Mira runs a premortem on a launch

Mira, solo founder of an 8-person seed-stage startup. The decision: launch a free tier in six weeks to drive signups. Prompting style: casual.

This is a complete run, prompt to artifact. Framework: Premortem. For the cast and the other journeys, see the Showcase index.

The situation

Mira’s board wants signup growth. A free tier is the obvious lever, and the team is excited, which is exactly when she gets nervous. She does not write a brief; she types what is in her head.

The prompt

/think-premortem "launching a free tier in 6 weeks to drive signups. nervous it
just cannibalizes our paid plan and that support gets swamped. small team."

That is the whole prompt. Casual, one situation, the two worries on her mind. The framework supplies the rest.

The output

Premortem Risk Register - Free-tier launch

Decision: Launch a self-serve free tier in 6 weeks to drive top-of-funnel signups. Intended outcome: A step-change in qualified signups that converts to paid, without starving the paid plan or the support queue. Horizon: 4 months after launch. Reversibility: One-way door in practice - pulling a free tier after launch is public and damages trust.

The failure, declared. It is four months after launch. The free tier shipped and the plan failed badly: signups tripled but paid net-new revenue went flat, two of our four engineers are now full-time on free-tier support and abuse, and the board read the signup chart as success while the cash chart said otherwise.

Top risks and what we will do. The three most likely ways this fails: (1) the free tier cannibalizes paid because the free plan is too generous - we will gate the two features paying customers value most and instrument the free-to-paid path before launch; (2) support and abuse swamp an 8-person team - we will cap free usage and ship self-serve docs before opening the doors; (3) vanity signups hide flat revenue - we will report free-to-paid conversion, not raw signups, from week one.

#	Cause of failure	Likelihood	Impact	Leading signal / tripwire	Mitigation	Owner	Kill criterion
1	Free plan is generous enough that paying customers downgrade	H	H	>5% of active trials choose free over paid in weeks 1-2; paid net-new flat while signups climb	Gate the top 2 paid-value features; instrument the free-to-paid funnel pre-launch	Mira	Paid net-new revenue below the pre-launch trend for 3 straight weeks, attributable to downgrades
2	Support + abuse load swamps the 8-person team	H	H	Support tickets per 100 free users above threshold by week 2; eng pulled onto abuse triage	Hard usage caps; self-serve onboarding + docs; basic rate-limiting at sign-up	Eng lead	More than one engineer-week per week on free support for a month with no path down
3	Raw-signup vanity metric masks flat revenue	M	H	Board deck leads with signups; conversion not yet instrumented at launch	Define and ship the free-to-paid conversion metric before launch; lead every update with it	Mira	Launch slips if conversion tracking is not live (it is a launch gate, not a follow-up)
4	The 6-week timeline forces an insecure self-serve flow	M	H	Auth/billing edge cases open in QA week 5; no security pass by week 4	Cut scope to a thin, secure path; freeze features at week 2; security gate at launch-minus-1-week	Eng lead	Security review not green by the launch-minus-1-week gate
5	”Free” attracts users with no fit, who never convert	M	M	Free cohort looks nothing like paying customers; week-4 activation low among good-fit users	Light qualification at sign-up; track activation by fit, not raw count	Mira	After 8 weeks, good-fit free-to-paid conversion is below the breakeven the model needs

Watch list. Brand read (“they went freemium, must be struggling”) - monitor sentiment, low likelihood. Free tier used for spam - standard rate-limiting likely sufficient.

Why this prompt worked

It named the decision, a horizon was implied by “6 weeks,” and it surfaced the two real worries (cannibalization, support). That was enough for the framework to do its job - declare the failure in the past tense, generate causes beyond the two she named, and force each into a tripwire and a kill criterion. No structure or polish was required of Mira.

What happened next

Mira did not run the launch off this; she ran the launch better. Risk #3 changed the plan immediately: conversion tracking became a launch gate, so the board would never again read signups as success. She carried the register into the launch doc, gave each top risk an owner, and re-read the kill criteria the week of launch, when momentum is highest and judgment is worst. Next in her thread: she could not choose between two pricing models for the new tier, so she ran the stress-test-a-decision recipe.