Stress-test a strategic narrative against likely questions

advancedClaude OpusExecutiveCommsframeworkmethodologycommunicationsstrategystorytelling

Use case

Use this before a board update, fundraise, or all-hands. The structure simulates the three audiences with different priors and surfaces the questions the narrative cannot answer — which is where it will break in the real meeting.

The prompt

You are stress-testing a strategic narrative. The goal is to find the questions that would expose its weak points BEFORE the audience does. Three audiences ask different questions; the narrative has to hold up to all three.

<context>
The narrative (or pitch) being tested:
<<<
{{narrative}}
>>>

Audience 1 (investor or board): {{investor_context}}
Audience 2 (employees or team): {{employee_context}}
Audience 3 (skeptical outsider — analyst, journalist, or a competitor's friend): {{skeptic_context}}
Stage of the company: {{stage}}
</context>

<task>
Step 1 — Read the narrative back.
In 3 to 5 sentences, state what the narrative is actually claiming. Separate the claims from the framing.

Step 2 — Generate questions per audience.
For each of the three audiences, generate 5 to 7 questions they would ask. Each question should be specific and uncomfortable. Lazy generic questions ("what's your biggest risk?") do not count. Tie each question to a specific claim or omission in the narrative.

Step 3 — Score the narrative's answer.
For every question, rate how well the current narrative answers it (1 = no answer / dodges, 5 = answers cleanly with evidence). Note where the answer is in the narrative, or note that the narrative does not address it.

Step 4 — Identify the weak points.
List the 3 to 5 questions that scored 1 or 2 across any audience. These are the holes that will get exposed.

Step 5 — Propose narrative repairs.
For each weak point, propose a specific edit to the narrative — either an added sentence, a reframing, or an honest acknowledgment with a plan. Repairs must not invent facts; if the underlying answer requires data the user does not have, name what data is needed.

Step 6 — Audience-specific framing notes.
Note 1 to 2 places where the same fact should be framed differently for different audiences (without contradicting itself).
</task>

<output_format>
## Narrative claims, separated from framing
## Investor / board questions (with score)
## Employee questions (with score)
## Skeptic questions (with score)
## Weakest answers (the questions that will hurt)
## Proposed narrative repairs
## Audience framing notes
</output_format>

<constraints>
- Questions must be specific. "What's your moat?" is generic; "Your moat claim rests on 18 months of data; what changed in the market in month 6 that the data doesn't capture?" is real.
- Do not invent answers. If a repair requires a number you do not have, say "needs: revenue retention by cohort 2023-2024" and stop.
- Score honestly. If half the answers are 4s and 5s, you are channeling a generous audience. Make the skeptic harder.
- Do not let the three audiences ask similar questions. If they do, you are writing one audience three times.
</constraints>

Variables

{{{{narrative}}}}Replace with your {{narrative}}
{{{{investor_context}}}}Replace with your {{investor context}}
{{{{employee_context}}}}Replace with your {{employee context}}
{{{{skeptic_context}}}}Replace with your {{skeptic context}}
{{{{stage}}}}Replace with your {{stage}}

Example

Input

narrative: "We're the AI-native CRM for sales teams who hate Salesforce. We've grown to $4M ARR in 18 months, our customers see 3x faster pipeline velocity, and we're raising to expand to mid-market."
investor_context: Series A round, lead is enterprise SaaS focused, has portfolio companies adjacent to the space
employee_context: 22 people, half hired in last 6 months, equity heavy comp
skeptic_context: Analyst who covers CRM and has seen 6 "Salesforce killers" fail
stage: Series A in progress, post product-market fit

Output

## Narrative claims, separated from framing
Claims: $4M ARR; 18-month timeline; "3x pipeline velocity" outcome; mid-market expansion as next step. Framing: "AI-native," "hate Salesforce."

## Investor questions (with score)
- "$4M ARR in 18 months — what's the breakdown by ICP, and what's net retention by cohort?" → 1, narrative says nothing
- "3x pipeline velocity is a customer claim — how do you measure it, and on what sample size?" → 2, no methodology stated
- "Mid-market is a different motion than your current customers — what proof points do you have that this expands?" → 1, asserted, not shown
- "Salesforce has 1500 engineers building AI features — what's defensible about your 22-person version 24 months from now?" → 1, hand-waves
- "What's your gross margin and CAC payback?" → 1, absent
...

## Employee questions (with score)
- "We hired half the team for the SMB motion — does mid-market expansion mean that work was wasted?" → 2, narrative is silent
- "Equity grants assume a much higher exit; what's the path to a real outcome at this market position?" → 1
- "If we lose, it'll be because Salesforce caught up. What's the plan if they do?" → 1
...

## Skeptic questions (with score)
- "Six AI-native CRMs have raised money in the last 24 months. What makes you the one that doesn't end up as a feature inside HubSpot?" → 1
- "'AI-native' is a positioning, not a moat. What's the actual moat?" → 1
- "3x velocity vs. what baseline — Salesforce, manual, no system?" → 2
...

## Weakest answers
1. Methodology behind the 3x velocity number
2. Defensibility against Salesforce + HubSpot adding AI
3. Net retention by cohort (or any unit economics signal)
4. Why mid-market is the next move and not deeper SMB or vertical SaaS

## Proposed narrative repairs
- 3x velocity: replace with "across 47 customers, median pipeline-to-close time dropped from 38 to 22 days; methodology and sample available." Needs: confirm the methodology actually holds across customers.
- Defensibility: add explicit "our edge is data structure, not model access — every customer's CRM gets a labeled outcome trail Salesforce cannot generate retroactively." Needs: confirm this claim is technically real.
- Mid-market choice: add "we hit a feature ceiling for SMB at ~$50k ACV; mid-market is our existing customers growing into us, not a new motion."
- Net retention: include the number, broken out by cohort, even if it is mixed. Hiding it loses more credibility than showing it.

## Audience framing notes
- The "Salesforce killer" framing plays well to investors ("big TAM") but cuts the wrong way with employees ("we lose if they catch up"). For the all-hands, lead with "our customers, our data" framing instead.
- Net retention numbers go in the deck for investors and skeptics; for employees, the relevant cut is "we are growing customers we already have" — same fact, different framing.

Tips for best results

1The three-audience structure is what makes this stronger than the typical 'rehearse the pitch' prompt — without it, you over-prepare for one audience and get caught by another. Each audience has different priors and asks different questions.
2If the skeptic's questions are not uncomfortable, you have not made the skeptic skeptical enough. Add specific context about what they have seen go wrong.
3Score honestly. A narrative full of 4s and 5s is one you have not actually stress-tested. If your gut says the meeting will be hard, the scores should reflect that.
4The 'needs: data' notes are the actual deliverable. They tell you what to go gather before the meeting, which is more valuable than rehearsing answers.
5Pair with devils-advocate on the strategy itself. Strategic narrative test makes the story durable; devil's advocate makes the strategy durable.

Related prompts

Devil's advocate — strongest case against a plan, then triage

advanced

Argue the strongest case against a proposed plan, then identify which of those arguments are decisive versus which are just noise.

Executiveframeworkmethodologystrategy

Executive summary from a long document

intermediate

Compress a long document into a structured executive summary with explicit decisions, action items, and open questions — not a paraphrase.

Executiveframeworkmethodologyexecutive-comms

Use Claude as a skeptical VC (diligence-grade questions, not generic feedback)

advanced

Frame Claude as a skeptical VC running diligence — pushing on TAM, durability, defensibility, founder-market fit, and capital efficiency.

Founders & Startupfounderfundraisingvc

Need help implementing this prompt in your workflow?

Book a call

You are stress-testing a strategic narrative. The goal is to find the questions that would expose its weak points BEFORE the audience does. Three audiences ask different questions; the narrative has to hold up to all three. <context> The narrative (or pitch) being tested: <<< {{narrative}} >>> Audience 1 (investor or board): {{investor_context}} Audience 2 (employees or team): {{employee_context}} Audience 3 (skeptical outsider — analyst, journalist, or a competitor's friend): {{skeptic_context}} Stage of the company: {{stage}} </context> <task> Step 1 — Read the narrative back. In 3 to 5 sentences, state what the narrative is actually claiming. Separate the claims from the framing. Step 2 — Generate questions per audience. For each of the three audiences, generate 5 to 7 questions they would ask. Each question should be specific and uncomfortable. Lazy generic questions ("what's your biggest risk?") do not count. Tie each question to a specific claim or omission in the narrative. Step 3 — Score the narrative's answer. For every question, rate how well the current narrative answers it (1 = no answer / dodges, 5 = answers cleanly with evidence). Note where the answer is in the narrative, or note that the narrative does not address it. Step 4 — Identify the weak points. List the 3 to 5 questions that scored 1 or 2 across any audience. These are the holes that will get exposed. Step 5 — Propose narrative repairs. For each weak point, propose a specific edit to the narrative — either an added sentence, a reframing, or an honest acknowledgment with a plan. Repairs must not invent facts; if the underlying answer requires data the user does not have, name what data is needed. Step 6 — Audience-specific framing notes. Note 1 to 2 places where the same fact should be framed differently for different audiences (without contradicting itself). </task> <output_format> ## Narrative claims, separated from framing ## Investor / board questions (with score) ## Employee questions (with score) ## Skeptic questions (with score) ## Weakest answers (the questions that will hurt) ## Proposed narrative repairs ## Audience framing notes </output_format> <constraints> - Questions must be specific. "What's your moat?" is generic; "Your moat claim rests on 18 months of data; what changed in the market in month 6 that the data doesn't capture?" is real. - Do not invent answers. If a repair requires a number you do not have, say "needs: revenue retention by cohort 2023-2024" and stop. - Score honestly. If half the answers are 4s and 5s, you are channeling a generous audience. Make the skeptic harder. - Do not let the three audiences ask similar questions. If they do, you are writing one audience three times. </constraints>

Example

Input

narrative: "We're the AI-native CRM for sales teams who hate Salesforce. We've grown to $4M ARR in 18 months, our customers see 3x faster pipeline velocity, and we're raising to expand to mid-market."
investor_context: Series A round, lead is enterprise SaaS focused, has portfolio companies adjacent to the space
employee_context: 22 people, half hired in last 6 months, equity heavy comp
skeptic_context: Analyst who covers CRM and has seen 6 "Salesforce killers" fail
stage: Series A in progress, post product-market fit

Output

## Narrative claims, separated from framing
Claims: $4M ARR; 18-month timeline; "3x pipeline velocity" outcome; mid-market expansion as next step. Framing: "AI-native," "hate Salesforce."

Tips for best results

1The three-audience structure is what makes this stronger than the typical 'rehearse the pitch' prompt — without it, you over-prepare for one audience and get caught by another. Each audience has different priors and asks different questions.

2If the skeptic's questions are not uncomfortable, you have not made the skeptic skeptical enough. Add specific context about what they have seen go wrong.

3Score honestly. A narrative full of 4s and 5s is one you have not actually stress-tested. If your gut says the meeting will be hard, the scores should reflect that.

4The 'needs: data' notes are the actual deliverable. They tell you what to go gather before the meeting, which is more valuable than rehearsing answers.

5Pair with devils-advocate on the strategy itself. Strategic narrative test makes the story durable; devil's advocate makes the strategy durable.