Website Anomaly Detection Across Device, Browser, and Location
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems fail to accurately identify and alert website owners about performance anomalies, particularly those affecting specific attributes like device type or browser, leading to inefficiencies in diagnosing and resolving issues.
Innovation Solution
An alerting system that evaluates user interactions on a website, determines attribute values, and compares performance metrics against historical ranges to identify anomalies, providing alerts when metrics fall outside the evaluation range, thereby pinpointing specific issues.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing monitoring systems track overall website performance, then general performance data is collected, but they fail to accurately identify anomalies affecting specific attributes like device type or browser
Solution Approach 1:
The patent segments website performance monitoring by dividing it into multiple attribute dimensions including device type, browser, operating system, and location. Each attribute is analyzed separately to identify anomalies specific to that segment, enabling precise detection of issues affecting particular user groups without requiring overly complex monolithic monitoring.
Solution Approach 2:
The patent introduces multiple dimensional attributes (device type, browser, OS, location) to transform single-dimensional performance tracking into multi-dimensional analysis. This dimensional expansion allows the system to identify anomalies that would be invisible in aggregate metrics, improving detection accuracy while maintaining manageable system complexity through structured attribute-based organization.
2Measurement precision
If the alerting system analyzes multiple attribute values and historical data, then anomaly detection precision improves, but processing time and computational resources increase
Solution Approach 1:
The patent implements preliminary action by pre-processing and organizing historical performance data according to multiple attributes before anomalies occur. The system establishes baseline performance ranges for each attribute combination in advance, so when anomalies are detected, the system can quickly compare current metrics against pre-computed ranges rather than analyzing all historical data in real-time, significantly reducing processing time.
Solution Approach 2:
The patent applies local quality by focusing analysis on specific attribute segments and their relevant historical data rather than processing all data uniformly. Each attribute value (e.g., mobile devices, Chrome browser) has its own localized performance baseline and anomaly detection logic, enabling efficient targeted analysis that reduces overall computational burden while maintaining high precision.
3Loss of information
If the system provides detailed alerts with attribute information, then diagnostic capability improves, but information overload may occur
Solution Approach 1:
The patent segments diagnostic information into organized groups by attribute type (device, browser, OS, location) and presents them in a structured format. Each segment contains relevant performance metrics and anomaly details for that specific attribute, making comprehensive diagnostic information accessible without overwhelming the user with unorganized data.
Solution Approach 2:
The patent applies local quality by providing tailored alert information specific to each affected attribute segment. Instead of generic alerts, the system delivers customized diagnostic details relevant to the specific attribute experiencing anomalies (e.g., mobile-specific issues for mobile devices), enabling website owners to quickly identify and address the root cause without sifting through irrelevant information.
Data Source
AI summary
Techniques are described herein for evaluating user activities on a website for detecting and alerting anomalies with the website. For example, an alerting system may determine a set of time-windows for each of a set of attribute value(s) and use a set of performance metrics and/or web session events for the time-windows from historical web sessions to detect an anomaly in a current time-window. The alerting system may determine an evaluation range based on weighted performance ranges, and alert a client indicating a performance anomaly if a performance metric at the time of evaluation are out of the evaluation range. Attribute value(s) related to a website issue can be identified based on a correlated to anomalous web session events or to an anomalous time-window.