Case 12 · Content Factory Multi-Model A/B · Client Conversion +21% · Per-Piece Cost -30% · Human Review -50%¶
TL;DR: Content Factory Multi-Model A/B · Client Conversion +21% · Per-Piece Cost -30% · Human Review -50%. Direction: Multi-Model Routing. Scale: content agency. Hard KPI: client conversion +21%, per-piece cost -30%, human review -50%. Tech: per-prompt A/B across GPT-4o / Claude 3.5 / AW36-Max / DeepSeek, auto-selection. My role: A/B platform, scoring model, client integration. Client name anonymized under NDA.
- Direction: Multi-Model Routing
- Industry: content-marketing
- Scale: content agency
- Hard KPI: client conversion +21%, per-piece cost -30%, human review -50%
Problem¶
Content-agency client outcomes vary widely; different models suit different styles; no single model fits all clients.
Solution¶
Per-prompt A/B across GPT-4o / Claude 3.5 / AW36-Max / DeepSeek, auto-select by outcome scoring.
Tech Stack¶
per-prompt A/B across GPT-4o / Claude 3.5 / AW36-Max / DeepSeek, auto-selection
Result (Hard KPIs)¶
client conversion +21%, per-piece cost -30%, human review -50%
My Role¶
A/B platform, scoring model, client integration
Related¶
Next: Download PDF · 30-min alignment
Client name anonymized · KPIs verified · Real client name disclosable under NDA.