Ad Placement Optimization Using Type-Based Slot Association
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional online systems fail to effectively account for advertisement size and content variations when placing ads, leading to suboptimal interactions between different advertisements in multiple slots.
Innovation Solution
The online system identifies candidate ads based on targeting criteria and determines optimal placement by associating each ad with a type, using rules and discount factors to maximize the combined value of ad placements across slots, considering factors like bid amounts, click-through rates, and ad characteristics.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If conventional methods place advertisements based only on bid amounts without considering advertisement size or content type, then the placement process is simple and fast, but the total value and user interaction across multiple advertisement slots is suboptimal
Solution Approach 1:
The system changes the parameters considered in advertisement placement from only bid amounts to include advertisement size, content type, and position discount factors. This allows the system to optimize total value across multiple slots by selecting advertisements that maximize overall effectiveness rather than simply following bid hierarchy.
Solution Approach 2:
The system applies different placement strategies to different advertisement slots based on their positions and characteristics. By assigning position discount factors to different slots and considering ad size and content type specific to each slot, the system optimizes each local placement decision to contribute to overall system value.
2Ease of manufacture
If the system considers advertisement size and content type in placement decisions, then total value and user interaction improve, but the complexity of the placement algorithm increases
Solution Approach 1:
The system performs preliminary classification of advertisements into types based on size and content characteristics before the actual placement decision. Position discount factors are pre-calculated for different slots. This preliminary organization simplifies the subsequent optimization process by reducing the search space and providing structured input for the placement algorithm.
Solution Approach 2:
The system segments the advertisement selection process into distinct steps: identifying candidate ads, classifying ad types, calculating position discount factors, and selecting final placements. This segmentation breaks down the complex optimization problem into manageable components that can be processed systematically.
3Ease of operation
If advertisements of different sizes are placed without considering their impact on other slots, then the placement process is straightforward, but the combined value of multiple advertisement slots is reduced
Solution Approach 1:
The system incorporates feedback from the placement decision on one slot into subsequent placement decisions for other slots. By calculating position discount factors that reflect the impact of each advertisement on overall slot value, and by considering ad size and content type interactions, the system adjusts placements to maximize combined value across all slots rather than treating each slot independently.
Data Source
AI summary
An online system selects advertisements for presentation in various advertisement slots to maximize the total value to the online system for advertisement presentation. Candidate advertisements for presentation to a user are identified and types of advertisements are associated with various advertisement slots. For example, types of advertisements are associated with advertisement slots based on values for presenting various types of advertisements in different advertisement slots and one or more ad placement rules. Candidate advertisements having an advertisement type associated with each advertisement slot are identified, and an expected value of various placements of the candidate advertisements having a type corresponding to advertisement slots is determined. Based on the expected values, a placement of candidate advertisements in advertisement slots is selected and presented to a user.