All answers

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

What share of production should iterate on winners versus invent new concepts?

The PrevailingUpdated Nov 12, 2026

Our answer

Put at least half of new creative production into genuinely new concepts. Spend the rest extending concepts that are still winning.

Confidence
Low
Strongest proof
2 Meta rules
1 dataset, 9 practitioners, 5 opinions or claims
Sources
14 sources
17 statements

New here? Take the 1-minute tour

Why this is the answer

3 reasons and our judgment

Practitioners lean toward new concepts. Jetfuel Agency budgets 50% of creative for new concepts, Opascope runs 5 to 10 new concepts a week and Wpromote weights testing toward concept tests.

New concepts, not iterations, bring more volume at Homestead / Mars Men, a consumer brand agency. Its teams spend little time on iterations as a result.

Meta's fix for a fatigued ad is a new one that looks different. Its help page asks for a new image or video materially different from the original, and says keeping the original active may maximize results.

Our judgment

  • Jetfuel Agency's split comes to about 60% new within new production. That is 50 of 80. Homestead / Mars Men is mostly new, and Wpromote weights testing toward concept tests.
  • Count a concept's first 2 or 3 versions as new. That follows Opascope's 2 to 3 variants per concept, so one new concept comes with every 4 to 6 new ads.
  • An iteration should change what the viewer sees. Change the hook, format, creator or length, not only the copy, since Meta's diversity score compares images and video thumbnails (What counts as a new creative).
  • Keep leaning toward new concepts when one winner carries the account. We side with New Engen over Segwise, which scopes its iteration framework to replacing a winner and writes for app marketers. Jetfuel Agency, Homestead / Mars Men, Wpromote and Ready Set lean new, against Nick Theriot and Fraser Cottrell, and both sides are stated practice.

Our judgment, built from Jetfuel Agency, Homestead / Mars Men, Wpromote, Ready Set, Opascope, New Engen, Meta, Segwise, Nick Theriot and Fraser Cottrell: our reading of these sources, not a finding from any one of them.

What the sources say

17 statements from 14 sources
  • Fraser CottrellNov 11, 2026Disagrees

    Fraser Cottrell, who runs an agency, says to base 80% of a new client's creative on what already wins and make 20% net new for the first 3 months. One client found a couple of winners.

    “base 80% of your ad creative on what's ALREADY winning - make the other 20% completely net new - for the first 3 months”

    Read it on x.com

  • RocketShip HQNov 11, 2026Disagrees

    Karina Tukanova, reported by RocketShip HQ, says Meta's push for 100 new concepts a week has not worked for any client she has seen. What helped was visual variety within what already works.

    “What actually shifted: visual diversity within what already works, not total reinvention weekly.”

    Read it on x.com

  • Nick TheriotNov 10, 2026Supports

    Nick Theriot, who runs an agency, says to test big swings like new desires and angles before small tweaks; he says one account went from $161K to $318K in 30 days, metric not stated.

    “Big swings first, then small tweaks on the winner”

    Read it on x.com

  • Nick TheriotNov 6, 2026Disagrees

    Nick Theriot, who sells ad management, says never rewrite a winning ad's hook; change one visual at a time and, once a version takes spend, make 80% of the next batch like it. A rule of thumb, no data.

    “Never rewrite the hook on a winning ad. Change the visuals instead.”

    Read it on x.com

  • Search Engine LandOct 30, 2026Supports

    Akvile DeFazio says more ads will not revive declining paid social results. Test distinct concepts, hooks and messages rather than small variations of one winner, isolating one variable at a time.

    “Test distinct concepts rather than launching minor visual variations of a single winning asset”

    Read it on newsletter:Search Engine Land:2026-10-30

  • Marin IstvanicOct 18, 2026Caveat

    Marin Istvanic relaunches old ads that spent over $10K and did well, using the same post ID to keep their social proof, in the testing campaign. He says 6 or 7 of 10 become winners again.

    “6-7 out of every 10 revivals become active winners again.”

    Read it on x.com

  • MetaOct 15, 2026Supports

    Meta flags 'Creative limited' when an ad's cost per result tops your past ads but is under twice as much, and 'Creative fatigue' at twice or more. It suggests a materially different new ad.

    “Keeping your original ad active instead of pausing or turning it off may maximize results”

    Read it on facebook.com

  • MetaOct 15, 2026Supports

    Meta's creative diversity score compares the images and video thumbnails in an ad set for similarity and labels it High, Medium or Low. It names no copy input and is still in development.

    “is calculated by comparing the images and video thumbnails in your ad set to each other to determine similarity.”

    Read it on facebook.com

  • Kurt ElsterSep 29, 2026Disagrees

    Jay Wright says most new creatives should be iterations on elements of winning ads in your account, not net-new concepts.

    “the majority of creatives shouldn't be net new It should be iterations from elements of winning ads that you're seeing in your account”

    Read it on unofficialshopifypodcast.com

  • Jetfuel AgencyAug 23, 2026Caveat

    Jetfuel, from agency practice with no data, says angles and formats fatigue at different rates: a new format can extend a concept one to two weeks, and a new hook on a tiring concept one to three weeks.

    “The practical implication: exhaust format variations before calling an angle dead.”

    Read it on jetfuel.agency

  • New EngenJul 29, 2026Supports

    New Engen says there is no fixed iteration-versus-new split, so run Scale, Extend and Explore lanes. Iteration gives Meta more versions of an idea; diversification gives it more ideas. No outcome data.

    “Iteration gives Meta more versions of an idea. Diversification gives Meta more ideas.”

    Read it on newengen.com

  • Homestead / Mars MenJul 17, 2026Supports

    Zach Stuck says his team spends less time on iterations and gets more volume from net-new concepts with a different person, format or style. After a winner he tests a new landing page. No data.

    “We actually try not to spend so much time on iterations anymore”

    Read it on motionapp.com

  • Ready SetJul 17, 2026Supports

    Ready Set says diversification explores and iteration exploits, and only diversification gives Meta new signal. It gives no data, only an example of two winning statics iterated into 11 variations.

    “iteration without diversification is motion without momentum”

    Read it on motionapp.com

  • OpascopeJul 17, 2026Supports

    Opascope, an agency citing 200+ audited accounts, says full concept tests give real signal and hook swaps don't; run 5 to 10 concepts a week with 2 to 3 variants each. No method shown.

    “Six versions of "get 20 percent off" on the same video produce no real signal.”

    Read it on opascope.com

  • WpromoteJul 17, 2026Supports

    Wpromote's Darren D'Altorio says to weight testing toward new concepts, not iterations, and limit iteration until a concept finds scale. He gives no fixed split. Vuori tests roughly 1,500 to 2,000 creatives a year.

    “Weight your testing roadmap on concept tests, not iterative testing”

    Read it on motionapp.com

  • Jetfuel AgencyJul 3, 2026Supports

    An agency practice with no data: split creative budget 50% to new concepts, 30% to refined variants of proven ads and 20% to proven ads run until the data says stop.

    Read it on jetfuel.agency

  • SegwiseJun 22, 2026Disagrees

    Segwise, a tool vendor, says when new ads cannot beat an evergreen winner, copy the winner's key attributes and vary the rest. For scale, it says feed Meta more distinct concepts.

    “your variants must be different enough to read as distinct entities while still preserving the winner's load-bearing attributes.”

    Read it on segwise.ai

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