JultikaUniversity of Oulu repository

Abstract

Fast website loading speeds can increase conversion rates and search engine rankings as well as encourage users to explore the site further, among other positive things. The purpose of the study was to find and compare free-to-use tools that can both evaluate the performance (loading and rendering speed) of a website and give suggestions how the performance could be improved. In addition, three tools were used to evaluate the performance of an existing WordPress site. Some of the performance improvement suggestions given by the tools were then acted upon, and the performance of the website was re-evaluated using the same tools. The research method used in the study was experimental research, and the research question was “How to evaluate and improve web performance using free-to-use tools?” There were also five sub-questions, of which the first two related to the tools and their features, and the last three to the case website.

Eight free-to-use web performance evaluation tools were compared focusing on what performance metrics they evaluate, what performance improvement suggestions they can give, and six other features that can be useful to know in practice. In alphabetical order, the tools were: GTmetrix, Lighthouse, PageSpeed Insights, Pingdom Tools, Test My Site, WebPageTest, Website Speed Test (by Dotcom-Tools) and Website Speed Test (by Uptrends). The amounts of metrics evaluated by the tools ranged from one to fifteen. The performance improvement suggestions given by the tools could be put into three categories, meaning that the suggestions largely overlapped between the tools. All tools except Lighthouse were web-based tools.

The performance of the case website was evaluated using GTmetrix, PageSpeed Insights and WebPageTest. On desktop, the performance was in the high-end range though varying between the three tools, and on mobile, the performance was noticeably slower due to the challenges of mobile devices (e.g. lower processing power compared to desktop computers) and mobile networks (e.g. higher latency compared to broadband connections). The common bottlenecks based on the suggestions given by the three tools seemed to be lack of using a CDN (Content Delivery Network), serving unoptimized images and serving large amounts of JavaScript. The results of the performance re-evaluation were mixed, highlighting the importance of carefully considering each performance improvement suggestion.

The main takeaways of the study for practitioners are to use multiple tools to get a wide variety of performance metrics and suggestions, and to regard the suggestions and relative performance scores given by the tools only as guidelines with the main goal being to improve the time-based performance metrics.