Every media buyer says they're "data-driven." Then you open their account and find ads that have been bleeding for three weeks because "they just need more time," sitting next to ads that got killed at $30 spend because the first-day CPA looked scary.
Kill rules exist so that the decision gets made by the system you built when you were calm, not by you at 11pm after a bad day. These are mine. They're not universal laws; they're the defaults I start every account with, then tune to the economics of that specific brand.
Rule 1: Spend to signal, not to feelings
An ad hasn't "failed" until it has spent enough to produce a readable signal. My default: don't judge an ad before it has spent roughly 1–1.5× your target CPA. If your target CPA is $40, an ad with $25 spend and zero conversions has told you almost nothing. Killing it there isn't discipline. It's noise-chasing.
The flip side: once an ad has spent 2–3× target CPA with nothing to show, it's done. No "one more day." The data already spoke; you just didn't like the accent.
Rule 2: Judge creative on leading indicators first
CPA is a lagging signal: it needs volume before it's trustworthy. Creative tells on itself much earlier:
- Hook rate (3-second views / impressions): is the first frame doing its job?
- CTR: did the promise land hard enough to earn a click?
- CPM drift: is Meta telling you the audience is rejecting it?
An ad with a strong hook rate and CTR but a bad early CPA often just needs time or a landing-page fix. An ad with a weak hook rate almost never recovers. The best offer in the world can't save an ad nobody stops for.
Pool rule 002Creative is the targeting.
Rule 3: Winners die of old age too
The mistake nobody talks about: riding a winner into the ground. When frequency creeps up and CTR starts sliding week over week, the ad is fatiguing, and the time to build its replacement was two weeks ago. This is why I deliver creative on a fixed monthly cadence instead of "when things break." Volume is insurance. You never want your account's performance depending on one tired ad.
Rule 4: Kill at the concept level, iterate at the variation level
If five variations of a concept all miss, the concept is dead. Stop making a sixth. But if one variation of a losing concept shows a pulse (decent hook rate, say), the angle might be right and the execution wrong. That's an iterate, not a kill. Knowing which one you're looking at is 80% of creative strategy.
Rule 5: The account has a memory. Respect it
Don't reset campaigns because you're bored. Constant restructuring throws away learning-phase data the algorithm paid for with your client's money. I touch structure rarely and creative constantly. Most "structure problems" are creative problems wearing a disguise.
The uncomfortable summary
Most buyers kill winners too early (panic on day one) and losers too late (hope on day twenty). The fix isn't more dashboards. It's deciding your thresholds before you launch, writing them down, and letting the rules do the killing so you don't have to.
Screenshots aren't results. P&L or it didn't happen.