手法を比較

	Multi-Touch Media Attribution	Uplift Modeling
分野	Marketing Science	Marketing Science
系統	Machine learning	Machine learning
提唱年	2016	2011
提唱者	Eva Anderl et al. (graph/Markov attribution); Ron Berman (Shapley-value attribution)	Nicholas J. Radcliffe & Patrick D. Surry
種類	Credit-assignment model for converting multi-channel exposure paths into channel contributions	Heterogeneous-treatment-effect model for targeting incremental responders
原典	Anderl, E., Becker, I., von Wangenheim, F., & Schumann, J. H. (2016). Mapping the customer journey: Lessons learned from graph-based online attribution modeling. International Journal of Research in Marketing, 33(3), 457-474.	Radcliffe, N. J., & Surry, P. D. (2011). Real-World Uplift Modelling with Significance-Based Uplift Trees. Stochastic Solutions White Paper TR-2011-1.
別名	Multi-Touch Attribution, Data-Driven Attribution, Markov-Chain Attribution, Shapley-Value Attribution	Incremental Response Modeling, True-Lift Modeling, Net-Lift Modeling, Persuadable Targeting
関連	3	3
概要	Multi-touch media attribution distributes credit for a conversion across the sequence of marketing touchpoints a customer encountered, replacing crude heuristics like 'last click gets everything' with models that respect the whole journey. Two principled approaches dominate: graph-based Markov-chain models, advanced by Eva Anderl and colleagues, which represent customer paths as transitions between channels and value a channel by its 'removal effect' on the probability of conversion; and Shapley-value attribution, analyzed by Ron Berman, which treats channels as players in a cooperative game and assigns each its average marginal contribution across all possible coalitions. Both reject single-touch rules because those rules systematically misvalue channels — Berman shows that last-touch over-incentivizes the final exposure and can lower advertiser profit, while Anderl et al. demonstrate that Markov models recover credit allocations markedly different from simple heuristics. The result is a defensible, data-driven map of which channels actually move customers toward conversion, used to reallocate budget and compute channel-level return on ad spend. Because attribution is fundamentally about the incremental effect of exposures, it sits at the boundary of measurement and causal inference.	Uplift modeling targets the people a marketing action actually changes, not the people most likely to buy anyway. Where a conventional response model predicts the probability of purchase, an uplift model predicts the difference a treatment makes — the incremental effect of, say, sending a coupon — and uses it to find 'persuadables' while avoiding 'sure things,' 'lost causes,' and especially 'sleeping dogs' who react negatively to contact. Nicholas Radcliffe and Patrick Surry, pioneers of the technique, formalized significance-based uplift trees that split on the difference in treatment-versus-control response rather than on response alone, and introduced the Qini curve to evaluate incremental gain. Pierre Gutierrez and Jean-Yves Gerardy's literature review situates uplift modeling squarely within causal inference, organizing the main estimation strategies and metrics. Because the quantity of interest is a conditional average treatment effect, uplift modeling is most reliable when built on randomized treatment and control data. The payoff is sharper, more profitable targeting: spend marketing effort where it produces genuine incremental response instead of rewarding behavior that would have happened regardless.