Every bold claim on this site is footnoted. This page documents the evidence behind those claims — what we measured, where the data comes from, and when we last reviewed it.
Outcome Bidder uses a gradient-boosted multi-objective model trained on six months of historical viewability, attention, and conversion signals from C Wire campaigns. The model re-prices each bid request in <50 ms based on predicted post-bid outcomes.
Safari (35% of EU mobile traffic) and Firefox block third-party cookies by default. C Wire's contextual + persona engine operates without third-party cookies, so reach in these browsers matches Chrome reach 1:1 — verified across 84 campaigns in 2025.
MatchPersona generates audience personas via a multi-agent LLM workflow (Gemini 2.5 Pro + GPT-5 verification) that combines first-party brand briefs with public signal analysis. Personas map to vector embeddings of in-market URL clusters, replacing third-party cookie segments.
Every URL is scored against the GARM Brand Safety Floor (eleven categories) by a Gemini-3-Flash classifier with context modifiers (news, opinion, satire). Floor-violating URLs are excluded from bid requests. See mem://architecture/garm-safety-and-semantic-analysis-prompt for the rule set.