Silktide does not see a webpage like most software. Instead, Silktide sees pages almost exactly as real users do – in a real web browser. This means we can see a tremendous range of content and detail that most other software cannot.
Silktide versus traditional bots
Silktide | Traditional bot |
Pages are tested on both mobile and desktop devices. | Simulation of a device is impossible, no distinction can be made between mobile or desktop. |
Can interact with pages just like a person would, e.g. logging in, performing searches, and testing these features work. | Interaction is not possible. |
Can test almost every part of any website, including areas behind JavaScript or AJAX. | Completely unable to test certain web pages. |
Full pages are loaded just like a real person, including images, scripts, fonts, and styles. Accurate testing of all parts of a page is possible. For example, we can assess font size, mobile layout, colour contrast and more. | Only the code for the page is loaded. Only testing of that code is possible. |
Able to measure how long pages take to become interactive (usable) in real browsers. | Only able to test how long page code takes to be transmitted. |
An analogy
Imagine you were trying to measure the intelligence of a person.
One way to do this would be to measure the physical properties of their brain, like its cells, size and weight.
A better way would be to talk to the person and assess their behaviour.
This is analogous to what website analysis tools do. Most traditional tools can only see the raw code (in this analogy, their brain). Silktide can experience the result of that code (talking to them).
The difference is between assessing the parts, and assessing what they do.
As you might imagine, there’s a tremendous amount of information you could only measure by looking at the end result. As webpages become more and more dynamic – with JavaScript, AJAX, etc. – this becomes more and more necessary.
For example, the HTML code for a page is usually the same, whether it is viewed on a desktop or a mobile device. But the end result can be completely different, and that same code may not even work on mobile. Looking at just the code can never tell you this, which is why we look at the end result.
Technical details
Silktide loads each webpage in the Chrome browser, which allows us to simulate a wide range of devices accurately. Our standard devices are as follows:
Desktop – A Windows desktop PC using Chrome
Mobile – A portrait iPhone 5 using Safari (this is intentionally old, so as to cover the smallest mobile screen size, a requirement of WCAG 2.1)
We load each page as a normal user would, executing any JavaScript the page requires. Where requested, we run custom interactions with each page as we go – for example, we might close any cookie law notifications, or check to confirm our age before entering a website.
We gather a tremendous amount of data from each page for testing, including:
A screenshot of the page, as it appeared on our device.
The layout of the page, e.g., what areas appear off the side, or below the fold.
List of all resources (images, stylesheets, etc.) and redirections used, including their timings.
The interactive state of the page over time, e.g., when could a page be used, when was it finished.
Any JavaScript or network errors.
The computed HTML, i.e., the final HTML of the page after scripts have finished modifying it. (This allows us to test areas only accessible via JavaScript).
The font size, typeface, colour and contrast of all text areas of the page.
And much more.
How devices are simulated
The following areas are covered when impersonating a given device (e.g., an iPhone):
The pixel and logical pixel resolution
The user agent
Explicit and calculated viewports (e.g. zoom to fit)
Device capabilities (e.g., Flash is not possible on mobile)
Bandwidth (e.g., 3G or 4G connectivity for mobile)
Most of Silktide’s analysis is conducted on the resulting computed source.