Definitions
How long to run an ad before deciding it's not working
Kill an ad after it has spent 2-3x your target cost-per-conversion with zero conversions, or after 2 weeks past the platform's learning phase — whichever comes first.
- 3 August 2026
- Free to read
- No signup
Add your website — see your ranked plan in about a minute, free.
How long should you run an ad before deciding it's not working?
Give a new ad through the platform's learning phase, roughly 7-14 days or 50 conversions (whichever is slower), before judging it. After that, kill it if it has spent 2-3 times your target cost-per-conversion with no conversions, or if cost-per-click is more than double your account average. Don't kill on day one or two — that's still the algorithm warming up, not a verdict on the ad.
Last updated 3 August 2026
| Signal | Keep running | Kill it |
|---|---|---|
| Days live | Under 7-14 (learning phase) | Past 14 with no movement |
| Spend vs target CPA | Under 2x target, no verdict yet | Over 2-3x target, zero conversions |
| Click-through rate | Within 50% of account average | Under half the account average |
| Cost-per-click | Near account average | Over double account average |
| Conversion rate on landing page | 1%+ (varies by industry) | Under 0.5% after 200+ clicks |
| Frequency (Meta) | Under 3 | Over 4-5, CTR falling |
Do this
The steps, in order.
- Step 1
Set the kill rule before you launch, not during
Write the exact number — 'kill if cost-per-lead exceeds $150 after $300 spent with zero leads' — before the ad goes live. Deciding in the moment almost always leans towards giving it more time than it deserves.
- Step 2
Separate 'not enough data' from 'genuinely bad'
Under about 500-1,000 impressions or 20-30 clicks, you don't have enough to judge — that's still noise. Don't kill an ad for a bad day one metric.
- Step 3
Check the landing page before blaming the ad
If click-through rate is fine but conversions are zero, the ad is doing its job and the landing page or offer is the problem — killing the ad won't fix that.
- Step 4
Kill the ad, not the whole campaign or channel
One underperforming ad or audience doesn't mean the channel failed. Keep 2-3 variants running at once so a bad one can be cut without losing the whole test.
- Step 5
Log what you killed and why
Keep a one-line record per killed ad: audience, hook, spend, result. After 10-15 kills you'll see a pattern in what doesn't work, which is more useful than any single test.
Worth knowing
The bits people get wrong.
Most platforms need a learning or delivery-optimisation phase before an ad's performance stabilises — Meta typically needs around 50 conversion events per ad set, Google Search needs a week or two of consistent delivery. Judging an ad before that point mostly measures the algorithm figuring out who to show it to, not the ad itself.
The most common mistake in the other direction is leaving a clearly dead ad running for a month because 'it might still pick up'. If it's spent three times your acceptable cost-per-result with no results, more time rarely rescues it — the audience or message is wrong, not the sample size.
It helps to distinguish three separate failures that look the same from the outside: nobody clicks (creative or targeting problem), people click but don't convert (landing page or offer problem), and people convert but don't become customers (lead quality problem). Each needs a different fix, and none is solved by simply waiting longer.
For businesses spending under $1,500/month, run at most 2-3 ad variants at a time rather than 6-8. A thin budget split too many ways means every individual ad stays stuck in 'not enough data' for weeks.
Questions
Follow-up questions.
Is it ever right to kill an ad on day one?
Only for obvious errors — broken link, wrong offer, rejected ad, or spend far above what you set. Kill for performance reasons, not errors, only after the learning phase.
How much should I spend before killing an ad with zero conversions?
A common rule of thumb is 2-3 times your target cost-per-conversion. If your target is $50 per lead, that's $100-150 spent with zero leads before you cut it.
Should I pause an ad or delete it?
Pause, don't delete. Pausing keeps the learning history and lets you compare it later; deleting loses the data and any accumulated social proof (likes, comments) on the post.
What if the ad is getting clicks but no sales?
That's a landing page or offer problem, not an ad problem. Fix or test the page before killing the ad — the ad is successfully doing the part it's responsible for.
How many ad variants should I test at once on a small budget?
Two to three per audience. More than that on a budget under $1,500/month spreads spend too thin for any variant to leave the learning phase within a reasonable time.
Elsewhere on gtm.help
Keep exploring.
Free marketing tools
Calculators and generators that run without a signup.
BrowseGrowth playbooks
Step-by-step plans by industry and goal.
BrowseTool guides
What to use, what it costs, when it earns its keep.
BrowseTool comparisons
Head-to-head picks for the calls people get stuck on.
BrowseMarketing checklists
Task lists by business type and competitive situation.
BrowseGet started
Want this answered for your business?
Paste your URL and you'll get a ranked plan, costed in hours and dollars, in about a minute.
Add your website and see your own ranked plan.