Capital Allocation & Diminishing Returns Engine • Authentic Facebook Ads Dataset
Chi-square test of independence confirms conversion differences across segments are statistically confirmed, not random variation (p = 8.89×10⁻⁶⁸). Recommended shifts are grounded in proven performance gaps—chasing signal, not noise. Statistically indistinguishable cohorts (Women 30–34 tied with Men 35–39; Women 35–39 tied with Men 45–49) are merged into shared reallocation tiers with equal shift percentages and unified rationales.
Slope β₁ = +0.01206. Incremental orders cost $82.89 each vs. Campaign 1178's average CPA of $63.83.
Chi-square test of independence confirms conversion rate differences across demographic cohorts are real, not random sampling variation. Reallocation targets proven statistical gaps instead of chasing noise.
Campaign 916 ($150 spend, p = 0.548) and Campaign 936 ($2.89k spend, p = 0.714) have statistically insignificant regression slopes. Ad spend was too low ($0-$20/ad) to support spend-scaling curves. Maintain as discovery sandboxes.