Should we test new creative in a separate testing campaign or inside the live campaign?
Our answer
The answer depends on spend.
Test new creative as new ads in the live campaign. If new concepts as a group get almost no spend for a week, give them their own ad set there, without older winners.
At $200K or more a month, test new concepts in a separate ad-set-budget campaign, one concept per ad set, and scale winners there. If you move one, use its post ID and keep the original running.
Why this is the answer
3 reasons and our judgmentWinners keep what Meta's delivery has learned when you test in place. Meta's help page runs creative tests inside an existing campaign, so there's no need to merge winners into another campaign where learnings reset.
An ad that wins in isolation isn't guaranteed to perform once moved. Jon Loomer calls testing inside the live ad set far superior to separate test campaigns for that reason.
Moving winners to a new campaign can backfire. Biddyco, an agency, calls graduating winners to a scaling campaign close to the worst thing you can do to an account. Eric Zaworski of Wispr Flow says graduated winners sometimes die.
Our judgment
- Keep testing in the live campaign, even though sources split on keeping a separate testing campaign. Loomer, Biddyco and Zaworski's account, which tests separately, all report winners that fail once moved, and Meta builds its testing tool to run inside the existing campaign. The main counterexample, Phil Kiel's switch, is one account with no control.
- At $200K or more a month, the narrower answer for consumer brands keeps a separate testing campaign instead. Kiel (about $850K a month), Marin Istvanic ($7M a month), Flighted and Skaleit Agency back it. Biddyco, which says its advice matches what it sees in accounts spending $1M to $3M a month, is the dissent there.
- Use one week of almost no spend on new concepts as the trigger to give them their own ad set. Check it with Kiel's measure, the share of weekly spend on ads under 14 days old, alongside Loomer's week-long push delivery check.
- Use Meta's creative testing tool when new ads get no spend. This extends Loomer's description of the tool spending budget evenly across test ads and Meta's note that test ads get their own delivery.
Our judgment, built from Jon Loomer, Biddyco, Eric Zaworski, Meta, Phil Kiel, Marin Istvanic, Flighted and Skaleit Agency: our reading of these sources, not a finding from any one of them.
What the sources say
34 statements from 23 sourcesPhil Kiel says not to pause a winning test ad when moving it to another campaign, since there is no guarantee it performs the same in a new ad set. No data.
“Never pause the ad in testing if moving an ad to another campaign.”
Charley Tichenor runs creative tests in their own ad sets inside one campaign, so turning off a losing test never touches the control. No results given.
“Turning off a losing test is a non-destructive edit. The control never gets touched.”
Phil Kiel, who runs an agency, tests in ABO at high creative volume: each new ad set starts at about 30% of target CPA a day, with no cost caps. He still uses CBO at 10 to 30 new ads a month.
“Every new ad set starts small, around 30% of target CPA a day.”
Meta says Advantage+ campaign budget suits campaigns with at least 2 ad sets and may spend most of the budget on one ad set if that gets the best overall results.
“Advantage+ campaign budget is best suited for campaigns with at least 2 ad sets.”
Phil Graham moves ads that get little or no spend into a new campaign with their own budget, to test them himself.
“I take out the ads. That are not getting spent. I turn them off in that campaign. I'll create a new campaign, new ad set and I'll give those ads or add its own budget and I'll test it myself.”
Meta's Ads Creative Studio page says the tool, not yet open to everyone, flags weak ad sets and ranks top single image or video ads by purchase ROAS over 7 days. It describes no creative testing feature.
“Ads Creative Studio is only available for single image and video ads at this time.”
Meta's Nik Ketkar says Ads Creative Studio will add creative testing and that the learning a creative gathers will carry into later campaigns. He gave no date, and Meta's help pages did not yet say so.
“all of those signals that you've acquired in one particular creative will translate to your ... subsequent campaigns”
Nick Theriot says successful Meta ad accounts do not need separate testing and scaling campaigns; he runs a single CBO campaign.
“I don't have a testing campaign. I don't have a scaling campaign. I just have a CBO campaign.”
Meta's related media feature suggests 1 to 10 existing assets for single image or video ads and may be on by default. Meta says one experiment showed 12.9% more conversions, with no test design given.
“Related media may be enabled by default, but you can disable it or edit the media recommendations”
Meta's creative test tool compares up to 7 copies of an existing ad inside the same campaign, so winners keep their learning. It suggests no more than 20% of budget and gives no confidence level.
“The test is set up in an existing campaign so that high-performing ads can continue to run after the test with delivery system learnings retained.”
Nick Theriot says he took over a flat brand in July 2025 and did $5M in 12 months without changing structure: one campaign for USA and Canada, three new creatives per ad set weekly, and he turned off non-buying ads.
“Three creatives per ad set, every single week, without exception”
A buyer managing $7M a month tests creative in ABO ad sets, one concept per ad set with 3 to 5 variations, starting at $100 a day on broad audiences.
“One concept per ad set. 3-5 variations.”
In one account spending about $850k a month, moving creative testing into ABO, one ad set per concept, raised weekly spend on creative under 14 days old from 10% to 38% over eleven weeks.
“CBO makes a new ad take budget off an old one. ABO lets it earn its own.”
Eric Zaworski says a testing winner sometimes dies when it moves to a scaling campaign with an order of magnitude more daily budget, so scale in steps you can afford and watch each one.
A vendor blog, with no data, suggests giving each creator its own ad set with a spend floor in one testing campaign, running seven days without edits, then moving winners to a CBO campaign.
“you lose per-creator budget control, so keep the test in a manual campaign first.”
Jon Loomer says that for low-volume accounts, isolating creative tests in separate campaigns or ad sets waters down the budget and gives meaningless results.
“if you're getting low volume results, you're watering down your budget even more with these tests.”
Biddyco, an agency, runs one main campaign: under about $50,000 a month, test new creative in its own ad set; at high spend, add it to the winners' ad set. It warns against graduating winners.
“close to the worst thing you can do to an account”
Biddyco, an agency, says when most ads in an ad set get almost no spend, turn off those that have had time and add new concepts in batches. In its accounts 10 to 30% of ads take 70 to 80% of spend.
“Turn off ads that have had enough time and aren't getting spend, and add new concepts in batches.”
Skaleit says above $5K a day, run a separate test campaign plus a single-ad-set CBO scaling campaign; running 30 to 50 ABO ad sets at that spend becomes a full-time job.
Flighted, an agency, has brands spending $5K to $100K a month run a sandbox test campaign at about 25% of spend, one ad set per concept, beside a winners campaign.
“One ad set per creative concept, grouped by theme or format”
Ralph Burns says to skip separate sandbox or testing campaigns and put all creative into one campaign for the algorithm to test.
“gone are the days of having a sandbox campaign or a separate testing campaign.”
Across 504 Common Thread Collective stores and about $3.35B of Meta spend, only 74% of ads got any spend. Whales were about 1% of new ads but 41% of new-ad spend.
“only seventy-four percent of ads got any spend at all”
Pushes back on full consolidation: an agency says in 70+ audits of brands spending $75K to $1.8M a month, it never saw one consolidated campaign sustain serious spend, so it runs separate campaigns.
CTC runs evergreen acquisition and retention campaigns, time-bound campaigns for marketing moments, and a recovery campaign for ads that spent less than 3 times the CPA target.
“Populate it with ads from evergreen where spend is less than 3x the CPA target.”
Zach Stuck of Motion says he wants about $300 a day in spend behind each creative test under a concept.
“I want to see, you know, $300 in spend per day behind a creative test”
Jetfuel says at $1K or more a day to run Advantage+ sales on 60 to 70% of budget plus a broad testing campaign on 20 to 25%, promoting ads that hit target cost per acquisition within 7 to 10 days. No data shown.
“Anything hitting your target CPA in the first 7-10 days gets promoted to ASC”
Jon Loomer prefers Meta's creative testing tool (2 to 5 ads in the live ad set, even spend) to separate test campaigns. Launching 20 to 50 ads at once is wasteful, he says, since Meta puts budget on a few.
“More ads won't get you there. Genuinely diverse ads will.”
William Kast says to test creative with one ad set per concept, each holding two to three variations that change a single thing.
“You have one ad set per concept that you want to test, and one concept is the same ad with two different variations.”
Jon Loomer says forcing Meta to deliver your budget through the creative testing tool is less efficient and gives suboptimal results.
“since you're forcing meta to deliver your budget via a test, it's less efficient and results will be suboptimal.”
Eric Zaworski says a healthy account keeps 15 to 25% of spend in testing and 75% or more on ads that graduated to scaling. Above 25% in testing is unmanaged scaling.
PPC Hero says to replace top, middle and bottom-of-funnel campaigns with 1 to 2 scaling campaigns plus 1 testing campaign, organized by business goal rather than audience.
Segwise, which sells creative tracking, says testing inside a target-ROAS campaign starves unproven ads. Its fix is a separate ABO test campaign with 20 to 30 percent of spend, promoting only proven winners. No data.
“unproven variants get starved”
Dara Denney says top brands mostly double down on what works. At high budgets ABO testing helps you learn from every creative while CBO is more efficient, and she warns against dogma.
A Meta field experiment by Pal and Susarla found the algorithm alone shifted female impression share by 2.27 points, so two-arm creative tests mix up a creative's effect with Meta's targeting.
“a conventional two-arm test would attribute the entire delivery shift to the creative.”
In the full answer
For readers who sign in- What to do this week3 steps
- The numbers behind it3 figures from cited sources
- Where each source standsA chart placing each source on the question
- The strongest disagreementThe best case against the answer, from a cited source
- How sure we areWhat its confidence rests on, and what would change the answer