold engineering, design, sales, and finance in one plan instead of winning each conversation separately and hoping it adds up. And their framing is simple enough that it travels without them in the room. I ended up calling it ORBIT (Orient, Reframe, Balance, Integrate, Transmit). I have been using it privately in coaching for a while, and today I am starting to write about it publicly. Here is the thing I keep noticing, though. Almost everyone is naturally strong on one or two of these, and blind to at least one. And the blind spot is usually there because nobody ever told you that part was your job too. So this week's question: which of the five is your weakest right now? Proposing a direction, reframing the problem, naming your trade-offs, holding the functions in one plan, or making your framing travel. And what did it cost you the last time it mattered? Specific stories beat agreement. I will go first in the comments.