FMCG · Creator Commerce
Creative should be a learning and scaling loop, not a video inventory.
D2 organized creative around hooks, angles, formats and creator profiles so UGC/KOC output could feed Ads, GMV Max and TikTok Shop learning.
Brand
E.O Wins Coffee
Focus
Creative testing, KOC/creator and commerce growth loop
Problem
The operating problem D2 needed to isolate
A useful case study connects the problem to source data and a repeatable decision rule.
- High content demand could turn production into a volume exercise.
- Creator bookings were disconnected from learning.
- Creative, Ads and shop performance lacked a shared measurement language.
Evidence
First-party data used in the analysis
The case is grounded in operating sources that answer different questions. D2 keeps sales-period, settlement and cost evidence separate before reconciling them.
- Creative production log
- Creator/KOC pipeline status
- Shop performance
- Ads / GMV Max signals
- Conversion data when available
Methodology
How the evidence was turned into a decision framework
The objective is a repeatable and traceable review method rather than a one-off optimization narrative.
- Tagged creative by hook, angle, format and creator profile.
- Recorded keep/kill/iterate signals after distribution instead of reviewing delivery count alone.
- Fed learning back into the next creator and creative brief.
Action
How the operating loop was reorganized
Execution changes are documented against the data and decision rule they are intended to improve.
- Built a testing matrix by hook, angle, format and creator profile.
- Tracked winning and losing patterns after distribution.
- Converted learning into the next creative brief.
Outcome
The most important output is better decision quality
D2 only publishes performance figures when the measurement period and reconciliation basis are sufficiently verified.
- Creative received clear keep/kill/iterate rules.
- Creator pipeline connected to commerce rather than delivery count.
- Performance learning fed back into production.
Publication standard
What this case does and does not claim
This case focuses on operating method and testing logic. D2 does not publish an absolute lift or ROAS claim without a confirmed comparison period and measurement basis.
Next step