Explainable Prediction System for CRM Attribution
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional CRM systems lack advanced predictive segmentation and attribution capabilities, failing to provide explainable predictions that account for causal events and offer insights into customer behavior across multiple communication channels.
Innovation Solution
A computer-implemented method and system for generating explainable predictions, including predictive segmentation and attribution, by receiving user objectives, processing data sets, determining activities, generating attribution models, and providing predictions with rationales, using techniques such as SPMF algorithms, Shapley models, and SHAP algorithms to identify causal sequences and channel effectiveness.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If conventional CRM systems use traditional data analysis methods, then they can provide basic reports and analyses, but they lack advanced predictive segmentation and attribution capabilities
Solution Approach 1:
The patent segments customers into distinct groups based on predicted behaviors and characteristics using machine learning models. This enables predictive segmentation by dividing the customer base into meaningful segments (e.g., high-value customers, at-risk customers) based on analytical results rather than traditional demographic criteria, thereby enhancing adaptability without proportionally increasing system complexity.
Solution Approach 2:
The patent introduces attribution models as intermediary components that connect marketing activities to customer outcomes. These models serve as mediators that analyze multi-channel interactions and determine causal relationships between marketing events and customer responses, enabling advanced predictive capabilities while managing system complexity through modular architecture.
2Loss of information
If conventional CRM systems provide backward-looking reports, then they can analyze historical data, but they cannot provide explainable predictions with rationales
Solution Approach 1:
The patent implements feedback mechanisms where the system not only generates predictions but also provides explanations and rationales for those predictions. The attribution models analyze historical data patterns and feed back causal relationships to explain why certain predictions are made, thereby reducing information loss about prediction reasoning while managing analysis complexity through structured feedback loops.
Solution Approach 2:
The patent performs preliminary analysis of historical data to establish attribution models and prediction frameworks before generating new predictions. By pre-processing historical interactions and establishing causal relationships in advance, the system can provide explainable predictions with rationales without performing complex real-time analysis, thereby reducing information loss while controlling analysis complexity.
3Loss of information
If conventional CRM systems track multiple communication channels, then they can collect comprehensive data, but they cannot evaluate causal events or provide attribution
Solution Approach 1:
The patent introduces attribution models as intermediary analytical components that process data from multiple communication channels and identify causal relationships. These models act as mediators between raw multi-channel data and business insights, evaluating which events causally influenced customer outcomes while managing the complexity of analyzing cross-channel interactions through structured attribution frameworks.
Solution Approach 2:
The patent replaces traditional mechanical tracking methods with machine learning-based attribution models that can automatically identify causal relationships across multiple channels. Instead of manually tracking and analyzing each interaction, the system uses computational models to substitute complex manual analysis with automated causal inference, thereby recovering causal attribution information while managing model complexity through algorithmic approaches.
Data Source
AI summary
Provided are computer-implemented systems and methods for providing explainable predictions, including receiving a prediction objective from a user; providing at least one data set from at least one data source; determining, at a processor, at least one activity from the at least one data set, the at least one activity comprising a feature of the corresponding data set; generating, at the processor, at least one attribution model from the at least one feature, the at least one attribution model operative to provide a prediction and an associated explanation; generating an explainable prediction comprising a prediction rationale based on the prediction objective received from the user and the at least one attribution model.