How should eco DTC test and kill fatigued ads without vanity creative tests?
Quick answer
Kill or iterate when frequency rises and CPA or new-customer share worsens, not when CTR dips for a day. Test new concepts and first-three-seconds hooks in a holdout cell; do not refresh by inventing stronger eco claims. Budget split lives in the creative vs media guide; weekly production in the UGC system.
Budget vs production vs this cadence
| Guide | Job |
|---|---|
| Creative budget | How much to spend on production vs media |
| UGC system | Weekly briefs, claims QA, rights |
| This guide | When to kill, iterate, or launch a new concept, and how to test it |
Fatigue signals (use together)
- Frequency on the winning audience climbing while CPA or ROAS worsens
- New-customer share falling (harvesting warm buyers)
- Hold rate or thumb-stop falling week over week on the same body
- Comment quality shifting to “seen this” or claim complaints
- Policy or claim tickets: pause immediately, do not “let it spend”
One noisy day of CTR is not a kill. Look at 3 to 7 days of contribution, not vanity rank.
Kill, iterate, or new concept
| Decision | When | What to ship |
|---|---|---|
| Iterate hook | Concept still converts; first 3 seconds tired | New open, same proof body |
| Iterate proof | Hook still stops; offer or materials line is stale | New proof clip, same claim kit |
| Kill | CPA and new-customer share both worse at high frequency | Stop. Do not raise budget to “give it another week” |
| New concept | Hook iterates failed twice, or audience saturation | Different job (demo, objection, refill, founder), still scoped claims |
Test design that is not vanity
- One variable: hook or concept, not five captions at once
- Enough spend: learning volume, not $20 “tests”
- Primary KPI: CPA or contribution and new-customer %, not CTR
- Claim freeze: variants use the same approved kit; see claims macros
- Holdout: keep a control winner live so you do not restart learning on the whole account
Advantage+ will remix whatever you upload. Fatigued libraries plus automation scale the tired ad. Guardrails: Advantage+.
Weekly cadence
- Monday: flag ads that hit frequency + CPA rules
- Midweek: ship 1 to 3 hook iterates or 1 new concept from the UGC pipeline
- Friday: kill losers, keep one control, log claim QA
If nothing new can go live, you have a production problem, not a media problem. Fund the pipeline per the budget guide.
Measurement
| Metric | Why |
|---|---|
| Days to kill after fatigue flag | Ops speed |
| Hook iterates vs net-new concepts / week | Balance |
| CPA and new-customer % on tests vs control | Real outcome |
| Claim incidents on “refreshes” | Eco failure mode |
Refresh the hook, not the claim
Tired ads need a new first three seconds or a new job. They do not need a louder unsourced “eco” line.
Frequently asked questions
How should eco DTC know an ad is fatigued?
Frequency up plus CPA or new-customer share down, confirmed over several days. CTR alone is not enough.
Should we kill a winner that still spends?
If contribution is decaying and iterates failed, yes. Raising budget on a tired winner usually harvests cheaper until it does not.
Hook test or new concept first?
Hook first if the body still converts. New concept if two hook iterates failed or the job is exhausted.
Can we refresh with a stronger green claim?
No. Same claim kit. Stronger language is not a creative test.
How often should new ads go live?
Weekly enough that 1 to 3 iterates or a concept can replace kills. Empty weeks mean the UGC pipeline is stalled.