{"id":2883,"date":"2026-05-29T16:43:28","date_gmt":"2026-05-29T16:43:28","guid":{"rendered":"https:\/\/optimix51-optimix-blog.hf.space\/methodology\/"},"modified":"2026-05-29T16:43:28","modified_gmt":"2026-05-29T16:43:28","slug":"methodology","status":"publish","type":"page","link":"https:\/\/www.aureliansystems.tech\/blog\/methodology\/","title":{"rendered":"Methodology &#8211; OptiMix | How Bayesian MMM Works"},"content":{"rendered":"<p>OptiMix uses <strong>Automatic Differentiation Variational Inference (ADVI)<\/strong> \u2014 a modern approximate Bayesian inference algorithm \u2014 to estimate marketing channel effects. Unlike MCMC methods that take hours or days, ADVI runs in minutes while producing full posterior distributions over every channel contribution.<\/p>\n<h2>The Media Mix Model<\/h2>\n<p>The core model is a hierarchical Bayesian regression accounting for: <strong>Baseline (Organic)<\/strong> sales, <strong>Saturation<\/strong> (diminishing returns at higher spend), <strong>Adstock<\/strong> (carry-over decay), <strong>Trend &amp; Seasonality<\/strong>, and <strong>Control Variables<\/strong> (price, promotions, competitor activity).<\/p>\n<h2>Privacy-First<\/h2>\n<p>OptiMix runs on <strong>aggregated weekly spend and revenue data<\/strong> \u2014 no PII, no cookies, no third-party tracking. GDPR and CCPA compliant out of the box.<\/p>\n<h2>Minimum Data Requirements<\/h2>\n<p><strong>26 weeks<\/strong> of weekly spend and revenue data across two or more marketing channels. 52+ weeks enables seasonal pattern detection. 2+ years separates long-term trend from media impact.<\/p>\n<h2>What OptiMix Does Not Do<\/h2>\n<p>It does not replace incrementality testing, does not provide real-time attribution, does not optimize creative or audience targeting, and does not manage automated bids. See the <a href=\"\/pricing\">pricing page<\/a> to get started.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>OptiMix uses Automatic Differentiation Variational Inference (ADVI) \u2014 a modern approximate Bayesian inference algorithm \u2014 to estimate marketing channel effects. Unlike MCMC methods that take hours or days, ADVI runs in minutes while producing full posterior distributions over every channel contribution. The Media Mix Model The core model is a hierarchical Bayesian regression accounting for: [&hellip;]<\/p>\n","protected":false},"author":1,"featured_media":0,"parent":0,"menu_order":0,"comment_status":"closed","ping_status":"closed","template":"","meta":{"footnotes":""},"class_list":["post-2883","page","type-page","status-publish","hentry"],"_links":{"self":[{"href":"https:\/\/www.aureliansystems.tech\/blog\/wp-json\/wp\/v2\/pages\/2883","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.aureliansystems.tech\/blog\/wp-json\/wp\/v2\/pages"}],"about":[{"href":"https:\/\/www.aureliansystems.tech\/blog\/wp-json\/wp\/v2\/types\/page"}],"author":[{"embeddable":true,"href":"https:\/\/www.aureliansystems.tech\/blog\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.aureliansystems.tech\/blog\/wp-json\/wp\/v2\/comments?post=2883"}],"version-history":[{"count":0,"href":"https:\/\/www.aureliansystems.tech\/blog\/wp-json\/wp\/v2\/pages\/2883\/revisions"}],"wp:attachment":[{"href":"https:\/\/www.aureliansystems.tech\/blog\/wp-json\/wp\/v2\/media?parent=2883"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}