uplift sunUplift.page
← library
promptexample · tested🧰 Engineering practicev1 · published 2026-10-15 · updated 2026-11-03

Pressure-test a product plan

Challenge a feature request, find the real job behind it, and cut it to a one-week version with metrics.

By Uplift Duo

Run it as a prompt

Paste this into any AI agent, or fetch it: curl -s https://uplift.page/api/v1/prompts/product-plan-pressure-test/raw

prompt.md
# Pressure-test a product plan

Before anyone writes code, find out whether this request is the right thing to build and what the smallest useful version is.

## Context
- Request or problem: [PASTE THE REQUEST]
- Product and who uses it: [PRODUCT AND USER]
- Constraints (time, team, budget, what must not break): [CONSTRAINTS]

## Steps
1. Restate the job the user is trying to get done in one sentence, from their side, not the feature's.
2. List what the current framing gets wrong or leaves out: hidden assumptions, users it ignores, problems it just moves elsewhere.
3. Describe the ambitious version: what would make users recommend it. Keep it coherent, not a wish list.
4. Cut that down to the smallest version that delivers real value and ships in one week. Name what is deliberately left out.
5. List the top 3 risks (user, technical, business) with one cheap way to test each before building.
6. Pick 1-3 metrics that would show within 30 days whether it worked, each with the threshold that means "keep going".

## Rules
- Be direct. Disagree with the request when the evidence points that way.
- Label every assumption as an assumption.
- If a key fact is missing, ask at most 3 questions first, then continue with stated assumptions.

## Output
A one-page memo with these headings: Job to be done, What the framing misses, Ambitious version, One-week version (in / out), Risks and how to test them, Success metrics, Recommendation (build / reshape / drop, one paragraph).

Done when every risk has a test with a pass/fail threshold, every metric has a number and a date, and the recommendation names the first task to start.

The full prompt

Before anyone writes code, find out whether this request is the right thing to build and what the smallest useful version is.

Context

  • Request or problem: [PASTE THE REQUEST]
  • Product and who uses it: [PRODUCT AND USER]
  • Constraints (time, team, budget, what must not break): [CONSTRAINTS]

Steps

  1. Restate the job the user is trying to get done in one sentence, from their side, not the feature's.
  2. List what the current framing gets wrong or leaves out: hidden assumptions, users it ignores, problems it just moves elsewhere.
  3. Describe the ambitious version: what would make users recommend it. Keep it coherent, not a wish list.
  4. Cut that down to the smallest version that delivers real value and ships in one week. Name what is deliberately left out.
  5. List the top 3 risks (user, technical, business) with one cheap way to test each before building.
  6. Pick 1-3 metrics that would show within 30 days whether it worked, each with the threshold that means "keep going".

Rules

  • Be direct. Disagree with the request when the evidence points that way.
  • Label every assumption as an assumption.
  • If a key fact is missing, ask at most 3 questions first, then continue with stated assumptions.

Output

A one-page memo with these headings: Job to be done, What the framing misses, Ambitious version, One-week version (in / out), Risks and how to test them, Success metrics, Recommendation (build / reshape / drop, one paragraph).

Done when every risk has a test with a pass/fail threshold, every metric has a number and a date, and the recommendation names the first task to start.

Served from the uplift.page library and refreshed within 5 minutes of every update.