SquishPixel Blog · Web performance · Core Web Vitals · Tools

How to compare CMS speed fairly using public PageSpeed and tech stack data

A DEV Community post explains how to test whether a CMS (content management system) really affects speed in your own niche. The method is to pick a sample of sites, detect what each is built with, audit each home page with PageSpeed Insights, and compare the groups. The author says collecting the data is quick and the hard part is not over-reading it.

The author lists several traps. Hosting can matter more than the platform, and server response time (TTFB, the time before a server starts sending a page) may explain more than the CMS. Large images slow any platform. Bigger companies tend to have custom builds, more tags and better engineers, so a platform gap may be a budget gap. Lighthouse runs on the same page can differ by a few points, and tiny groups tell you little.

For sampling, the article advises choosing one niche from a list that does not depend on speed, such as a trade association directory. It warns against using the first page of Google results, because those sites already rank partly on page experience. It suggests 50 or more sites and auditing only the home page, so every page is the same type.

The post uses two open-source Apify Actors that the author built: a tech stack detector and a PageSpeed Insights audit tool. Audits should be run three times on different days so each site gets a median. A script then joins the data and prints median scores with group size, TTFB and image savings beside them. The author gives a cost example of $1.20 for 60 sites and notes limits: detection reads only served HTML, so tags loaded later through Google Tag Manager are missed, and some sites block scripts.

Why it matters

Platform debates often rest on one memorable bad site. This article shows how to replace that with a small, honest comparison, and how easily the numbers mislead. A slow group of sites may reflect hosting, heavy images or company size rather than the CMS itself.

If you are deciding on a rebuild or blaming marketing tags for a slow site, the same caution applies. Check server response time and image weight first, treat any gap as correlation, and test a cause by changing one thing at a time, such as removing tags on a staging copy of one page and auditing it again.

Source: DEV Community. This brief summarises their reporting; read the original for the full detail.