Audience Classification Model Using Consolidated User Event Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems struggle to accurately differentiate high-value customers from low-value customers and predict future purchases due to disconnected sources of customer interaction data across multiple devices and identifiers.
Innovation Solution
The system combines information from multiple users and sources to predict future behavior by identifying personas through activity consolidation and using machine-learning models to generate scores indicating the probability of a user belonging to a specific audience or making a purchase.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If customer data is collected from multiple sources and devices, then predictive accuracy is improved, but data consolidation complexity increases
Solution Approach 1:
The patent segments customer data into distinct identifiers (e.g., email, phone number, device ID) and consolidates them into unified customer profiles. This segmentation allows the system to handle multiple data sources systematically by processing and linking identifiers in discrete steps, reducing the complexity of consolidating heterogeneous data from various sources.
Solution Approach 2:
The patent introduces an intermediary data consolidation layer that acts as a mediator between multiple data sources and the predictive model. This intermediary structure standardizes and harmonizes data from different sources before feeding it into the machine learning model, thereby improving predictive accuracy while managing the complexity through a structured intermediate representation.
2Measurement precision
If multiple customer identifiers are used to track users, then customer recognition accuracy is improved, but system complexity increases
Solution Approach 1:
The patent merges multiple customer identifiers (email, phone, device ID, IP address) into a unified customer profile or persona. By combining these identifiers and linking them to the same customer record, the system achieves accurate customer recognition across different touchpoints while managing complexity through a centralized profile structure that consolidates rather than proliferates identification methods.
Solution Approach 2:
The patent creates a universal customer profile structure that can accommodate multiple identifiers and data sources through a single standardized framework. This multi-functional profile serves as a common reference point for recognizing customers across different devices and channels, improving recognition accuracy without requiring separate complex systems for each identifier type.
3Adaptability or versatility
If anonymous browsing data is collected, then customer coverage is improved, but data reliability decreases
Solution Approach 1:
The patent applies partial action by collecting and processing only the necessary data from anonymous browsing sessions to create provisional customer profiles. Rather than requiring complete and reliable data from all sources, the system can operate with partial information from anonymous sessions, improving customer coverage while accepting that some data may be less reliable until confirmed through authenticated sources.
Solution Approach 2:
The patent implements feedback mechanisms where anonymous browsing data is continuously refined and validated as customers interact with the system through authenticated channels. Over time, the system uses feedback from confirmed customer behavior patterns to improve the reliability of profiles initially created from anonymous data, allowing the system to maintain broad coverage while progressively enhancing data reliability.
Data Source
AI summary
Methods, systems, and computer programs are presented for estimating if a user belongs to an audience category. One method includes an operation for accessing events generated at a website. Each event comprises a data structure describing an operation performed by a user, from a group of users, when accessing the website. Further, the method includes an operation for providing event information and information of a first user, for a predefined time window, as input to an audience machine-learning (ML) model. The audience ML model is trained with training data comprising values for features that include event features, user information features, and audience labels. The method further includes operations for generating, by the audience ML model, a score for the first user indicating a probability that the first user belongs to the audience, and for determining if the user belongs to the audience based on the score.