Prototype Test Plan Builder

How to use it

1. 1Paste the prompt into ChatGPT, Claude, or Google Gemini.
2. 2Fill the bracketed inputs (prototype, what changed, risks, timeline).
3. 3Answer its clarifying questions before it drafts the plan.
4. 4Iterate: ask it to compress the stages to fit your deadline.

See it in action

Hypotheses for the Solstice lower-acid recipe (ranked by risk):
1. H1 — The raised pH keeps the culture stable and safe through target shelf life (kills the project if false).
2. H2 — Lower acidity reads as 'smoother', not 'flat', versus the current recipe.
3. H3 — Existing Solstice drinkers rate the new recipe at parity or better.

Stage plan:
Stage 1 (weeks 1-2) — Bench tasting: three acidity variants vs control, internal panel, blind.
Stage 2 (weeks 2-6) — Stability protocol: pH drift, culture viability and package pressure checks with your lab; flag the safety review to a qualified technologist.
Stage 3 (weeks 6-8) — Co-packer pilot batch plus a small consumer read with current drinkers.

Decision rules:
- Kill if any variant shows unstable pH drift or culture loss in Stage 2.
- Iterate if 'smoother' wins but aftertaste flags at bench — adjust and re-run Stage 1 only.
- Advance to pilot only with a Stage 1 winner and a clean Stage 2.

What this plan cannot tell you: real-world repeat rate and velocity — those need in-market data.

Want me to draft the Stage 1 tasting sheet and variant matrix?

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 senior product development lead for food and beverage companies who has taken dozens of prototypes from bench to pilot. You believe cheap tests come before expensive ones, every stage needs pass criteria agreed in advance, and a plan without kill rules is just a schedule.

# Context I'll provide
- Prototype and category: [PROTOTYPE]
- What changed vs the current or reference product: [WHAT CHANGED]
- Biggest risks or worries: [RISKS]
- Timeline to a decision: [TIMELINE]
- Testing resources: [RESOURCES e.g. lab, kitchen, co-packer pilot slot, panel access]
- Success definition (optional): [SUCCESS e.g. parity with current recipe, preferred by target consumers]

# Your task

Frequently asked questions

What is a prototype test plan in food and beverage R&D?

A prototype test plan is the sequence of evaluations a new recipe or product goes through between the bench and a launch decision — what gets tested, in what order, against what pass criteria, and what happens on a fail. A good plan runs cheap tests first, ties every stage to a decision, and includes explicit kill rules so weak prototypes die early instead of expensively.

How is this different from the Stage-Gate Review Deck Builder?

They are neighbors in the same process. This skill designs the tests that generate evidence — hypotheses, stages, pass criteria, kill rules. The Stage-Gate Review Deck Builder packages accumulated evidence into a gate decision for a committee. Use this one when planning the work between gates, and the deck builder when the gate date arrives.

What sample sizes will it recommend?

Directional ones, honestly labeled. It suggests workable panel and sample counts for bench and kitchen stages, states plainly what a small internal read can and cannot support, and tells you when a question — usually consumer preference or shelf-life validation — requires a research partner or an accredited lab rather than a bigger office tasting.

Will this work in any AI chat tool?

Yes. The prompt is model-agnostic and runs in ChatGPT, Claude, Google Gemini, or any capable chat model. Teams often save it as a Custom GPT or a Claude Skill so every project builds test plans with the same stages and decision rules, which makes plans comparable across the pipeline and easier to audit after launch.