Optimizing marketing buying using auto-bidding algorithms

What the talk is about

Serhii Sukach explains how to automate marketing buying with auto-bidding algorithms. He first covers how to judge buying quality: ROI, a player-level LTV forecast, payback period and the share of inefficient spend. He then compares traffic sources: video networks, where a marketing manager sets bids by hand, and DSP networks that optimize per device. Optimization works at the unit level (project, partner, platform, publisher, country), and there can be tens of thousands of units. For units with history, bids are raised to scale volume or lowered to keep volume at a lower price; for new units, a discovery mode starts from a default bid and switches off weak ones quickly. After adopting the algorithm, all spend came under control and the inefficient share dropped from about 30% to under 7%. Useful for marketers, analysts and user acquisition teams.