Overlapping Experiments Layered Traffic Diversion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for optimizing website conversion rates through experiments are limited by the need for disjoint experiments, which restricts the number of tests that can be run and hinders the analysis of interactions between parameters, leading to inefficiencies in testing and optimization.
Innovation Solution
The implementation of overlapping experiments, where multiple experiments can be performed on the same user queries across different layers, allowing for independent diversion methods such as cookie identifiers, random traffic, and user identifiers, enabling the analysis of interactions between parameters and increasing the number of experiments that can be run.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If conventional A/B testing or multivariate testing is used with disjoint experiments, then each experiment can be conducted independently, but the total number of experiments that can be run is limited and interactions between parameters cannot be analyzed
Solution Approach 1:
The patent segments experiments into multiple independent layers, where each layer can conduct experiments independently. This segmentation allows numerous experiments to run simultaneously across different layers without interfering with each other, thereby increasing the total number of experiments that can be conducted while maintaining manageable complexity through modular organization.
Solution Approach 2:
The patent introduces a vertical dimension by organizing experiments into multiple layers stacked on top of each other. Instead of running experiments sequentially or in parallel at the same level, the system adds a layer dimension, allowing experiments to coexist at different heights in the experiment space. This dimensional transformation enables many more experiments to run simultaneously without increasing horizontal complexity.
2Adaptability or versatility
If multiple experiments are run on the same user queries, then the number of experiments increases and parameter interactions can be analyzed, but the diversion and management of traffic becomes more complex
Solution Approach 1:
The patent segments the traffic diversion process into layer-specific diversion mechanisms. Each layer independently diverts traffic according to its own experiment requirements, rather than having a single complex diversion system. This segmentation simplifies traffic management while enabling multiple experiments to run on the same user queries across different layers.
Solution Approach 2:
The patent applies partial action by allowing only certain layers to divert traffic for specific experiments, rather than diverting all traffic to all experiments. This selective approach enables parameter interaction analysis where needed while maintaining simpler diversion rules for other layers, balancing versatility with manageability.
3Ease of operation
If experiments are organized in a nested structure of domains and layers, then experiments can be efficiently managed and gradual launches are enabled, but the system complexity increases
Solution Approach 1:
The patent implements a nested structure where layers are contained within domains, and experiments are contained within layers. This nesting organization allows efficient management by providing clear hierarchical boundaries and containment relationships. The nested structure enables gradual launches by allowing control at multiple levels (domain level, layer level, experiment level) while organizing complexity in a manageable hierarchical fashion.
Solution Approach 2:
The patent introduces dynamic control mechanisms that allow the nested structure to adapt to different experiment needs. The system can dynamically adjust which layers are active, which domains are enabled, and how traffic is distributed across the nested hierarchy. This dynamic capability improves operational efficiency while the modular nested design keeps the underlying structure manageable.
Data Source
AI summary
User queries are received, with each query requesting a service from a server. Overlapping experiments are performed on at least a portion of the queries, with each experiment modifying one or more parameters associated with the queries or parameters associated with processing of the queries, and with the experiments organized into layers. Two or more experiments in different layers are allowed to be performed on the same query, and for any given layer, at most one experiment is allowed to be performed on the same query.