Tip #2: Custom Algos in Programmatic Ads

This piece explains Custom Algos as bespoke, machine-driven optimization for individual advertisers that sits outside vanilla DSP parameterization. A private third-party algorithm can override traditional DSP decisioning, enabling optimization across numerous levers (creatives, ad sizes, audiences, devices, inventory) to find privileged strategies. The concept is linked to the Multi-Armed Bandit problem, illustrating how the algo tests many combinations to maximize outcomes with less human input. It notes typical DSP tech fees of around 10%, and cautions that adding a private custom algo can introduce an additional ~10% in fees, potentially increasing disintermediation in the supply chain. The narrative contrasts the brute-force, human-based approach with algorithmic precision, and highlights considerations about whether the cost and complexity deliver commensurate results. It also mentions environmental implications of programmatic redirects and previews Tip #3 on working media cost waterfalls.