Short answer: Judge a creative testing ad in two different ways depending on whether it has sold. An ad with no sales is judged on spend: once it has spent a fair amount, a multiple of your target cost per sale, with nothing to show, pause it. An ad with sales is judged on cost per new customer and return on ad spend together, against your floor, because either number alone can mislead.
Most media buyers judge every ad on the same metric, usually return on ad spend. That mixes up two very different situations. An ad that has never sold and an ad that sells at a poor price need different answers, and the Ad Set they sit in needs its own answer too.
What does it mean to judge a creative testing ad?
Judging a creative testing ad means deciding, from its recent results, whether to pause it, hold it, move it to a cost capped campaign, or promote it to your Scaling campaign.
Those four outcomes are the only ones. Every ad in testing ends up in one of them sooner or later. This article covers how we pick between them. If you want the bigger picture of where these campaigns fit, read the CTSG system.
A few terms first:
- CPA (cost per acquisition): what you paid in ad spend for one sale.
- ROAS (return on ad spend): revenue divided by ad spend.
- New customer CPA and ROAS: the same numbers counting only first time buyers. In testing, new customers are what we care about most, because testing exists to find ads that bring in new people.
- Floor: the worst CPA and ROAS your business can accept. Below the floor, an ad is losing money in a way you cannot live with.
How do you judge an ad that has zero sales?
Judge it on spend. Until it has spent a fair amount, hold it. Once it passes that amount with zero purchases, pause it.
The problem is that "zero sales" means different things at different spend levels. After a few dollars, zero sales tells you nothing. After a meaningful amount, it tells you a lot.
The solution is to set a spend line tied to your own economics. A common rule in the industry, and the logic we use, is to tie it to a multiple of your target CPA. If a sale is supposed to cost a certain amount, an ad that spends a couple of times that amount without one sale has had a fair chance. The exact multiple depends on your price point, your margins and how quickly people buy.
Two details make this work better:
- Check more than one source before you call it zero. Meta might show no sales while your third party attribution shows a new customer. That is not a zero sale ad. Read both before you pause.
- Use early signals when you are close to the line. If an ad is near the spend line with no sales, look at cost per add to cart. Cheap add to carts mean people want it, and sales are often close behind. Cost per outbound click and outbound CTR (the share of people who clicked through to your site) are secondary signals. Healthy clicks with no add to carts usually means the ad attracts the wrong person. Expensive clicks with weak outbound CTR means the ad is not stopping anyone.
There is no "kill" category beyond this. A zero sale ad below the line is a hold. Above the line, it is a pause.
How do you judge an ad that has sales?
Once an ad has sold, spend no longer decides. Judge it on new customer CPA and new customer ROAS together, against your floor.
Here is why both numbers matter. Low priced products make CPA look good. High priced products make ROAS look good. An ad that sells your cheapest item can have a great CPA and a weak return. An ad that sells your most expensive bundle can have a great return on one order while its CPA is far too high. One number alone can fool you.
So the decision has four branches:
- Passes. ROAS vs floor: Passes. Call: Keep, then check for more budget or promotion. Why: It sells at a price that works
- Passes. ROAS vs floor: Fails. Call: Hold, recheck in a few days. Why: One number is misleading you; wait for more data
- Fails. ROAS vs floor: Passes. Call: Hold, recheck in a few days. Why: Same, in the other direction
- Fails. ROAS vs floor: Fails. Call: Move to the Graveyard. Why: It sells, just not at a good cost
That last row is important. An ad that sells, but too expensively, is not a loser. People respond to it. We move those ads to a cost capped campaign where Meta only spends when it can hit an acceptable price. That is the Graveyard. Only ads with sales go there. An ad with zero sales has nothing for a cost cap to rescue.
What does "heading the right way" mean?
It means an ad that is below floor over the window you judge on, but has sold in the last couple of days or whose last few days are clearly better than the days before. That ad gets one more check before you act.
The problem: a strict window can catch an ad right as it turns. If you pause it on the window average, you kill it just as it starts working.
The solution: give it one hold, once. At the next review, it is judged normally. If it is still below floor, it gets the normal call. This rule exists to catch real turns, not to give endless second chances. That is why it only applies once.
When does an ad earn a promotion to Scaling?
An ad earns a promotion when it has more than one sale, at or better than goal on both new customer CPA and ROAS, and those numbers hold on your deciding source.
A few guardrails:
- One sale is not enough. One sale can be luck. More than one, at goal, is a pattern.
- Watch for one large order. If an ad's return rests on one big order, judge it without that order. A pass that depends on a single big order is not a pass.
- Check frequency, but do not let it block the promotion. Daily frequency (how many times each person saw the ad that day) tells you who the ad is selling to. An ad selling while frequency stays near one is reaching new people, and tends to get spend in Scaling. An ad that only sells at higher frequency sells like a remarketing ad, and often gets little spend once promoted. We flag this, but the bar for promotion stays the same.
- Check that it is not already in Scaling. Look in every Scaling campaign and the Graveyard by the exact ad name first.
How is the Ad Set judged, separately from the ads?
The Ad Set is a budget decision. The ad is a creative decision. Judge them separately, then reconcile.
This is where many accounts go wrong. People look at the Ad Set's total numbers, see a poor return, and pause the whole thing. But an Ad Set can be failing as a whole while one ad inside it is working well.
How we reconcile the two:
- Ad Set failing, every ad failing: pause the Ad Set.
- Ad Set failing, one ad passing: do not pause the Ad Set. Pause the failing ads and let the winner have the budget.
- Ad Set passing, one ad failing: keep the Ad Set, pause that ad.
- Ad Set paused, but passing on your attribution tool: consider reopening it, minus the failing ads.
Never pause an Ad Set on its average alone. Always look inside first.
When does the Ad Set get its verdict?
At a set spend checkpoint for the whole Ad Set, not for a single ad. When the Ad Set reaches that checkpoint, every ad inside is reviewed. The checkpoint is when you look, not a trigger to pause.
We wrote a full article on this, the first verdict, because it is one of the most misunderstood parts of testing. Two points matter here:
- Before the checkpoint, only individual zero sale ads that pass their own spend line get paused. Every other ad waits.
- At the checkpoint, how the money was split matters. If one ad took most of the spend, judge the Ad Set mostly on that ad. If the money was spread thin across many ads, most of them never got a real test, so you do not judge the Ad Set as a whole yet. You may extend it a little, with a hard stop.
What happens when the top ad stops selling?
When the ad carrying an Ad Set stops selling and no other ad inside is selling at goal, the Ad Set should go.
We call the leading ad the main ad, picked on recent results, not lifetime. A lifetime leader can be fading while a newer ad carries the Ad Set. The main ad is "gone" when it has been paused, when its recent numbers fall below floor, or when it spends a fair amount recently with no sales.
If another ad inside is selling well, that ad carries the Ad Set and it stays on. If nothing is carrying it, pause the Ad Set. And importantly: pausing the top ad does not restart the test. The ads left behind do not get a fresh budget runway. They are judged on the Ad Set's real age and spend.
What about ads that slowly waste a little money?
Separate urgent pauses from cleanup. They are two different jobs.
Urgent pauses are ads wasting real money right now. Cleanup is for ads that have been running for a while, have spent a modest amount with no sales, and that Meta keeps feeding a little each day. They are not an emergency, but they add up. Handle them as housekeeping, on their own list, so they do not crowd out the urgent calls. More in pause vs cleanup.
Key takeaways
- An ad with zero sales is judged on spend; an ad with sales is judged on new customer CPA and ROAS together.
- Tie your zero sale spend line to a multiple of your own target CPA.
- When one number passes and the other fails, hold and recheck, because price point can make either number misleading.
- Ads that sell, but too expensively, go to a cost capped Graveyard, not the trash.
- The Ad Set is a budget decision and the ad is a creative decision; never pause an Ad Set on its average alone.
- More than one sale at goal, without leaning on one big order, is the bar for promotion.
FAQ
Should I judge testing ads on ROAS alone?No. ROAS alone favors ads that sell expensive products, even when they bring in too few customers. Use new customer CPA and ROAS together, and hold when they disagree.
How much should an ad spend before I pause it with zero sales?Tie it to your own target CPA, usually a small multiple of it. A low priced product with fast decisions can be judged sooner than a high priced one with a longer buying cycle.
Why move ads to a Graveyard instead of pausing them?Because an ad that sells is proof people respond to it. A cost capped campaign lets it keep selling only when the price is acceptable. Ads with zero sales get paused, since there is nothing to rescue.
What if Meta shows zero sales but my attribution tool shows sales?Do not pause on Meta's numbers alone. Meta can under report. If your other source shows new customers at a good price, the ad may be a real winner Meta cannot see.
Is one sale enough to promote an ad?No. One sale can be chance. We want more than one sale at goal on both numbers before promoting.
If you want to know how your testing campaign is being judged today, and what it might be missing, we offer a free Meta account audit. It includes a full account audit plus net new reach and cumulative reach reports. Request one at our contact page.
Keep reading
More on the CTSG system:
What Is the CTSG System? Creative Testing, Scaling and Graveyard for Meta Ads
How to Run a Daily Meta Ads Optimization (Without Overreacting)
How Early Can You Judge a New Meta Ad? The First Verdict Explained
Why Do Some Meta Ads Start Selling Late? Late Bloomers Explained
How to Tell if a Meta Ad Set Got a Real Test (Spread vs Concentrated Spend)
See every article on the CTSG system
Ready for your free Meta audit?
We will review your account and run net new reach and cumulative reach reports, free.
Check what you learned
Tap an answer to see if you got it right.
1. How do you judge an ad with zero sales?
2. CPA passes but ROAS fails. What do you do?
3. An ad sells, but too expensively. Where does it go?
Ready for the full test? Take the Media Buying exam and get certified. Want this run on your account? Get my free audit.
Module 1: Media Buying · article 4 of 19
Next: How Early Can You Judge a New Meta Ad? The First Verdict Explained
Previous: Should Meta Creative Testing Run to Cold Audiences Only?
About the author
Jake Mehani, founder of Creative Analytics
Jake has spent 15 years running Meta and paid social ads for consumer brands. Meta published his RBX Active Advantage+ case study, he taught the DMAnc Meta and TikTok ads masterclass, and he spoke at Motion's 2024 Creative Strategy Summit. His senior team runs media buying, creative and measurement like an in house team.