Ad Valuation via Randomized Performance Estimates
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In online advertising, determining the expected value of advertisements, especially new ones, is challenging due to limited data, leading to potential revenue losses for publishers and ad networks, as they struggle to select profitable ads amidst scarce information and inflated estimates from advertisers.
Innovation Solution
Implementing a cost-per-transaction (CPT) pricing model with strategies such as randomized conversion estimates, data mining for new advertisements, and transitioning between pricing models like CPM and CPT to ensure accurate valuation and maximize revenue, including artificial boosting of new ads to compete with mature ones.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional pricing models (CPM, CPC, CPA) are used to value advertisements, then advertisers can pay based on impressions, clicks, or actions, but publishers cannot accurately determine the expected value of new advertisements due to limited performance data
Solution Approach 1:
The system performs preliminary actions by collecting and analyzing various data sources (advertiser attributes, industry data, seasonal factors, etc.) before the advertisement is published. This allows the formation of a probability distribution for expected conversion rates even before actual performance data is available, enabling publishers to make informed decisions about new advertisements.
Solution Approach 2:
The patent introduces an intermediary mechanism - a probability distribution model - that mediates between the lack of direct performance data and the need for accurate value determination. This intermediary uses multiple data sources and statistical methods to bridge the information gap, allowing accurate valuation without requiring extensive historical performance data.
2Ease of operation
If publishers rely on advertiser-provided conversion estimates, then the valuation process is simplified, but the estimates may be inflated and lead to revenue losses
Solution Approach 1:
The system incorporates feedback mechanisms by continuously monitoring actual advertisement performance and using this information to refine the probability distribution models. This feedback loop allows the system to learn from real performance data while maintaining the simplicity of the valuation process, gradually improving reliability as more data becomes available.
Solution Approach 2:
The patent dynamically changes parameters in the valuation model based on available data. For new advertisements with limited data, the system uses broader parameters (advertiser attributes, industry averages). As performance data accumulates, the model transitions to using more specific performance-based parameters, thereby improving reliability while maintaining operational simplicity.
3Reliability
If publishers are conservative in selecting new advertisements, then revenue losses from poor performers are minimized, but profitable new advertisements may be missed
Solution Approach 1:
The system applies partial action by allowing publishers to selectively publish new advertisements based on their calculated expected values. Rather than being completely conservative or completely aggressive, the system enables a balanced approach where advertisements with sufficiently high expected values (above a threshold) are published, achieving moderate risk-taking that captures profitable opportunities while avoiding obvious poor performers.
Solution Approach 2:
The patent enables dynamic adjustment of risk parameters in the valuation model. Publishers can modify parameters such as the confidence level required for publication or the minimum expected value threshold. This allows the system to adapt to different risk appetites and market conditions, balancing reliability and productivity based on specific needs.
4Productivity
If more screen real estate is allocated to advertisements, then revenue opportunities increase, but the difficulty of selecting profitable ads among more options increases
Solution Approach 1:
The system extracts the essential valuation factors from complex ad selection considerations, focusing on the most critical parameters (expected conversion rate, bid amount, probability distribution confidence). By extracting and prioritizing these key factors, the system simplifies the selection process while still enabling effective utilization of extensive ad inventory and screen real estate.
Solution Approach 2:
The patent transforms the complex multi-dimensional ad selection problem into a more manageable form by changing the parameters used for evaluation. Instead of considering numerous factors simultaneously, the system converts them into a unified expected value metric that combines conversion probability, bid amount, and confidence levels, thereby simplifying the selection process while maximizing revenue potential.
Data Source
AI summary
Strategies are described for conducting an advertising campaign using a cost-per-transaction (CPT) pricing model. In this model, the advertiser is charged when an end-user takes some express action in response to viewing the advertisement, such as by clicking on the advertisement, purchasing the advertised asset, performing a registration procedure, and so forth. Various solutions allow for the computation of the expected value of a CPT advertisement when there is a scarcity of data pertaining to the actual performance of the CPT advertisement.