The course aims at providing students with a conceptual framework and a toolbox for optimization of online advertising campaigns (inside sites, apps, games). At the end of the course the student should be able to design and possibly implement real-life systems for optimization of campaigns performance, intended in financial and marketing terms. The required mathematical background is limited to basic differential calculus and probability theory. The treatment is quantitative and concepts will be translated in formulas and algorithms. Nevertheless, focus will be on intuition and business meaning more than on formal rigor.
Contents
• The online advertising ecosystem. Advertisers, publishers, business intermediaries, technology providers, data providers. Trends and Programmatic Advertising.
• Online advertising campaign management: design, targeting, creation, monitoring, optimization and reporting.
• Data about people and their behavior. Classical segmentation, micro-segmentation, one-to-one relationships. Data management platforms.
• The publisher problem. Basic micro-economic concepts and decision theory: expected utility, marginal utility, pricing, decision trees, value of information, risk and uncertainty, opportunity cost, equilibrium and optimality.
• The advertiser problem. Market segmentation, customer profiling. The advertisers-publishers game.
• Forecasting visitors and campaigns behavior. Classical methods: linear regression, logistic regression, time series analysis. Factorization methods. Markovian methods.
• Learning and optimization. Facing uncertainty. The Exp-Exp dilemma. Multi-armed bandits. Reinforcement learning.