# Estimate the impact of a feature before we build it

A transparent back-of-envelope impact estimate with ranges, assumptions, and the one number that matters most.

- Topic: Thinking & Decisions
- Framework: Few-Shot
- Level: Standard
- Tags: Research, Fill-in Template
- Source: https://www.thinkwithprompt.com/prompts/estimate-the-impact-of-a-feature-before-we-build-it

## Prompt

```text
Help me estimate the likely impact of a feature before we commit to building it.

**Feature or bet:** [ONE-LINE DESCRIPTION]
**Metric it should move:** [e.g. trial-to-paid conversion, weekly active teams, support tickets]
**Baseline data I have:** [CURRENT VOLUMES, RATES, REVENUE PER USER - with dates]
**Who it reaches:** [SEGMENT, SHARE OF USERS, ENTRY POINT]
**Comparable evidence:** [PAST LAUNCHES, COMPETITOR DATA, RESEARCH - or "none"]
**Estimated effort:** [e.g. 2 engineers for 6 weeks]

Build the estimate as a driver tree: reach x adoption x effect per user x value per unit of effect. For each driver:
- State my input or a clearly labeled assumption.
- Give a low, likely, and high value, and why.

Then:
1. Calculate low / likely / high impact over 12 months, showing the math.
2. Identify which assumption the result is most sensitive to.
3. Suggest the cheapest way to test that assumption before building.
4. Compare impact to effort in one line, and flag if the low case would not justify the work.

Rules: never present an assumption as data. Mark every assumption [ASSUMPTION] and every figure I must check [CONFIRM]. Avoid false precision: round to sensible numbers. Output as a short table plus no more than 150 words of commentary.
```

## What to fill in

- ONE-LINE DESCRIPTION
- CURRENT VOLUMES, RATES, REVENUE PER USER - with dates
- SEGMENT, SHARE OF USERS, ENTRY POINT
- PAST LAUNCHES, COMPETITOR DATA, RESEARCH - or "none"
- e.g. 2 engineers for 6 weeks
- ASSUMPTION
- CONFIRM

Free to copy and use. From ThinkWithPrompt (https://www.thinkwithprompt.com/), a library of 2,663 curated AI prompts.