This field record asks: What share of head-to-head spend does an identical creative copy receive against its still-running original in the first days after launch? It is part of a seven-study Darwin IQ Labs series on early signals for newly introduced Meta creatives (Nos. 113–119).

Among natural duplicates—identical image/video sets re-added while the original still spent—from clients sampled of various industries (67 head-to-head pairs in 38 ad sets), the copy takes a median 74% of head-to-head spend in the first three days (95% CI 58–82%), falling to ~even by week 2 and 36% by week 3. New ads receive a launch bonus, not a handicap.

Sample

Dataset EMR-2026-10E: ads created in a ~150-day lookback across active Meta accounts; 67 head-to-head duplicates after excluding swaps where the original stopped delivering. Clients sampled of various industries.

Method

Creatives are matched on the exact set of image hashes and video IDs. Head-to-head share = copy spend ÷ (copy + original). Late-add checks compare a new ad’s early share versus the incumbent top spender against later spend retention. Observational natural experiments — buyers chose relaunches.

Findings

Launch bonus, then fade. There's no handicap; new ads get a launch bonus. An identical copy of a running creative takes 74% of the head-to-head spend against its original in the first 3 days ($2.85 per $1, 67 natural duplicates), falls back to about even by week 2 and trails by week 3. A challenger has to beat that bonus, not parity, to look better than the incumbent.

Identical copies take most early spend, then fall back toward parity

Median copy ÷ (copy + original) by day since the copy’s first impression

50% barDay 061%Day 188%Days 0–274%Week 248%Week 336%

Source: Darwin IQ Labs, dataset EMR-2026-10E · 50% = parity

Clearing roughly half the copy baseline predicts later spend retention (~1.6× in the weekly sample); clearing the full early bonus does not.

What this record does not claim

  • No creative assets are released with EMR-2026-10E.
  • Relaunches are selected by buyers (often prior winners); fatigue on the original may inflate the bonus.
  • Swaps where the original was paused are excluded from the head-to-head sample.