Dynamic Digital Campaign Bid Adjustment via ML Consumer Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Merchants face inefficiencies in generating digital component campaigns to target specific consumer subsets, often leading to computationally intensive processes and self-bidding during auctions, as existing systems lack the ability to dynamically adjust bids and conversion values based on real-time consumer identification.
Innovation Solution
A machine learning model is trained to predict whether a consumer is new or a high-value consumer, allowing for dynamic adjustment of bid and conversion values in real-time, using ground truth data and consumer qualifications to determine qualifying consumers and adjust campaign values accordingly.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If merchants generate multiple digital component campaigns for different consumer subsets, then they can target specific consumers effectively, but it increases time consumption and computational complexity
Solution Approach 1:
The patent combines multiple consumer subset targeting capabilities into a single digital component campaign by using a machine learning model that dynamically identifies and segments consumers in real-time. Instead of creating separate campaigns for new consumers, existing consumers, high-value consumers, and low-value consumers, the system processes all consumer types through one unified campaign with automated ML-based segmentation and bid adjustment.
Solution Approach 2:
The system dynamically adjusts bid values and conversion values based on real-time consumer identification using a machine learning model. The bid value adjustment is probabilistic, calculated as: adjusted_bid = base_bid + (additional_bid × probability_that_consumer_is_new_or_high_value). This dynamic adjustment occurs automatically during the auction process without requiring pre-configured separate campaigns for different consumer segments.
2Adaptability or versatility
If merchants generate multiple digital component campaigns for different consumer subsets, then they can target specific consumers effectively, but it reduces productivity due to time-consuming manual creation
Solution Approach 1:
The system enables self-service automated campaign management where the machine learning model automatically identifies consumer types, determines appropriate bid adjustments, and optimizes conversion values without merchant intervention. The ML model continuously learns from conversion data and ground truth information about consumer characteristics, allowing the system to autonomously manage targeting for multiple consumer subsets within a single campaign.
3Reliability
If merchants bid for specific consumer subsets using multiple campaigns, then they can optimize bids for new and high-value consumers, but they may bid against themselves during auctions
Solution Approach 1:
The system segments consumers into different categories (new vs. existing, high-value vs. low-value) using a machine learning model that processes consumer data in real-time during the auction. This segmentation occurs within a single campaign framework, allowing the system to apply different bid adjustments for different consumer types without creating multiple separate campaigns that could conflict with each other.
Solution Approach 2:
The system dynamically changes bid parameters based on consumer classification. The bid value is adjusted using the formula: adjusted_bid = base_bid + (additional_bid × probability_that_consumer_is_new_or_high_value). This parameter change occurs automatically during the auction process based on ML model predictions, eliminating self-bidding conflicts while maintaining optimized bids for different consumer segments.
Data Source
AI summary
The technology is generally directed to determining whether a consumer being presented with a digital component is a new and/or new qualifying consumer at the time the digital component is being selected. A machine learning model may be trained and used to predict whether the consumer is a new and/or new qualifying consumer. The prediction may be a probability representing the likelihood that the user is a new and/or new qualifying consumer for a given merchant. The probability may be used to dynamically adjust the merchant's bid at the time of auction to have the merchant's digital component selected. The probability may, in some examples, may be used to dynamically adjust the conversion values after the merchant's digital component is provided for output.