Skip to content
Back to Idea Validation

Cut Criteria

Define in advance what results would make you kill the idea — pre-commitment to prevent sunk-cost escalation.

Why

The most expensive failure mode is not killing a bad idea — it is keeping it alive past the point where the evidence says stop, because of sunk cost. Cut criteria are pre-committed thresholds: "if the MVP does not reach 10 paying users in 4 weeks, I kill it." Written before launch, before emotional attachment to the work, they prevent the rationalization that sets in after: "well, we just need better marketing" / "one more feature and it will take off". Without pre-committed cut criteria, teams spiral — adding features, running ads, changing messaging — for months past the point where an honest assessment would have killed or pivoted the product.

When

Before MVP launch, in the same session as defining the hypothesis. Takes 1 hour. Review at each milestone.

How

1. Define the success metric
   - The single number that proves the hypothesis
   - Example: "trial → paid conversion rate" or "weekly active users who completed the core loop"

2. Set the threshold for success AND failure
   - Success: "conversion > 5% in 4 weeks → proceed to v1"
   - Failure: "conversion < 1% in 4 weeks → kill or pivot"
   - Gray zone: "1-5% → iterate on the biggest friction point, re-test"

3. Set the time window
   - Long enough to get statistical signal (usually 2-4 weeks)
   - Short enough that you do not waste months (max 6 weeks for an MVP)

4. Define what "kill" means
   - Kill the idea entirely? Or pivot to a different segment/problem?
   - What happens to existing users? (refund, migrate, sunset)
   - What do you do next? (new idea, or rest from this one)

5. Define what "pivot" means
   - Which dimension changes? (segment, problem, solution!Which dimension changes? (segment, problem, solution, channel)
   - What carries over? (tech, insights, users)
   - How many pivots before full kill? (recommend max 2)

6. Write it down and share it
   - Physical document, shared with a trusted advisor
   - They hold you accountable when the gray!They hold you accountable when the gray zone tempts you to keep going

Example

MVP: AI changelog generator.
Hypothesis: "solo devs will pay $15/mo to auto-generate changelogs."

Cut criteria (written before launch):
  Metric: trial → paid conversion rate
  Time window: 4 weeks post-launch
  Success: conversion > 5% → build v1 with top 3 requested features
  Failure: conversion < 1% → kill the idea, refund all pre-orders
  Gray zone (1-5%): interview 10 trial users, fix the #1 friction, re-test for 2 more weeks

  Kill means: sunset the product, refund all users, write a post-mort!Kill means: sunset the product, refund all users, write a post-mortem, start a new idea
  Pivot means: if the problem is real but the solution is wrong, change the solution (max 1 pivot before kill)

  Accountability: shared with my indie hacker accountability partner; I must show them the data at week 4 and follow the criteria.

Week 4 result: 180 trials, 9 paid = 5.0% conversion → gray zone edge → interview 10, find "output quality inconsistent" → fix prompt → re-test 2 weeks → 7.2% → proceed to v1.

Common Problems

  • ⚠Setting the threshold after seeing the result — "well, 3% is pretty good" moves the goalposts; set thresholds before launch.
  • ⚠No definition of "kill" — without specifying what kill means (refund? sunset? pivot?), you avoid the decision by keeping options open.
  • ⚠Gray zone without an action — "1-5% is ambiguous" without a defined next step becomes an excuse to keep going indefinitely.
  • ⚠No accountability — criteria you keep to yourself are easy to rationalize away; share them with someone who will hold you to them.

Best Practices

  • ✓Write cut criteria before launch and share with an accountability partner!Write!Write cut criteria before launch and share with an accountability partner — the social commitment makes rationalization harder.
  • ✓Define the gray zone action — "if 1-5%, interview 10 users and fix the top friction" is actionable; "if 1-5%, think about it" is not.
  • ✓Set a maximum number of pivots before full kill — infinite pivoting is just slow killing; 1-2 pivots is honest.
  • ✓Review the criteria at each milestone with your accountability partner — they will see rationalization you cannot see in yourself.