The LoopDecide how you'll quit before you start
Most operators know when to start something and almost never decide, in advance, when to stop it. A kill rule — a real metric, a window, and an automatic consequence, written down before launch — is how a small budget keeps flowing toward what's working instead of what you're only hoping for.
Decide how you'll quit before you start
Most small operators know when to start something. Almost none decide, in advance, when to stop it. So the campaign that isn't working keeps running, because pulling it now would mean admitting the money already spent was wasted — and the money already spent is exactly the thing you can't get back no matter what you do next. The fix isn't willpower in the moment. It's a rule you wrote down before the moment arrived: decide how you'll quit before you start.
The sunk cost is the trap, and it's invisible from the inside
Once you've put real effort or real dollars behind something, your judgment about it quietly stops being about the future. It becomes about justifying the past. A week of zero results reads as "give it more time" instead of "this isn't landing," because more time is the only thing that could make the week you already spent feel like it meant something. That's the sunk-cost reflex, and you cannot reason your way out of it in the heat of it — the whole point of the trap is that it feels like patience while it's happening.
You beat it the only way it can be beaten: by making the decision earlier, when you had nothing invested and could think clearly. A kill rule set in advance is a message from the version of you who wasn't emotionally committed yet — and that version is the one you should trust.
What a kill rule actually looks like
It's a concrete, boring, pre-committed threshold that says this route ends here. Ours is a starve rule: if something is spending and produces two straight weeks with zero sales, its budget goes to zero. Not "we'll reassess." Not "let's tweak the copy one more time." To zero. The number was chosen before any campaign launched, precisely so that no single disappointing campaign gets to argue its own case later.
The rule needs three things to work. A metric that isn't a vibe — sales, not impressions or "engagement." A window — long enough to be fair, short enough to matter. And a consequence that's automatic, so triggering it isn't a fresh painful decision every time; it's just the rule doing what you already told it to do.
Killing one thing is how you fund the next thing
A kill rule sounds like it's about loss. It's really about redeployment. Every dollar and every hour still trapped in a route that isn't working is a dollar and an hour not available for the route that might. The operator who cuts the dead campaign fast isn't more ruthless than the one who lets it limp — they're more funded, because they freed the resources to concentrate somewhere with a pulse.
This is the same discipline from the other direction as earning the budget in the first place. You make a product prove organic demand before it gets spend; and you make spend keep proving itself, on a clock, or it loses the budget back. Both gates exist so a small pool of money always flows toward whatever is actually working and away from whatever is only hoping.
The whole rule in one line
Before you start, write down the number that ends it — a real metric, a real window, an automatic consequence. Then when the disappointing week comes, and it will, you're not making a hard call under pressure. You're keeping a promise you made yourself when you could still see straight.
Ready to try PromptGarden?
Grow, organize and share your prompt library
Browse Prompt Garden →Keep reading
Ship the smallest complete thing
When you have a stack of ideas, the pull is to build wide — a little of each, all at once. The discipline that actually ships is the opposite: make one thing whole before you touch the second. A small product a stranger can find, buy, and use beats an ambitious one that's mostly built and breaks near the end. Complete is defined by the buyer's full path, not your feature list.
The sale isn't done until the download works
The payment goes through, the success screen loads, and it feels like the work is done. It isn't. The buyer paid for a thing, and until that thing is in their hands and working, you haven't sold anything — you've taken money and promised to deliver. Fulfillment is the half of the sale that hides on the far side of the checkout, and it's the half easiest to leave half-built.
Show the work, not the star rating
A new product has no reviews, no follower count, no "trusted by" logos — and the temptation is to invent them. Don't. There is a kind of proof you can show on day one that fabricated proof can never match: the work itself. Show how it was made, let the thing be tried, and let provenance do what a borrowed star rating can't.