All answers

The Prevailing
Answers for
Business model:
Business model, pick up to two

What should we do with new ads Meta won't spend on: push delivery to them, move them to another campaign, or let them go?

The PrevailingUpdated Nov 12, 2026

Our answer

The answer depends on spend.

Leave a starved new ad alone and launch visibly different new ads. Push delivery only for a weak ad set, at about 10 to 30% of budget for a few days, or for a short promotion until it ends.

Above $200K a month, leave starved ads running instead of pushing them one by one, and keep launching distinct concepts. Optionally retest them in a recovery campaign before writing them off.

On a small budget, add genuinely new ads to your live ad set a few at a time. If the batch gets nothing for a week, test it in its own fixed-budget ad set.

Why this is the answer

3 reasons and our judgment

Most new ads get little spend, and that's normal. Winners take 56.5% of spend at $200K to $1M a month and 63.7% above $1M (2026 benchmark from Motion, a creative analytics software maker).

A handful of ads carries most of the spend. Across about 70,000 ads, 5 to 10% of ads carry 80 to 90% of spend (Breezeway, a Shopify ad analytics vendor).

Pushing a starved ad usually just confirms Meta had it right. Jon Loomer, a Meta ads educator, expects that outcome and says habitual pushing risks driving up costs.

Our judgment

  • Our push delivery share of 10 to 30% of the ad set's budget spans Jon Loomer's figures. It runs from his 10% per ad to his example move from about 10% to 30%, and a few days matches his advice to push a low-delivery ad in a weak ad set for a few days.
  • Keep push delivery for promotions too small for their own campaign. Loomer advises pushing for promotions instead of building a separate campaign, and we reconcile that with separate scheduled campaigns for planned sale windows (Campaign structure for sale promotions, Leading with a discount or value and Budgets and launch timing for seasonal peaks). Adding the promotion's ads is a significant edit that restarts the ad set's learning.
  • When new concepts as a group get no spend, treat it as a structure problem (Where to test new creative). Loomer calls new, distinct ads the most reliable lever and warns against habitual pushing, and Common Thread Collective adds conceptually different creative first when delivery falls short.
  • Tying Common Thread Collective's under-delivery sequence to $200K or more a month is our extension. This question's narrower answer applies Common Thread Collective's under-delivery sequence, which comes with no spend level.

Our judgment, built from Jon Loomer, Meta and Common Thread Collective: our reading of these sources, not a finding from any one of them.

What the sources say

14 statements from 8 sources
  • Barry HottNov 9, 2026Caveat

    Barry Hott trusts Meta's system more than humans to pick ads and says $5 of spend can teach it something, but with little data it favors attention-getting ads over efficient sellers.

    “With limited data, the system will tend to favor more attention-getting ads, which isn't always what sells efficiently.”

    Read it on x.com

  • SingularOct 26, 2026Supports

    Singular, a creative analytics seller, finds a few creatives carry delivery: on Facebook the top five get about 80% of impressions. Installs per 1,000 impressions peaked at 13 to 54 creatives per $1,000 of spend.

    “In campaigns running more than 6, the top asset gets about a third of impressions, and the top five get roughly 80%.”

    Read it on singular.net

  • MetaOct 15, 2026Caveat

    Meta says targeting, creative or optimization event changes and adding a new ad are always significant edits, while a budget change may or may not restart learning, depending on its size. No safe percentage is given.

    “if you increase your budget from $100 to $101, that isn't likely to cause one or more ad sets to reenter the learning phase.”

    Read it on facebook.com

  • MetaOct 15, 2026Caveat

    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.”

    Read it on facebook.com

  • MetaSep 20, 2026Supports

    Meta's help pages, as reported by Terrads, say it is fine when one ad gets few results while others on the same budget do. Its creative test only aims to spend the share set, at no more than 20% of budget.

    “If your ad is not receiving enough results but other ads using the same budget are, that's okay.”

    Read it on terrads.com

  • AgentioAug 12, 2026Disagrees

    Agentio's data on 65,000 Meta Partnership Ads ($130M, 137 brands): 1 in 5 tested ads becomes a winner, and 45% of winners look like losers before $100 of spend. It suggests about $1,000 before killing one.

    “45% of eventual winners still look like losers when performance is assessed before reaching $100 in spend.”

    Read it on agentio.com

  • BreezewayJul 13, 2026Supports

    Breezeway, a seller, says in its data on nearly 70,000 Meta ads the average ad gets about 8 days of spend and only 5% of ads carry 80 to 85% of budget, so launch distinct new ads.

    “only 5% of your ads will ever carry 80 to 85% of your total budget”

    Read it on alleypodcast.buzzsprout.com

  • BreezewayJul 13, 2026Supports

    Josh Coffy says switching his agency's Meta ads to industry call-outs cut CPA almost in half for four weeks; in a client test, ads differing only in call-out text got no spend on 9 of 10. No control group.

    “How calling out specific verticals in ad creative cut CPA nearly in half for four straight weeks”

    Read it on alleypodcast.buzzsprout.com

  • Common Thread CollectiveJul 7, 2026Supports

    For under-delivery, CTC adds conceptually new creative, tries other bid types, then moves ads that spent under 3x the CPA target into a recovery campaign before loosening targets. No outcome numbers.

    “Give them another shot in a dedicated environment before writing them off.”

    Read it on commonthreadco.com

  • Jon LoomerJul 2, 2026Supports

    Jon Loomer says one ad taking 50% of budget is normal and splitting ad sets to force delivery hurts. Push delivery can test an underused ad, though it usually confirms Meta had it right.

    “one ad taking 50% of budget is normal and usually not a problem.”

    Read it on jonloomer.com

  • Jon LoomerJun 11, 2026Supports

    In a small test by Jon Loomer, Meta split budget accurately across five pushed ads over the first $75.02 of spend. Meta's warning about minimum spend per ad was only a soft warning. No result figures.

    “Using the first $75.02 of the budget spent since making this change and ignoring partial days, Meta is pushing budget accurately.”

    Read it on jonloomer.com

  • Jon LoomerJun 9, 2026Supports

    Jon Loomer tested Meta's push delivery, which gives one ad a chosen share of ad set budget for 1 to 30 days, and advises using it sparingly, such as for promotions. No CPA data.

    Read it on jonloomer.com

  • Jon LoomerMay 27, 2026Supports

    Jon Loomer says test new Meta features only to solve a specific problem, not when results are already good, and treat AI connectors with extra caution.

    “why you should only experiment when you have a specific problem to solve rather than tinkering when results are already good”

    Read it on pubcast.jonloomer.com

  • MotionApr 17, 2026Supports

    Motion's benchmark of $1.29B of Meta spend across 6,015 advertisers found winning ads take 23% of spend in accounts under $10K a month and 63.7% at $1M+. Spend on losers is part of finding winners.

    “isn't pure waste”

    Read it on motionapp.com

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

Read next

More in Creative volume and testing