You’re staring at an underperforming Facebook ad campaign. It’s been running for five days. CPA is double your target. ROAS is 1.5x when you need 4x. Your instinct says kill it. But your account rep says “give it time, the algorithm is still learning.”
Meanwhile, the ad is spending $150/day and delivering nothing.
This is the hardest ongoing decision in paid advertising: when to be patient and when to cut your losses. Wait too long and you burn budget. Kill too early and you never find out if the campaign would have worked. Most advertisers have no framework for this decision — they go on gut feeling, which means they’re inconsistent and often wrong.
Here’s a data-driven framework for deciding when to pause, when to optimize, and when to kill a Facebook ad.
The Three Decisions
Every underperforming ad or campaign leads to one of three actions:
- Wait — The data is insufficient. Keep running without changes.
- Optimize — The signal is clear enough to act. Adjust targeting, creative, or bidding.
- Kill — The campaign is fundamentally not working. Stop spending.
The mistake most advertisers make is jumping straight to #3 without going through #1 and #2. The second mistake is staying on #1 forever because they’re afraid of making a decision.
The Statistical Minimum: When You Have Enough Data
Before making any decision, you need enough data to be statistically meaningful. A campaign with 200 impressions and zero conversions tells you nothing. A campaign with 10,000 impressions and zero conversions tells you something important.
The 2x Spend Rule
You haven’t spent enough to decide until you’ve spent at least 2x your target CPA on an ad or ad set.
If your target CPA is $40, you need to spend at least $80 on an ad before deciding it doesn’t work. With $80 spent and zero conversions, you can be reasonably confident the ad isn’t converting at your target rate.
If your target CPA is $40 and you’ve spent $25 with zero conversions — that’s not a failing ad. That’s insufficient data.
The Click Threshold
Beyond spend, look at clicks. You need at least 100 link clicks to an ad set before judging landing page conversion rate. With 50 clicks and zero conversions, the sample is too small. With 200 clicks and zero conversions, the landing page is the problem.
The Impression Threshold
For creative performance (CTR), you need at least 1,000 impressions before judging. A 0% CTR on 200 impressions is noise. A 0.3% CTR on 5,000 impressions is a genuine signal.
The Decision Framework
Once you have sufficient data (2x CPA spend, 100+ clicks, or 1,000+ impressions), apply this framework:
Level 1: Is the Creative Working?
Metrics to check: CTR (link click-through rate), Hook Rate (3-second video views / impressions), CPC
| Metric | Healthy | Warning | Kill Zone |
|---|---|---|---|
| CTR (link) | >1.5% | 0.8-1.5% | <0.8% |
| Hook Rate (video) | >25% | 15-25% | <15% |
| CPC | Below target | At target | 2x+ target |
CTR below 0.8% after 1,000+ impressions:
- The creative isn’t resonating. People see it and scroll past.
- Action: Kill the specific ad creative. Don’t kill the entire ad set or campaign — replace the creative.
CTR between 0.8-1.5%:
- Mediocre but not dead. Test new creative alongside it.
- Action: Optimize. Add new ad variations to the ad set.
CTR above 1.5%:
- Creative is working. If conversions are still low, the problem is downstream (landing page, offer, audience).
- Action: Move to Level 2.
Level 2: Is the Audience Responding?
Metrics to check: CPC, CPM, Frequency, Audience size
| Metric | Healthy | Warning | Kill Zone |
|---|---|---|---|
| CPM | $5-15 (varies by industry) | $15-25 | $30+ |
| Frequency | <2 (prospecting) | 2-3 | 4+ (prospecting) |
| CPC | <$2 (ecommerce) | $2-5 | $5+ |
CPM above $30:
- You’re paying a premium to reach this audience. Either the audience is too small, too competitive, or Meta’s algorithm is struggling to find convertible users.
- Action: Broaden the audience. If using a narrow interest or small lookalike, test broader targeting.
Frequency above 3 (for prospecting):
- The same people are seeing your ad multiple times without converting. You’ve saturated the audience.
- Action: Expand audience size, add new creative (fresh creative resets effective frequency), or pause and let the audience cool off for 1-2 weeks.
CPC above $5 for ecommerce (adjust for your industry):
- Each click is expensive. Even if the landing page converts well, CPA will be high.
- Action: Test new creative angles (creative drives CTR which drives CPC). Check if audience overlap is inflating costs.
Level 3: Is the Landing Page Converting?
Metrics to check: Conversion Rate (CVR), Cost Per Add-to-Cart, CPA
| Metric | Healthy | Warning | Kill Zone |
|---|---|---|---|
| CVR (ecommerce) | >2% | 1-2% | <1% |
| CVR (lead gen) | >5% | 2-5% | <2% |
| CPA vs target | At or below | 1-1.5x target | 2x+ target |
Conversion rate below 1% (ecommerce) after 200+ clicks:
- Clicks are coming. People are landing. They’re not buying.
- Action: Fix the landing page, not the ad. Check message match, page speed, mobile experience, offer clarity. The ad is doing its job (getting clicks). The page isn’t doing its job (converting).
CPA at 1-1.5x target after 3x target spend:
- Close but not there. Worth optimizing.
- Action: Adjust bid strategy. If using lowest cost, switch to cost cap at your target CPA. Refine audience. Test landing page variations.
CPA at 2x+ target after 3x target spend:
- Not close enough to optimize into profitability.
- Action: Kill the ad set. Something fundamental isn’t working — wrong audience, wrong offer, or wrong creative-to-landing-page match.
The Learning Phase: When to Be Patient
Meta’s algorithm enters a “learning phase” when you create a new ad set or make significant changes. During this phase (approximately 50 conversions), performance is volatile. CPA may spike, then drop, then spike again.
Don’t Kill During Learning Phase
If your ad set hasn’t exited the learning phase, CPA data is unreliable. The algorithm is experimenting — showing your ad to different people to learn who converts.
Patience threshold: Wait for 50 conversions OR 7 days (whichever comes first). If after 7 days you haven’t reached 50 conversions and CPA is 3x+ target, the campaign structure may need adjustment (budget too low, audience too narrow, or event optimization too far down funnel).
Don’t Reset the Learning Phase
These actions reset the learning phase:
- Changing targeting
- Changing budget by more than 20% in a single day
- Changing bid strategy
- Pausing for more than 7 days and restarting
- Changing creative (adding/removing ads)
Each reset means another 50 conversions of unstable performance. Make changes infrequently and deliberately.
The Kill Checklist
Before killing a campaign, run through this checklist:
Is the Tracking Working?
A campaign that appears to have zero conversions might actually be converting — but the pixel isn’t firing correctly. Before killing anything, verify:
- Meta Pixel is installed and active (check Events Manager)
- Conversion events are firing on the correct page
- Conversions API is sending server-side events
- No duplicate or missing events
- Attribution window in Ads Manager matches your evaluation window
We’ve seen advertisers kill profitable campaigns because a tracking update broke their pixel and conversions stopped being reported. Check tracking first. See our Meta Pixel vs Conversions API guide for verification steps.
Have You Tested Enough Creative?
The #1 lever on Meta is creative. Before concluding “this audience doesn’t convert,” ask whether you’ve tested 5+ meaningfully different creative angles. An audience that doesn’t respond to a product photo carousel might respond to a UGC testimonial video.
Is the Offer Competitive?
Sometimes the ad and the audience are fine. The offer just isn’t compelling enough. Compare your offer to competitors:
- Are others offering free shipping? A discount? A guarantee?
- Is your price competitive?
- Is your landing page as clean and trustworthy as competitors’?
Are External Factors at Play?
Check for external factors:
- Seasonality (CPMs spike during Q4, back-to-school, etc.)
- Competitor activity (new competitor entering market)
- Market events (economic changes affecting buying behavior)
- Platform changes (Meta algorithm updates, new ad formats)
A campaign that underperforms during Black Friday CPM spikes might perform great in January.
When to Kill: The Hard Rules
Kill with confidence when ALL of these are true:
- You’ve spent at least 3x your target CPA
- The campaign has exited learning phase (or been in learning for 7+ days)
- CPA is 2x+ your target with no downward trend
- You’ve tested 3+ creative variations
- You’ve verified tracking is working correctly
- Landing page conversion rate has been checked independently
If all six conditions are met, kill it. Don’t second-guess. Redirect that budget to what’s working.
When to Optimize Instead of Kill
Optimize when you see these patterns:
Good CTR, bad conversion rate:
- Creative works, landing page doesn’t. Fix the page, not the ads.
Good conversion rate, bad CPA (high CPC):
- Landing page works, but clicks are expensive. Test new creative to improve CTR/CPC.
CPA close to target (within 25%) and trending down:
- The algorithm is still learning and improving. Give it more time and data.
Strong performance in one segment, weak in another:
- Check breakdown by age, gender, device, placement. Exclude the underperforming segment instead of killing the entire ad set.
After You Kill: What to Do Next
Killing a campaign isn’t failure. It’s data. Use what you learned:
- Document what didn’t work — Creative angle, audience, offer, landing page. Build a “negative learnings” list.
- Analyze the data — What segment of the audience DID engage? What creative had the best CTR even if conversions were low? Are there signals buried in the data?
- Test the inverse — If a broad audience didn’t work, try a specific one (and vice versa). If a product demo video didn’t work, try UGC. If a discount offer didn’t work, try a value proposition.
- Consider platform mix — Maybe this product or offer works better on Google Ads (intent-based) than Meta (interruption-based). Check ROAS by platform.
- Try the Advantage+ approach — If manual targeting didn’t work, test Advantage+ Shopping campaigns which let Meta’s algorithm handle targeting entirely.
The Monthly Campaign Audit
Don’t wait for campaigns to fail. Run a monthly audit:
- Check frequency across all active ad sets. Anything above 3 needs fresh creative or expanded audience.
- Compare CPA trends week over week. Rising CPA is the earliest warning sign.
- Review creative performance — Which ads are still performing? Which have fatigued?
- Verify conversion tracking is still working. A tracking break can masquerade as a campaign performance decline.
Trust Your Data, Not Your Feelings
The framework above works because it removes emotion from the decision. “I feel like this ad should work” isn’t a strategy. “CTR is 2.1%, CVR is 0.4% after 300 clicks, and CPA is 3.2x target — kill it” is a strategy.
Document your thresholds. Apply them consistently. Review and adjust the thresholds quarterly based on your actual performance data.
Start With Reliable Tracking
Every number in this framework — CPA, ROAS, CVR, conversion count — depends on accurate tracking. If your Meta Pixel is missing conversions, you might kill a profitable campaign. If it’s double-counting, you might keep a money-losing campaign alive.
Run a free scan on your site to verify your conversion tracking is accurate before making any pause/kill/optimize decisions. The worst outcome isn’t a failed campaign — it’s killing a winner because your data was wrong.