Measured vs. Perceived Web Performance

Abstract

We considered the measured vs. perceived web performance case study. The work aimed to find a relationship between the automatically measured times of loading elements on web pages with zoological articles and the feelings of users regarding these loading times. We found that the following automatically measured performance metrics, translate the best into the perceived performance evaluated by the users: (i) page_loading, measured by Selenium; (ii) first_meaningful_paint, measured by PageSpeed Insights; (iii) domInteractive measured by Browsertime; and (iv) contentEventLoad, measured by Browsertime. In research carried out with the use of Selenium; however, there is a problem of changes in the interface that make the measurement unable to be performed correctly. A possible solution to this problem could be to change the used CSS selectors from selectors picked from the website itself to selectors defined based on the analysis of the selected HTML code. Another solution would be to use tools that allow, for example, to detect all graphic files on the website and design navigation based on them.