Feature Spec / PRD

What this tool actually does

You're three Slack threads deep, the engineers want to know what 'works for teams' means, and the founder keeps adding 'oh also it should...' in DMs. The PRD lives in your head and it's not transferring. This tool turns a fuzzy feature idea into a spec engineers can actually build from: problem statement that explains why this is worth doing now, user stories with real acceptance criteria, the edge cases your team will otherwise hit in QA on day 9, success metrics that aren't 'usage goes up', explicit scope cuts for v1 vs v2 so you ship something, an open-questions list to surface before estimation, and a kill-criteria section that gives everyone permission to stop building if the data says so. Output is markdown that pastes cleanly into Notion or Linear without reformatting. Built for founders, solo PMs, and tech leads who hate writing PRDs but know nothing ships without one.

Example run

Brief used: “Feature: in-app referral program for our SaaS (project management tool, $29/user/mo, 800 paying teams). User clicks 'Invite friends' in nav, gets a unique link, both sides get 1 month free when the friend converts to paid. Target user: existing paid customer, team lead persona. Constraints: needs to ship in 3 weeks, only 1 backend engineer, can't change billing logic mid-month, fraud is a worry (Stripe testers gaming it).”

# Feature Spec: In-App Referral Program

**Owner:** [PM]
**Engineering:** [name]
**Status:** Draft → Ready for estimation
**Target ship:** [date, 3 weeks out]

## Problem statement
Our paid customers are our highest-converting source of new signups (anecdotal — word-of-mouth lands closing in onboarding chat), but we have no in-product way for them to do it. Today they share us in Slack, but there's no incentive, no tracking, and no signal back to us about who's recommending. If we don't build this, growth stays dependent on paid ads (CAC up 30% QoQ) and the organic motion stays invisible.

## User stories

**1. As a team lead on a paid plan, I want to invite peers to the product so I can get rewarded when they sign up.**
Acceptance criteria:
- Logged-in paying user sees 'Invite' entry in the main nav.
- Clicking opens a modal with a unique referral URL, a copy button, and a 'share via email' field.
- URL contains a stable user-scoped token (not the user id).
- Modal shows current state: invited count, converted count, credits earned.

**2. As a referred user, I want to land on a signup page that acknowledges the referrer so trust is established before I enter a credit card.**
Acceptance criteria:
- Landing URL accepts the token, validates server-side.
- Signup page shows 'Invited by [Referrer's first name]' (only first name, no email).
- Token persists across the signup flow even if the user closes and reopens the tab within 14 days.

**3. As the referrer, I want to get my month free automatically once the friend converts.**
Acceptance criteria:
- 'Converted' = friend subscribes to a paid plan AND remains paid for 7 days (not just trial start).
- Credit applied to referrer's next invoice as a line-item discount.
- Referrer gets an in-app notification + email when credit is applied.

**4. As an admin, I want a referral dashboard so I can see signups, conversions, and fraud signals.**
Acceptance criteria:
- Internal-only page lists referrals by referrer with status (pending / converted / flagged).
- Flag column shows automated checks: same IP, same email domain on free Gmail patterns, signup-then-immediate-paid (under 60 seconds).

## Edge cases checklist
- Referrer cancels their subscription before the friend converts → friend still gets their month free, referrer credit voided.
- Friend signs up but uses a different email than the invite email → still credit if the token matches.
- Friend signs up, downgrades to free, comes back later and goes paid → credit applied at the moment paid status holds 7 days.
- Same person signs up twice with different emails from the same IP → flagged, no credit.
- Referrer is on annual plan → credit converts to a $29 line-item discount, not 'free month'.
- Token tampering / scraping → server validates token signature; invalid tokens fail silently to /signup with no referrer.
- Friend signs up before clicking the link (already had account) → no credit.
- Mid-month signup → credit applied on next invoice, never current.

## Success metrics
- **Primary:** Referral-attributed paid signups as % of total paid signups. Target 8% within 60 days of launch.
- **Secondary:** Activation rate of referred users (must be ≥ baseline activation; if lower, the referral channel is unhealthy).
- **Operational:** Fraud flag rate < 5% of approved credits.
- **Anti-metric:** Total credit dollars given out vs MRR from referred users — credits must stay under 25% of attributed MRR.

## v1 scope (ships in 3 weeks)
- Nav entry, modal with link + copy + email-share field.
- Token generation and signup-side attribution.
- Conversion logic (paid + 7 days hold).
- Credit application as invoice line item.
- In-app + email notification on credit applied.
- Admin dashboard, basic.

## v2 (next quarter)
- Tiered rewards (refer 3 → bigger reward).
- Public leaderboard.
- Custom referral landing pages per referrer.
- Integration with Twitter/LinkedIn share intents.

## Open questions (resolve before estimation)
1. Does 'paid for 7 days' include trial? Assumed no — but confirm with finance.
2. Annual plan referrers — credit is $29 flat, not 'one month equivalent'. Confirm.
3. What's the email-share copy? Default template needed; assume the marketing team can write 3 versions.
4. Do we need a terms-of-program page for legal? Probably yes — ask counsel before launch.
5. Existing free-trial signups: are they 'converted' for referral purposes? Assumed only paid counts.

## Kill criteria
Kill or pause the feature if, at the 60-day mark:
- Referral-attributed signups are < 3% of total paid signups (signal: not enough lift to justify the credit cost).
- Fraud flag rate > 15% (signal: gaming faster than we can fix it).
- Activation rate of referred users < 50% of baseline (signal: the channel brings low-quality users).

If two of three trigger, the feature ships to a 'hidden until invited' state and we run the v2 redesign instead of expanding.

Static example — your run uses Claude live on your specific brief.

Who it’s for

Solo founders writing the first PRD for a feature their two engineers will build next sprint, head-of-product hires inheriting a team that's never seen a real spec, tech leads who keep getting 'just one more thing' requests because the scope was never written down. Not for: enterprise PMs who need stakeholder sign-off workflows or compliance fields — this is a working doc, not a process artefact.

What you get on the paid tier

A complete PRD in markdown ready to paste into Notion or Linear: (1) problem statement that frames why this is worth building and what changes if you don't, (2) 3-5 user stories in 'as a / I want / so that' format with acceptance criteria for each, (3) edge cases checklist — empty states, failure modes, permission boundaries, the stuff QA will catch otherwise, (4) success metrics defined as measurable thresholds, not vibes, (5) scope-cut suggestions splitting v1 (what ships in 2 weeks) from v2 (what waits), (6) open questions list to resolve before estimation, and (7) kill criteria — the conditions under which you stop building this. Plus a header block with feature name, owner field, status, and target ship date.

Common use cases

How it works

1. Describe your input. One or two sentences about your startup, audience, or product. The more specific you are, the sharper the output.
2. Get an instant preview. Claude Haiku generates a free sample in ~30s. No signup, no card.
3. Unlock the full version. Drop your email for 3 full-length runs/month, or pay $29 one-time for unlimited.

FAQ

Founders who use this tool also try