Blog

How Hilenium Communicates to Prospective Clients If They’re Fast & Found

Fast & Found by the Numbers

When we first engage with a client it’s important to demonstrate the current state of their websites. Once documented it gives us a baseline of data from which we can propose a range of tasks to upgrade their Fast & Found performance to Hilenium standards. 

For clients who have only one website or require more comparative information we can also produce the data for their competitors. In the table below we have mocked up the results of Fast & Found analysis to indicate the general ranges of data returned.

Website Load Speed

To generate our website load information we utilise the Google Chrome web browser. Whilst Chrome was relativley late to the browser wars, appearing well after Firefox(Netscape), Internet Explorer, Safari and Opera it is now the most popular browser worldwide with around 55% market share. More importantly the Chrome browser posses an excellent debug environment with its Chrome Inspector. With launch of Chrome developer tools protocols we can automatically deploy instances of Chrome and programmatically run our Fast & Found test. Its a growing ecosystem with great tooling.

HTML Load

To determine the base responsiveness of the web server we record the base HTML Document load. This is the web page source you can see if you inspect the code in a browser (ignoring any inclusions or modifications performed by javascript). This basic test is typical of what is is used by server-side load testing libraries such as Apache AB  HTTP server bench-marking tool and by load test software as a service providers like loader.io.  Load times for  the HTML Doc give us an indication of how well the website and web server are configured to respond to the initial request.

HTML Load time is a good basic performance indicator without being obscured with all the requests to load a web page. The aforementioned programs Apache AB and Loader take things further by testing maximum concurrent users to determine the web servers capacity to handle large volumes of traffic. We avoid this on initial consultation as these tools load tests the website until it is unresponsive – it can also cause an poorly configured server to fall over. Intentionally crashing a prospective clients website is not something we want to do!   

Total Load

To determine the effective time a website takes to load there are several events to consider:

• DOMContentLoaded: Indicates when the browser has finished parsing the document. Other resources such as images and stylesheets may not have finished being downloaded at this stage.
• Load: Indicates when all the initial resources including images, stylesheets, JavaScript have been downloaded.
• Finnish: Indicates when all resources including asynchronously, non-blocking, loading has completed. 

For our snapshot we use the load event as it best approximates the performance of a correctly configured website and server. In a web applications with complex asynchronous calls or DOM rendering other events would also need to be considered. for the majority of websites the load event will be fine.

Fast Metrics

In the Fast Metrics section we focus on analysis of the content and code of a website to detect if there is anything that may hinder performance. The code they refer to is front-end only as it needs to be accessible by the browser. Its important to note fast metrics is not concerned with load time specifically but with technical factors that effect load and fast user experience criteria. We have experienced sites with great Fast Metrics that are slow to load and conversely, sites with poor Fast Metrics that are blazingly fast. Generally though, the two go hand-in-hand.

Google PageSpeed

PageSpeed is a series of tools released by Google. We will be generally referring to PageSpeed Insights in this article. PageSpeed requests a copy of the website from the web server and then checks to see if a page has applied common performance best practices. After analysis a score is provided ranging from 0 to 100. As PageSpeed is a performance optimisation tool, its recommendations are biased in favour of speed and are continually updated to measure new performance techniques. Hence, a site’s PageSpeed score may vary over time. 

PageSpeed Insights measures performance of:

• Elapsed time for above-the-fold load i.e. browser rendering of content initially visible to user.
• Elapsed time for whole page load i.e. browser rendering of the whole page whether visible to user or not (below the fold).

Yahoo YSlow

YSlow was an internal project at Yahoo aimed at helping website developers improve the performance of their websites. It grades a web pages based on pre-defined rule-sets. The Exceptional Performance Team at Yahoo identified 34 performance rules and YSlow analysis includes 23 of these.

Although the YSlow project appears to have less momentum than PageSpeed these days, we feel a second opinion never hurts.

SEO Analysis

In our SEO analysis we scan the website as rendered in the browser to determine issues that will adversely affect search engine rankings. These are broken into three types with decreasing Severity; Errors, Warnings & Issues

SEO Issues

The SEO Issues tally presented in the snapshot table is the total number of all SEO issues. Hence, the SEO Issues tally includes both SEO Errors and SEO Warnings in addition to lower importance issues. 

SEO Errors

SEO Errors tally is measure of the most serious issues include:

• Broken external links
• Broken external images
• Pages with duplicate Meta descriptions
• Pages returning a 5XX or 4XX HTTP status codes

SEO Warnings

SEO Warnings are issues that should be fixed but are not the most urgent and include:

• Images are missing an alt attribute
• Page are missing an H1 tag
• Page has multiple H1 tags
• Page is missing a meta description
• Missing sitemap XML reference in Robots file

Other SEO Issues

Low priority issues in the tally include:

• Missing HTTP encryptiom
• Too long a page
• Blocked crawl by robots.txt
• Missing www CNAME
• Broken internal links
• Duplicate content
• Page have no viewport tag
• Duplicate titles
• Internal links with nofollow attributes
• Underscores in URLs
• Flash Player usage
• No DocType declaration
• No language declaration
• Temporary redirects
• No encoding declaration
• Too long or short page title
• Too many query parameters in URLs
• Use of iframes
• Too many on-page links
• Pages with few words
• Low text to HTML ratio
• Too long URLs
• No robots.txt file

Summary

As Hilenium’s focus is on Fast & Found we give a simple status for each that helps prioritise where optimisation should be focused. 

Fast Status

After considering website load times and the Fast Metrics we then award a general Fast Status to the website:

• Fast: The website is performing well only minor tweaks may be required.
• Fix: The website needs a reasonable amount of work.
• Urgent: The website is performing poorly and need much work as soon as possible.

Found Status

After performing an overview of the nature for the SEO isues, errors and warnings we award a general Found Status to each website:

• Found:  The website is optimised well and only minor tweaks may be required.
• Fix:  The website needs a reasonable amount of work.
• Urgent:  The website is performing poorly and need much work as soon as possible.

We feel the Hilenium Fast & Found Snapshot delivers just the right amount of technical data with two simple statuses for each website.