Non-parametric Attribution Estimation via Coalitional Game Theory
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional marketing attribution models, whether rule-based or algorithmic, often fail to accurately capture the influence of multiple marketing channels on user conversion decisions due to assumptions about relationships between channel exposures and user responses, leading to inaccurate estimations.
Innovation Solution
A non-parametric estimation approach is implemented in a computing environment to analyze user interactions across multiple marketing channels, using a coalitional game method to estimate attributions without assuming relationships, allowing for real-time attribution analysis and resource allocation optimization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If rule-based models are used to assign attribution to marketing channels, then the attribution process is simple and fast, but the accuracy of capturing the influence of multiple channels is poor
Solution Approach 1:
The patent transitions from rule-based models to algorithmic models that use regression functions with multiple parameters (beta coefficients) to represent channel influences. This allows the system to capture complex relationships between marketing channels and conversions while maintaining computational efficiency through established statistical methods.
Solution Approach 2:
The patent introduces an intermediary layer of algorithmic processing that mediates between simple rule-based assignments and complex actual relationships. The regression model acts as an intermediary that can approximate non-linear relationships while remaining computationally tractable and interpretable.
2Measurement precision
If algorithmic models with assumed relationships are used to estimate marketing attributions, then the model structure is simple and computable, but the estimation accuracy is poor due to incorrect assumptions
Solution Approach 1:
The patent moves from static assumed relationships to dynamic learning approaches where the model adapts to actual data patterns. The system learns the true relationships between marketing channels and conversions from historical data rather than relying on pre-specified assumptions, allowing it to capture non-linear effects and synergies.
Solution Approach 2:
The patent incorporates feedback mechanisms where the model continuously learns from observed conversion data to refine its estimates of channel influences. This allows the system to improve attribution accuracy over time by adjusting its understanding of channel relationships based on actual performance feedback.
3Measurement precision
If non-parametric estimation is used to capture all channel combinations and synergies, then the attribution accuracy is improved, but the computational complexity and data processing requirements increase
Solution Approach 1:
The patent segments the analysis into manageable components by examining marketing channel combinations systematically. Rather than attempting to model all possible interactions simultaneously, the approach breaks down the problem into individual channel effects and their combinations, making the computational task tractable while still capturing synergistic effects.
Solution Approach 2:
The patent combines multiple estimation approaches into a composite model that leverages the strengths of different methods. By integrating parametric and non-parametric techniques, the system achieves high attribution accuracy while managing computational complexity through the complementary nature of the combined approaches.
Data Source
AI summary
Techniques for analyzing marketing channels are described. Users are exposed to the marketing channels. User responses (e.g., purchases and no-purchases) to the exposures are tracked. Upon a request from a marketer to analyze an attribution of a marketing channel, the user responses are analyzed. The attribution represents the credit that the marketing channel should get for influencing the users exposed thereto into exhibiting a particular user response (e.g., a purchase). The analysis involves multiple steps. In a first step, a non-parametric estimation is used to generate a value function at a user-level. In a second step, a coalitional game approach is used to estimate the attribution based on the value function. A response is provided to the marketer with data about the attribution.