Response Function for Online Advertising Bidding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Online advertising campaigns face challenges in efficiently managing and optimizing bidding strategies based on dynamic price and event rate data, leading to suboptimal impression allocation and event generation.
Innovation Solution
The system employs a method of generating discrete and continuous functions from campaign transaction signals, using recursive binning and temporal filtering to create a response function that informs bidding decisions, optimizing the relationship between price and event rate for improved campaign performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional bidding strategies are used in online advertising campaigns, then implementation is simple, but impression allocation is suboptimal and event generation is inefficient
Solution Approach 1:
The patent segments the continuous price and event rate data into discrete bins, creating a binned representation that simplifies computation while preserving essential patterns. This segmentation enables efficient processing of bidding data without losing critical information for optimization decisions.
Solution Approach 2:
The system performs preliminary processing of price and event rate data by creating binned histograms and response functions before actual bidding decisions are made. This pre-computation of statistical relationships allows for faster, more efficient real-time bidding without complex calculations during the bidding moment.
2Measurement precision
If detailed transaction data is processed for every bidding decision, then bidding accuracy is improved, but computational costs and memory requirements increase
Solution Approach 1:
The patent extracts essential statistical patterns from detailed transaction data by creating binned histograms that capture the relationship between price and event rate. Instead of processing every individual transaction, the system extracts the underlying distribution patterns, significantly reducing computational requirements while maintaining bidding accuracy.
Solution Approach 2:
The system creates a simplified copy of the complex transaction data in the form of binned histograms and response functions. This copied representation preserves the essential relationships needed for accurate bidding decisions while requiring minimal computational resources for processing.
3Productivity
If real-time optimization of bidding strategies is implemented, then campaign performance is improved, but system complexity and processing time increase
Solution Approach 1:
The system performs preliminary computation of binned histograms and response functions that capture the price-event rate relationship. This pre-computation enables real-time bidding optimization without requiring complex calculations at the moment of bidding, thus improving campaign performance while minimizing processing time.
Solution Approach 2:
The patent transforms continuous price and event rate parameters into discrete binned representations, changing the parameter space to enable faster computation. This parameter transformation allows for real-time optimization by working with simplified discrete distributions rather than continuous data.
Data Source
AI summary
A method for monitoring an advertising campaign is disclosed and includes receiving a campaign transaction signal, generating a discrete function based on the signal, and generating a continuous function based on the discrete function. The signal indicates a price and an event rate for impressions in the campaign. The discrete function indicates a discrete cumulative distribution of the impressions. The price dimension corresponds to an impression price independent variable. The event rate dimension corresponds to an event rate independent variable. The price dimension and the rate dimension are segmented into a plurality of bins based on threshold values. The continuous function indicates a continuous cumulative distribution of the impressions and is continuous in the price dimension and the rate dimension. The response function is also a continuous function. The response function is a function of control variable that parameterizes a correspondence between the price and event rate variables.