Consumer Performance Index Scoring for Web Applications
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for assessing user engagement and performance of websites and web applications in cloud computing lack comprehensive real-time analysis and benchmarking, particularly in correlating landing page load times with bounce rates across diverse user experiences.
Innovation Solution
A cloud-based system utilizing real user measurement (RUM) data to calculate a consumer performance index (CPI) score, which aggregates and analyzes beacon data from multiple domains to create benchmarks that correlate landing page load times with bounce rates, employing logistic regression models to estimate user tolerance and generate CPI scores.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If real-time user measurement data is collected and analyzed to create comprehensive performance benchmarks, then measurement precision and analytical depth are improved, but system complexity and computational requirements increase
Solution Approach 1:
The system segments the complex analysis task by separating data collection, data processing, and benchmark generation into distinct modules. The beacon collector gathers raw RUM data independently, the statistical processor handles calculations separately, and the benchmark generator creates comparisons autonomously, reducing overall system complexity while maintaining precision.
Solution Approach 2:
The patent introduces intermediary components such as the statistical processor and data aggregator that mediate between raw beacon data and final benchmarks. These intermediaries simplify the relationship between complex data collection and benchmark generation, allowing each component to focus on specific tasks without increasing overall system complexity.
2Loss of information
If comprehensive real-time analysis of user experience data is performed across multiple domains, then the quality of performance insights is improved, but processing time and computational resources increase
Solution Approach 1:
The system performs preliminary actions by pre-processing beacon data in real-time as it is collected, immediately calculating statistical metrics and preparing data for benchmark comparison. This preliminary processing reduces the time required for subsequent comprehensive analysis and allows quick generation of performance insights without delaying information completeness.
Solution Approach 2:
The patent implements continuous processing where beacon data is constantly collected, processed, and compared against benchmarks in real-time. This continuous action ensures that information remains complete and up-to-date while avoiding batch processing delays, maintaining both information completeness and processing efficiency through ongoing operations.
3Loss of information
If statistical models are used to correlate load times with bounce rates, then analytical depth and insight quality are improved, but model complexity and computational requirements increase
Solution Approach 1:
The patent applies parameter changes by transforming raw beacon data into standardized statistical parameters and metrics that can be directly correlated with user behavior. By converting complex user experience data into manageable parameters like load time, bounce rate, and engagement metrics, the system reduces model complexity while maintaining deep analytical insights into user behavior patterns.
Data Source
AI summary
A computer implemented method creating a benchmark for consumer satisfaction of website or web application performance includes aggregating real user measurement (RUM) data captured from devices of users during sessions on a website or web application, the RUM data including a landing page load time of each user and whether a bounce occurred. The probability that each tenant will bounce is then modeled. Next, an ordered discrete set of lethal dose (LD) values is calculated and the tenants are then filtered for erroneousness. An ordered set of mean LD values is calculated, and that set is reduced to exactly 10m ordered mean LD values. The 10m ordered mean LD values are mapped to a corresponding number of ordered CPI scores ranging from 0 to 100, wherein the mapping results in the benchmark.