Momentum Product Co.
Signal to Strategy · Course resources

Session 2: Stress-Test the Bet

Everything from Session 2 of Signal to Strategy. The slides, the prompt pack, the skill, the full run we did live, and the Launchpad bet pack. Free, same as everything we run.

Session 1 was about making the signal visible. This one is about what you do when the signal is in and the decision still isn't obvious.

Session2 of 4
DateSeptember 16, 2026
CourseSignal to Strategy on Maven
Cohort access

Enter your cohort code.

Session resources are for Signal to Strategy cohort members. Your code is in the welcome email and pinned in the cohort Slack channel.

Not in the cohort? See the course →

Start here

Everything from Session 2.

Prompt pack

The eight prompts, in order, with the Launchpad versions you can paste as-is and what the run actually found. Start here if you want to just run it.

Download the PDF →

Session 2 slides

The full ninety minutes, including the Launchpad stress test.

Download the PDF →

The run itself

The actual conversation: every prompt, every reply, and the point where we pushed back on it.

Open the chat →

Launchpad bet pack

The four files we ran live, plus the prompt pack in markdown so you can load it straight into the chat.

Download the .zip →

The skill

Drop this into Claude and it runs the whole method on your own bet, without you pasting anything. Works in any chat once it's installed.

Download the .skill →
The four verdicts

Not a score. A verdict.

Take every piece of evidence in your write-up and give it exactly one verdict against the strategy you already committed to.

  • Confirms. The signal and the strategy point the same way. Useful, and the least informative of the four.
  • Contradicts. They point in opposite directions, or the bet quietly breaks a commitment you already made. This is the valuable one.
  • Flatters. It looks like support. It's only support because you framed the bet to match something you already believed.
  • Silent. Nothing says anything, and you've been filling the gap with assumption. Most silences are a phone call, not a quarter.

If everything lands in Confirms, that isn't validation.

That's a bet you framed to win.

The five moves

This is the whole loop.

Unchanged from the live run. Each one is a move, not a magic string: you're copying the reasoning, not the words. Run them in order, in one chat.

These are written against the Launchpad bet pack above. Upload all four files into one chat and they run as-is. Swap the bracketed parts for your own, or grab the prompt pack, which has both versions side by side.

00

Load the context

I'm about to make a bet and I want it stress-tested, not improved. Here's what's in this workspace: [my strategy doc / vision and principles], [the write-up of the bet], [the customer signal behind it], and [transcripts or notes showing how my stakeholders actually talk]. Read all of it before you say anything. Then tell me what decision you think I'm making and what I seem to be optimizing for. Don't recommend anything yet.

If the reply could be about any company, your context isn't specific enough. Fix it before you go on.

01

Restate

Before you critique anything: restate my bet in your own words. What am I actually optimizing for, what constraints did I give you, and what assumptions are you making that I did NOT state? List the assumptions separately, and make them specific: "the user wants to save money" is not an assumption, it's filler.

The move most people stop halfway through: take a surfaced assumption, tell it that assumption is wrong, and re-run. One corrected assumption can flip the whole answer. You can't correct an assumption you can't see.

02

Sort it against the strategy

Now go through my strategy document line by line against this bet. For every piece of signal in the write-up, and every claim the bet makes, give it exactly one verdict: CONFIRMS: the signal and the strategy point the same way CONTRADICTS: they point opposite ways, or the bet breaks a commitment I already made FLATTERS: it looks like support, but only because the bet was framed to match something I already believed SILENT: the strategy and the data say nothing, and I've been filling the gap with assumption For each one, quote the specific line or number you're relying on. Put the bet's central claim in CONTRADICTS or FLATTERS if that's where it belongs, don't protect it. Then tell me, in one paragraph, the trade-off I've been avoiding naming.

This is the move that isn't in a generic pushback loop, and it's the one that matters. The highest-value output is a commitment in your own strategy doc that your bet quietly breaks.

03

Constrain

Re-run this against my real limits: [time], [budget], [headcount], [technical or compliance constraints], [what's already committed]. If something has to give, tell me explicitly what you'd trade and why that trade rather than another. Then name one thing I should NOT trade even under pressure, and why.

Push back if it quietly drops a constraint: "You ignored the headcount limit. Stay inside the constraints or tell me which one to relax, don't pretend it's free."

04

Compare

Give me the top three paths side by side, not just the winner, including the smallest possible version and doing nothing. For each: what it's best at, what I give up, who it's wrong for, and the one risk I'd be signing up for. No straw men, make the honest case for each. Then tell me which one you'd pick and the single fact that would change your mind.

If you'd never pick option B, it isn't a comparison.

05

Challenge

Argue against your own recommendation. Build the strongest good-faith case that I should do the opposite, and the strongest case that I should do nothing and wait two quarters. What would have to be true for this to be the wrong call? What's the single piece of information that, if I gave it to you, would change the answer? And what option have you been underweighting this whole time?

The move everyone skips. If it caves the moment you push, that's a tell, not an insight: "Don't just agree with me. If the original call was right, defend it."

06

The stakeholder lenses

Here are the notes and transcripts from [last quarter's business review / the architecture review / the design critique]. Read how each person actually talks: what they measure, what they're tired of hearing, the tests they apply out loud. Then, for each of [CEO], [CFO], [engineering lead] and [a peer in my own discipline], give me the three hardest questions they will ask about this bet, in their own register, using their own numbers. Quote them where the transcript gives you the hook, never invent a quote. For each question, tell me honestly whether the bet as written can answer it.

"Pretend you're a skeptical CFO" gets you a stock CFO. Your CFO is more specific and much harder. Once a lens sounds like the real person, tell your AI to package it as a skill and it's reusable next quarter.

07

The validation-test canvas

Before I commit anything to this, draft the cheapest tests that could kill it. Order them by how much risk they retire per dollar, not by what's interesting to build. For each one give me exactly: the assumption it tests, the test itself, what result would KILL the bet, what result would greenlight it, the cost and elapsed time, who runs it, and a decide-by date working back from [the date the decision locks]. At least two must be runnable in under a day. At least one must use data or documents we already have rather than talking to customers. And tell me which tests are only worth running if an earlier one comes back a particular way.

The rule underneath all of it: if no result could kill the bet, it isn't a test, it's a demo. Write the kill condition first.

Install it as a skill

The method, installed.

The prompts are the method written out. The skill is the method installed: once it's in, you say "stress-test this bet" and it runs the whole thing on your own material.

Download stress-test-the-bet.skill above. In Claude, go to Settings → Capabilities → Skills and upload it, or drag the file into a chat and Claude will offer to install it. Then in any chat: give it your bet and your strategy doc and ask it to stress-test the bet.

It works on whatever you have.

No strategy document is the common case. A vision statement, last quarter's goals, or the list of things you agreed not to do this year all count, and the skill will tell you what it can and can't conclude from what you gave it.

The reusable block

Keep this one in your notes.

Paste it into any chat, after any first answer, on any decision.

Any decision · any chat

Pressure-test it properly

Before I act on this, pressure-test it properly: RESTATE my decision in your own words and list the assumptions you're making that I didn't state. SORT every piece of evidence against my strategy: confirms, contradicts, flatters, or silent. Quote the line you're relying on. Then name the trade-off I've been avoiding. CONSTRAIN: re-run it against my real limits [list them] and tell me what you'd trade, and one thing I shouldn't. COMPARE your pick against the top two alternatives and doing nothing. Honestly, no straw men. CHALLENGE: argue the strongest case against your own recommendation, and tell me the single fact that would change it. Ground every claim in something I gave you. Never invent a quote or a number. If you need something you don't have, say it's missing.
This week's project

Before next Wednesday.

  1. Run the full loop on your real bet. All five moves, with your own stakeholder transcripts loaded as lenses.
  2. Ship one validation test. The cheapest one on your canvas. Actually run it this week, not design it, run it.
  3. Post the trade-off in the cohort channel. One sentence. The thing you'd been avoiding naming.

The signal tells you what's true.

It doesn't tell you what you're willing to give up. That part is still yours.

Momentum Product Co. · Heather Gawel + Babajide Okusanya · Launchpad is fictional. For course use only.