Why Your Elementor or Divi Site Is Slow: Measured, Not Guessed

Owners come to us with the same sentence: “my site got slow and I think it is Elementor.” Sometimes it is. Often the builder is one cause among four. This article is how we measure a slow WordPress site that was built with Elementor or Divi, and the order in which we fix it. It sits in our series on keeping a site alive, next to the pillar on what a WordPress maintenance plan should include, because speed is one of the things a plan watches.

Measure first

Open PageSpeed Insights, enter your home page and one inner page, and read two sections separately.

Field data comes from real Chrome users, covers the previous 28 days, and is reported at the 75th percentile. Lab data comes from Lighthouse running one simulated load on a fixed device and network: a mid-tier phone on a mobile connection for the mobile test. Google describes lab data as useful for debugging and notes that lab scores can change from run to run, and that field data is a historical record of many devices and connections. If your site does not have enough visits for field data on a page, PageSpeed Insights falls back to the whole origin, and with too little data there it shows none.

Read the field data first, because that is what your visitors had. Then use the lab run to find out why.

The three numbers

Google’s Core Web Vitals “good” thresholds, at the 75th percentile of page loads, are¹:

Write the three field numbers for the home page and for your top inner page on paper. That is the baseline everything below is compared with.

Where the time goes

Take LCP, since it is the metric small-business sites fail most visibly. Google’s guidance splits it into four parts and gives rough shares of a good score³. Applied to the 2.5-second target, they look like this³:

The shares are guidelines, not rules, the guidance says. The seconds in the third column are our arithmetic from those shares. The useful point is where the time can be lost: the server answers late, the image is discovered late, or the image is heavy.

What the builder adds

A page builder wraps your content in more markup, more styles and more scripts than a theme written for one site would. Two things in Google’s documentation explain why that matters. Large pages cost more to render, and “large DOMs do require more work to render than small DOMs,” which affects both the first render and the response to a tap. And scripts that load, parse and run on the main thread can create long tasks that delay the browser’s response to a visitor.

The builders’ own documentation shows they know this. Elementor lists performance settings on its help page, last updated June 19, 2026:

• Optimized Image Loading applies a high fetch priority to the LCP image and lazy loading to images below the fold.
• Optimized Gutenberg Loading removes unused block-editor scripts and styles.
• Lazy Load Background Images lazy-loads every background image except the first.
• Inline Font Icons draws icons as inline SVG, so the icon fonts and their styles do not load.
• Element Caching and Optimized Control Loading are listed as experimental features.

Check whether the stable ones are switched on.

Elegant Themes makes a speed claim for its newer builder. It says Divi 5 is “two to four times faster than Divi 4,” with server response times down 41% on a simple layout, 51% on a larger one and 74% on a very long landing page⁴. Read the conditions: the test was run in an uncached environment while Divi 5 was in alpha, the article gives no sample sizes or repeat runs, and it comes from the company that sells the product⁴. It also notes that on the longest page about 90% of the weight was images, so the builder’s code was not the main cost⁴. A site that is slow because of its photos will not be fixed by a builder upgrade. Whether to move is a separate decision, covered in Divi 5: should an existing site migrate.

What to do, in order

The order below goes from the cheapest and most common causes to the most expensive.

1. Fix the images. Compare the size an image is displayed at with the size of the file. Resize and compress before upload, serve a next-generation format where the builder or a plugin offers it, and do not lazy-load the main image: Google’s guidance says never to lazy-load your LCP image. Elementor’s image setting does this automatically. If LCP is poor, this is the first place to look, because the image is both the thing measured and the part that most often weighs too much.
2. Turn on caching and check the server’s answer. A cached page is served without rebuilding it on every visit. Look at the time to first byte in the lab report: Google’s guidance warns that a high one can make a 2.5-second LCP “challenging, or even impossible,” and names redirects, distant visitors and query strings that defeat caching among the causes³. A slow answer on every page, caching on, points at the host or its plan, not the builder.
3. Switch on the builder’s own performance settings. For Elementor, the list above. For Divi, look in your version’s settings for its performance options, and read Elegant Themes’ own notes for your version before a redesign.
4. Count the scripts. Open the lab report’s list of third-party and plugin scripts. Each one that loads on every page adds main-thread work, which is what drives a poor INP. Remove what you do not use: the old slider, the second analytics tag, the chat widget nobody answers.
5. Re-measure. Lab results show at once; field data takes a 28-day window to move². Do not judge a change from a single test.

When a rebuild is the honest answer

Sometimes the measurements say the same thing after every fix: the markup is heavy on every page, the plugins are load-bearing, and the lab score for an empty page is already poor. At that point more tuning costs more than a lean site that does the same job. A rebuild on a theme with no builder keeps the content, the address and the redirects, and replaces the layers. It is a decision with a price, so ask for the measurement that justifies it in the proposal, not a promise.

What we do about it

We do not use a page builder on the sites we build. The themes are written for each client, which keeps the markup small and the number of plugins low, and updates go on a staging copy first. On a site built with a page builder, our method is the one in this article: measure with PageSpeed Insights, write down the field numbers, fix the first four items, and re-measure. A rebuild comes only after those have been tried, with the before and after numbers on the table. This is part of what we call optimization. It is also one reason the sites we build stay in service for years.

When to get help

Get help if the field data is in the “poor” band on the pages that bring you calls, if the site feels slow to edit as well as to visit, or if a previous round of tuning made no difference. Do not pay for speed work on a site that already passes all three metrics at the 75th percentile; a faster score will not bring a customer. The consultation is free, and we will run the measurement with you on the call.

Frequently Asked Questions

What is a good PageSpeed score for a small-business site?

Google does not set a target for the Lighthouse score; it sets thresholds for the three Core Web Vitals: LCP 2.5 seconds or less, INP 200 milliseconds or less, and CLS 0.1 or less, at the 75th percentile of real visits¹. Look at those numbers in the field-data section.

Why is my PageSpeed score different every time I test?

The lab score comes from one simulated load, and Google says lab scores can vary from run to run. Run it three times, and trust the field data more².

Is Elementor slower than Divi?

We cannot say in general. Each vendor describes its own performance features, and the answer depends on the hosting, the content and the plugins. Measure your own site.

Will upgrading to Divi 5 make my site faster?

Elegant Themes reports faster server responses in its own uncached alpha-stage tests. On a site whose weight is mostly images, the builder’s share is small. Measure before and after on a staging copy.

Does my hosting matter more than my page builder?

For the first part of the load it can. Google’s guidance puts about 40% of a good LCP on the server’s first response and says a high time to first byte can make 2.5 seconds hard or impossible³. Test the server’s response before you change the builder.

Sources