Web Performance Monitoring

Web Performance Monitoring is the measurement of a website’s or web service’s ability to respond efficiently to end-user interactions. Data gathered from monitoring helps to improve the service’s speed and ultimately increase user satisfaction resulting in higher user retention while reducing bounce rates and shopping cart abandonment.

Why use Web Performance Monitoring

Research shows that a site only has three seconds to load before losing 40 percent of its visitors ( source , and research by Google found that a half-second delay was enough to annoy users and turn 20 percent of a Google’s users away. To safeguard a brand’s reputation and revenue stream businesses conduct regular Web Performance Monitoring on their web assets to prevent the costly loss of site visitors.

What causes poor web performance?

Many things can cause a website or web service to perform poorly. File quantity, file sizes, system architecture, and user variables such as location, device, browser type, operating system, and connection speed all come together to form what could be a great or poor end-user experience. With dynamic distributed content, performance problems can manifest themselves at any point due to site changes or failing or inadequate hardware. Third party content often plays a role in poor performance, but identifying sluggish third party content can be difficult without Synthetic Web Performance Monitoring.

How does Web Performance Monitoring work?

Depending on the type of service offered, a company may choose between many different types of Web Performance Monitoring. Different solutions offer different levels of reporting granularity along with active ( Synthetic Monitoring ) and passive ( Real User Monitoring ) approaches.

Synthetic Performance Monitoring

For an “active” approach to performance monitoring, companies use “synthetic” performance monitoring. Synthetic Monitoring uses a computer application that conducts the performance checks on a routine basis. Because the testing does not require waiting for user interaction, the monitoring industry considers synthetic monitoring an “active” form of monitoring. Every minute or so, a checkpoint sends a request to the website or web service, measures the response times, and generates reports and alerts based on the site’s performance during the test. A site may choose basic to advanced performance monitoring solutions based on what needs monitoring.

Basic Web Performance Monitoring

For brochure sites and other non-critical pages, companies may choose the most basic form of performance monitoring. Basic Website Monitoring (also called HTTP and HTTPS monitors or uptime monitoring) uses checkpoints that send a request for the page and records the time it takes to download the page. A basic monitor does not process the returned content, so any subsequent requests never happen. For example, basic web monitors do not download images, so speed issues due to image file sizes will go unrecorded. This very basic performance indicator can capture some back-end issues, track uptime (excellent for SLA tracking), and help identify performance trends but not much more in respect to performance.

Mid-level Web Performance Monitoring

Mid-level performance monitoring reduces the uncertainty of basic monitoring by using a real browser like Chrome to make the request, execute scripts, request and download additional content, and load the page. The returned content loads into the browser just as it would for an actual user. This level of performance monitoring gives a more accurate account of the sites response and load times, but it does not provide the element-by-element performance results of Advanced Web Performance Monitoring. Because Real Browser Monitoring loads the page content, if the page errors, the monitor generates a screenshot of the resulting page when possible.

Advanced Web Performance Monitoring

Advanced web performance monitoring provides the most granular performance data obtainable from Synthetic Monitoring. The request initiates from a real browser just like mid-level monitoring and measures the resolve, TCP connection, HTTPS handshake, send, wait, and receive times for each element on the page. Displaying the data in a waterfall chart provides a visual account of the page load progression. With advanced web performance monitoring, a website or web service can discover bottlenecks, failing content, slow third party content, infrastructure issues, and generate error screenshots.

The importance of a large checkpoint network in synthetic Web Performance Monitoring

Latency due to an end-user’s location is a serious problem for many websites and services. For a synthetic solution to work as a true indicator of user experience, the tests must originate from a location near the site’s actual users. Using a service that provides a large network of checkpoints allows a site to designate where the checks originate. This provides data that better resembles the site’s actual user experience and helps identify location specific issues.

Real User Monitoring and Performance

Real User Monitoring (RUM) is a passive approach to performance monitoring. The monitoring industry considers Real User Monitoring “passive” because the monitoring only occurs when a user accesses a page running a RUM script. The RUM script gathers detailed information about the user’s environment and reports the performance as experienced by the actual user. Real User Monitoring reports on:

Load time

• Time to first byte
• Page ready time

Network duration

• Redirect duration
• DNS duration
• Connect duration

Backend duration

• Send duration
• Receive duration

Frontend duration

• DOM duration
• Render duration

Download time

The performance data sorting options include page viewed, user location, browser type, operating system, and device type. Real User Monitoring gives detailed insights into a site’s performance when the site is available, but its passive nature makes it an unreliable tool for uptime reporting and other alerting needs. Synthetic monitoring is better suited to alerting a company to uptime and performance issues, while Real User Monitoring provides insights on how to structure the synthetic options and to get real insights into the site’s actual users’ experiences.

Conclusion

After uptime monitoring, performance monitoring may be the single most important type of monitoring a company can conduct. Synthetic Performance Monitoring and Real User Monitoring offer active and passive monitoring allowing a brand to monitor their performance while seeing exactly how end users experience the site. The dual approach protects reputation and revenue.