Channel Incrementality Ratio Generation Using Machine Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Assessing the efficiency of digital traffic channels is complex and inefficient, as existing methods like online controlled incrementality testing waste resources, time, and provide reduced accuracy.
Innovation Solution
A method using a machine learning model to programmatically generate a channel incrementality ratio based on historical data, including transaction and touchpoint timestamps, sorted lists, and weighting factors for each channel.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If online controlled incrementality testing is used to assess channel efficiency, then measurement capability is provided, but resource waste increases and time consumption increases
Solution Approach 1:
The system performs preliminary actions by collecting and storing touchpoint and transaction data in advance through data extraction modules. Historical data is pre-processed and stored in databases, allowing the machine learning model to generate incrementality ratios without requiring real-time controlled testing, thereby eliminating resource waste while maintaining measurement accuracy.
Solution Approach 2:
The system creates a virtual copy of the controlled testing environment by using machine learning models to simulate incrementality measurements. Instead of conducting actual controlled experiments that consume resources, the system generates predictive incrementality ratios from historical data copies, achieving the same measurement objective without the associated resource costs.
2Measurement precision
If online controlled incrementality testing is used to assess channel efficiency, then measurement capability is provided, but time consumption increases
Solution Approach 1:
The system performs preliminary data collection and model training in advance. Historical touchpoint and transaction data are continuously accumulated and pre-processed, allowing the machine learning model to generate incrementality ratios instantaneously when needed, eliminating the time-consuming controlled testing process while maintaining measurement precision.
Solution Approach 2:
The system replaces time-consuming controlled experiments with a virtual copying approach using machine learning models. The model learns from historical data patterns and generates predictive measurements without requiring actual testing time, thus dramatically reducing time consumption while preserving measurement accuracy.
3Loss of energy
If machine learning model is used to generate channel incrementality ratio, then resource waste is reduced and time consumption is reduced, but measurement precision may be compromised
Solution Approach 1:
The system optimizes measurement precision by carefully selecting and adjusting key parameters including the attribution window period, the specific features fed to the machine learning model, and the model architecture itself. These parameter optimizations ensure that the model generates accurate incrementality ratios from historical data without requiring resource-intensive controlled testing.
Solution Approach 2:
The system implements feedback mechanisms where the generated incrementality ratios are continuously validated and the machine learning model is retrained with new data. This feedback loop ensures measurement precision is maintained and improved over time, compensating for any potential accuracy loss from not conducting controlled experiments.
Data Source
AI summary
Embodiments of the present disclosure provide methods, systems, and apparatuses for computing a channel incrementality ratio using a machine learning model.