A shiny idea shows up and research starts. Tabs multiply. Evidence for the idea pops up everywhere because you are hunting for it. Kill criteria written first change the hunt. Write kill criteria before you research a shiny idea so the project can die cheaply when it should.
Kill criteria are not vibes. They are observable stops: if X is true, we drop this for thirty days. Without them, research turns into advocacy with citations and a growing pile of almost-proof.
List three falsifiers before the first tab
Write three conditions that would kill the idea. Price floor above budget. No owner available this quarter. A dependency that fails a five-minute ping. Keep them measurable. Soft feelings do not count as criteria and should not enter the list.
Arguing the opposite for a few minutes sharpens weak criteria. Use the timed opposite brief from argue the opposite case hard for five timed minutes before you lock the list.
Timebox research to one cheap test each
Each criterion gets one check, not an essay. Fifteen minutes per check is enough for most early ideas. If a check is unclear, redesign the check once. Do not open a third theory draft. That pattern matches splitting stuck work into one next cheap test.
Curiosity tabs love shiny ideas. Mute them with a door so research stays on the falsifiers, as when you mute curiosity tabs before they eat your next hour.
Record pass, fail, or unclear in public
Write the outcome beside each criterion where teammates can see it. Private advocacy thrives in private notes. Public fail marks make it easier to stop. Unclear means the test was weak, not that the idea deserves more romance or another week of quiet browsing.
When decisions spin, put a clock on naming the next criterion using timebox looping decisions so choices stop spinning.
Protect the stop rule from sunk-cost talk
After two fails, stop for the agreed window. Do not renegotiate criteria mid-fail because the idea feels clever. Changing criteria after evidence arrives is how shiny ideas become permanent. Schedule a revisit date if you must, then close the folder.
Park unfinished sentences so the fail record survives the next meeting switch: park unfinished sentences before switches.
Reuse criteria templates for recurring idea types
Tooling ideas, hiring ideas, and content ideas fail for different reasons. Keep short templates. Reuse beats inventing philosophy under excitement. Review monthly which ideas died cleanly versus which limped without criteria and still sit in someone’s sidebar.
Clean deaths free attention for work that still has a live path. Speed of killing bad ideas is a focus skill as much as a strategy skill, and it compounds when the team trusts the stop rule.
FAQ
What is a kill criterion?
An observable condition that means you stop the idea for a set window without further research.
How many criteria should I write?
Three is enough for early ideas. More becomes a stall tactic.
When do I write them?
Before the first research tab or outreach message.
What if a test is unclear?
Redesign the check once. Do not treat unclear as a pass.
Can I change criteria later?
Only on a scheduled revisit, not in the same session as a fail.
Does this slow good ideas?
It slows advocacy. Good ideas survive cheap falsifiers and move faster afterward.
