Concept Red-Team Reviewer

How to use it

1. 1Drop the prompt into ChatGPT, Claude or Gemini.
2. 2Fill the bracketed inputs (concept, target, business case, stage).
3. 3Answer its clarifying questions — it attacks specifics, not vibes.
4. 4Iterate: ask it to argue the bull case back, then reconcile the two.

See it in action

Bear case (as a skeptic would say it): Peak & Pantry asks shoppers to pay a premium for ambient versions of items they already buy fresh a few steps away — the occasion is real, but the deli case next to it solves it better, fresher, and cheaper per serving.

Failure modes:
- Trial: 'shelf-stable meat' reads as a quality compromise at a premium price point.
- Repeat: entertaining is episodic; purchase frequency may not clear velocity hurdles.
- Margin: multi-component pack drives COGS and co-pack complexity.
- Operations: total shelf life is set by the weakest component.
- Retail: buyers may not know where it lives — deli, crackers, or seasonal.

Must-be-true assumptions (most fragile first):
1. Shoppers accept ambient charcuterie as occasion-worthy (assumption to validate).
2. The kit beats building a board from the deli case on convenience alone.
3. A placement exists where velocity survives outside the holiday peak.

Kill criteria to agree now: no clear home aisle after placement testing; repeat intent below your action standard in a home-use test.

Verdict: survivable IF the placement question resolves — run the cheapest test first.

Want me to design the cheapest test for assumption #1?

The prompt

Here's the start of the prompt. Download the free bundle for the full, ready-to-paste version — plus the installable Claude Skill and Custom-GPT instructions.

# Role
You are a veteran CPG red-teamer — a former innovation VP retained to attack concepts before companies invest in them. You do not improve ideas and you do not soften findings. Your value is precision: every attack names a mechanism, not a mood.

# Context I'll provide
- The concept: [CONCEPT — as currently written, however rough]
- Target consumer and occasion: [TARGET]
- Business-case logic: [BUSINESS CASE e.g. price point, channel, margin expectation, volume hope]
- Stage and decision at hand: [STAGE e.g. pre-pilot, seeking gate approval]
- The team's claimed winning insight (optional): [INSIGHT]
- Evidence gathered so far (optional): [EVIDENCE]

# Your task

Frequently asked questions

What is a red team review for a product concept?

A red team review is a structured attack on a concept by someone whose only job is to find how it fails — before the market does it for you. It surfaces the bear case, the failure modes across trial, repeat, margin, operations, and retail, and the assumptions that must hold. The output is not pessimism; it is a list of cheap tests and agreed kill criteria.

How is this different from the Innovation Brief Builder?

The Innovation Brief Builder constructs a concept — tension, RTBs, format, claims. This skill assumes the concept already exists and attacks it. It never rewrites or improves the idea, because mixing construction and criticism softens both. A strong sequence is brief first, red team second, then revise the brief yourself with the findings in hand.

Will it just tell me my concept is bad?

No — it ends with a verdict and conditions, not a mood. Every attack is tied to a mechanism, every fragile assumption gets the cheapest test that would settle it, and the kill criteria are specific results you agree on in advance. Many concepts leave a red team stronger, because the weaknesses were named early enough to fix or test cheaply.

Can I run this in any AI chat tool?

Yes. ChatGPT, Claude, Google Gemini, or any capable chat model will run it identically — the rigor comes from the prompt structure, not the model. Some teams keep it saved as a Custom GPT or a Claude Skill and make a red-team pass mandatory before any concept enters a gate meeting with a spend request attached.