Moving a 150 Tool Site Off Elementor: What Changed in PageSpeed
In November 2026 I started moving every page on convertnow.tools off Elementor and onto the plain WordPress block editor. Not the whole site in one weekend. A few pages per run, each checked by hand.
This is why I did it, how each move works, what broke, and what changed in PageSpeed.
Why leave Elementor
Elementor is quick for getting a page up. It is less good at staying light. Each page loaded its own stack of stylesheets and scripts, and a simple tool page carried far more code than the tool needed.
On a site with 150+ tools, that adds up, and PageSpeed agreed.
How each move works
Every tool page has the same parts: a short introduction, the tool, related tools, how it works, an FAQ and a row of related posts. For each page I:
- Rebuild every section as plain blocks, keeping headings and text exactly as they were
- Put the tool markup in one HTML block with its own scoped styles
- Keep the tool JavaScript as a separate file, loaded only on that page
- Replace the Elementor posts carousel with a plain query block for the latest posts in that category
- Compare section heights with the old page and test on a 390 pixel wide phone screen
Every page goes into a spreadsheet tracker, so I always know what is done and which tool scripts I still need to update.
Things that broke
Four problems came up more than once:
- WordPress turns straight quotes into curly ones, which breaks code samples inside a page
- Some JavaScript operators get mangled when the editor saves, so tool scripts avoid them
- Blocks that rely on preserved whitespace, like code output panes, collapse unless protected
- Elementor and Rank Math each printed FAQ schema, so some pages had it twice
On the API Response Diff Checker, the two comparison panes also grew wider than the screen with long lines. A one line grid change fixed it.
The numbers
Here is what PageSpeed Insights reported on mobile for the pages I moved, such as the JSON Formatter:
- Performance went from the 70s to 95 and above
- Accessibility went from 87 to 95 and above
- Agentic browsing, the newer check for how well AI agents can use a page, went from 2 or 3 out of 4 to a full 4 out of 4
The accessibility gain was a surprise. Elementor wraps every element in several layers of divs, and some of its widgets miss labels and heading order. Plain blocks produce clean, semantic HTML, so the same content scores better without any extra work. That same clean structure is what lets AI agents read and use the page.
You can see the JSON Formatter report for yourself.
Was it worth it
Elementor kept my layout saved, so any page can roll back with one click. That safety net made it easy to start. The rest of the site moves a few pages at a time, and every page gets the same before and after check.
The same week this migration went live, search impressions dropped sharply. I wrote up what actually caused it in how I fixed a traffic drop after a website redesign.
