The single most expensive sentence in paid acquisition is 'let's give it another week.' It is expensive because every dollar that goes into a dying ad set is a dollar that did not go into the winner you have not yet found. The teams that compound on paid spend are not the ones with the best creative. They are the ones with the strictest kill rules.
The thresholds we use
Our rules are deliberately rigid. Rigid is the point. The moment you let yourself negotiate with the data, you stop running a system and start running a vibe.
- Spend cap before kill: 2.5× target CPA. If a creative has burned through 2.5× the cost we are willing to pay for a customer and still has not produced one, it is dead.
- Time cap before kill: 72 hours. After three full days, any creative that has not crossed our hook-rate floor is killed regardless of spend.
- Hook rate floor: 25% three-second view-through on cold traffic. If you cannot hold a third of the audience past three seconds, the rest of the creative does not matter.
- CTR floor at tier two: 1.4%. We let a creative through if it survived tier one (cold testing). At tier two budgets, CTR matters more than hook rate, because we have already filtered for hooks.
The opportunity-cost math
Here is why patience is expensive. A creative running at 1.2× your target CPA is not breaking even. It is bleeding margin. Every day you let it run, you are denying budget to whatever creative would have been tested in its place. If your testing cadence is two new creatives a week and you let a marginal creative run for an extra two weeks, you have skipped four tests. The probability that one of those four would have been a winner is high enough that the math almost never favors keeping the marginal creative alive.
Patience in paid acquisition is not a virtue. It is a budget leak with a story attached.
The exception we allow
There is one case where we let a creative survive past the kill threshold: it has an unusually low CPM and the conversion rate is the failure point, not the engagement. That signals an audience-or-landing-page problem, not a creative problem. We pull the creative off the original audience and test it against three new ones. If two of the three new audiences convert, we have learned something about audience targeting that the kill would have wasted. That is the only exception. There is no other.
What this looks like as a habit
The teams that win at paid acquisition over years, not months, all have the same habit: every Monday morning, every ad set gets reviewed against the kill rules. No exceptions, no 'let me think about it,' no founder favorites. The rules execute, the budget reallocates, and the testing engine keeps producing winners because the dying creatives are not crowding it out. That is the whole game.