How to Write an Ad Kill Rule Before You Spend a Euro
How to write an ad kill rule: three numbers to decide before launch, a seven-line template to copy, and what to do on the last day of the test, however you feel
To write an ad kill rule, decide three things before you launch: the budget, the duration, and the result that ends the test. Write them in one sentence, put it where you'll see it at the end, and treat it as binding.
How to write an ad kill rule: one sentence, written while you're calm
Here is a kill rule.
"I will spend this much over this many days. If the cost per real lead is above my ceiling at the end, I stop."
Here is not a kill rule.
"Let's see how it goes."
That second sentence has funded more dead campaigns than any bad creative ever did. It sounds reasonable. It's actually a decision to decide later, which means a decision to keep spending.
Why you can't trust yourself at the moment of decision
The end of a test is the worst time to decide anything. Think about what's in the room.
There's a near-miss. The cost per lead is a bit over your ceiling, not far over. There's a promising trend: yesterday was better than the day before. There's the platform itself, politely suggesting you raise the budget because it is "learning". And there's you, who spent a week on that creative and quite likes it.
On a team, someone would say "this isn't working". You work alone. Nobody holds you to the number except the number, and only if it was written down.
A written ad kill rule is the only colleague you have. Treat it that way. It was written by the calm version of you, with nothing to defend. The tired version at the end of the test doesn't get a vote.
Free guide · 21 pagesJudge ads by deals, not clicks.The Paid Ads Playbook: how to spend on ads without lighting money on fire.Get the free guide →The three numbers every kill rule needs
An ad kill rule with fewer than three numbers is a wish.
| Number | What it answers | How to set it | What goes wrong without it |
|---|---|---|---|
| Budget | What is this finding worth to me? | The price of a finding, paid in full or not at all | Spend creeps up one reasonable day at a time |
| Duration | When does the test end? | A fixed period, chosen before launch | "Another week" becomes a habit |
| Stopping result | What outcome ends it? | Your ceiling for a real lead, from your own numbers | Any result can be called good enough |
The third is where the work is. You can't invent the stopping result in a hurry. It comes from arithmetic you do first.
Get the ceiling before you write the rule
Every ad decision comes back to one comparison: what a customer costs against what a customer is worth.
Take what a customer pays you over a period you can honestly count on. Multiply by your gross margin, so you're working with what you keep, not what you invoice. Decide how long you can wait to be paid back. That's the most you can spend to win one customer.
Now work backwards through your own conversion rates. Leads to conversations. Conversations to closed deals. That gives you the most you can pay for a lead, and from there for a click.
Write all three ceilings down. The one that goes in your kill rule is the cost per real lead. Real, meaning distinct people you recorded, not the count the platform reports. If you want to see why those two numbers differ, read how to judge ads by deals, not clicks.
A rule you can copy and fill in
Paste this into a note. Fill the blanks. Do it before you open the ad manager.
- This test is research / acquisition. (Pick one.)
- I will spend a total of ___ over ___ days, and no more.
- I am testing one offer, one audience and these ads: ___.
- The result I report to the platform is ___.
- At the end, I compare the cost per real lead with my ceiling of ___.
- If it is above the ceiling, I stop it that day. If it is at or below, I earn a slightly larger test, not a blank cheque.
- I will not edit the campaign before the end date unless something is genuinely broken.
Seven lines. Print it if you must. Put it where you'll see it on the last day, not buried in a folder.
What "genuinely broken" means
You'll want an escape hatch. Fine. Define it narrowly.
Broken means conversions aren't arriving at all, a link is dead, or the page is down. It does not mean "today was bad", "a competitor launched" or "I had a better idea for the headline".
Why so strict? Because every edit to a live campaign (a new ad, a budget change, a tweak to the audience) resets part of the platform's learning and contaminates your comparison. You asked a question. Edit mid-test and you can't tell what the answer means.
So verify at the start that conversions arrive. Then leave it alone. The setup for that is in how to test ads on a small budget.
Free assessmentHow trapped are you, really?Take the Corporate Suffocation Index, the free solopreneur readiness assessment.Take the assessment →What to do when the test ends
Compare. Don't negotiate.
- Passed? You have earned a slightly larger test. Increase in a step, hold it long enough to read, and apply the same rule at every level.
- Failed? Stop it today. Not after one more week.
- In between? That's the case the rule exists for. Apply it as written.
Then write one paragraph: what you tried, what it cost, what it produced, and the decision. That record is the most valuable advertising asset you can build, because it's the only one the platform can't see.
A test stopped on time is not a failure. It's a finding you bought at the price you agreed to pay.
Five ways a kill rule quietly dies
You wrote it. You meant it. Here's how it still gets broken.
- The soft deadline. The end date arrives and you extend by a few days "to get a cleaner read". A cleaner read of what? The rule said stop.
- The moved ceiling. The cost per lead lands above your number, so you decide the number was too strict. If the ceiling was wrong, fix it for the next test, never mid-flight.
- The platform's nudge. It suggests raising the budget. It will never suggest stopping. It has no reason to.
- The proxy slide. Leads disappoint, so you start admiring clicks, views or reach. Those were never the outcome you chose.
- The hidden rule. It lives in your head or in a folder you never open. If you can't see it on the last day, it doesn't exist.
Each one feels like flexibility. Each one is the same old habit in a better coat: letting a research budget become a permanent cost.
Why an ad kill rule beats your judgement
You might think experience will do the job. Surely after a few campaigns you'll just feel when to stop?
You won't. The feeling is the problem. Spending quietly leaves your account while you're busy with everything else, and the dashboard reports generously on itself. A feeling can't compete with that. A number you wrote down earlier can.
That's also why this is a solo skill more than a team one. A team has someone to say "that's enough". You have a sentence on a page. Make it a good one.
Free guide · 21 pagesJudge ads by deals, not clicks.The Paid Ads Playbook: how to spend on ads without lighting money on fire.Get the free guide →The catch with leads versus deals
One honest warning. A rule built on cost per lead can still reward the wrong ads. A cheap lead that never becomes a conversation is not cheap.
That's where the lag comes in. A lead today may close weeks or months from now. If you judge on revenue too early, every ad looks like a failure. Wait forever and you never kill anything. So decide the wait in advance. Find your typical time from first contact to closed deal, and judge cohorts on revenue only once they've had that long. Until then, steer by the furthest-down milestone that reliably predicts revenue, and call it a proxy out loud.
Put that in the rule too. One more line, written now.
Free assessmentHow trapped are you, really?Take the Corporate Suffocation Index, the free solopreneur readiness assessment.Take the assessment →Frequently asked questions
What is an ad kill rule?
A threshold written before launch that tells you when to stop a campaign. It has a fixed budget, a fixed duration and a stopping result. "Let's see how it goes" is not one.
What number should I base the kill rule on?
Your own ceiling on cost per real lead, worked backwards from what a customer is worth after margin and from your own conversion rates. Use real people you recorded, not the platform's conversion count.
Can I change the rule if the test looks promising?
No. A promising trend is exactly what the rule is there to resist. If the result passes, you earn a slightly larger test with its own rule. You do not earn a bigger budget on a hunch.
Should I stop early if it is clearly failing?
Only if something is genuinely broken, such as conversions not arriving. Otherwise let it run to the end date, then stop that day. Early edits and early exits both make results hard to read. Stick to the ad kill rule you wrote.
The full method, with the checklists, is in The Paid Ads Playbook, a free 21-page playbook.
How trapped are you, really?
Take the Corporate Suffocation Index, the free solopreneur readiness assessment.