If you’ve ever wondered why two WordPress sites built on the same hosting plan can have completely different speed scores, the answer is usually sitting right in the theme and page builder.
Page builders are the single biggest variable in WordPress performance. Not your image sizes. Not your caching plugin. The builder.
This guide compares the most widely used WordPress page builders on the metrics that matter most: Lighthouse performance scores, HTTP requests, total page weight, and real-world consistency. Every number cited here comes from published, independent tests you can verify yourself.
The Sources Behind This Data
Two primary studies inform this comparison:
What We’re Comparing
Six builders covered here:
- Elementor – the most widely installed page builder in the world
- Divi – the most popular all-in-one theme + builder combo
- WP Bakery (formerly Visual Composer) – bundled with thousands of premium themes
- Beaver Builder – a developer-favored builder known for reliability
- Kadence Blocks – a Gutenberg-native block builder with global styling
- GenerateBlocks – an ultra-minimal block builder designed for zero overhead
What the Scores Mean
Lighthouse Performance Score (0-100) – Google’s open-source audit tool. Scores above 90 are “Good.” 50-89 is “Needs Improvement.” Below 50 is “Poor.” Google uses this as a proxy for Core Web Vitals, which are a confirmed ranking signal. Mobile scores matter most – Google indexes mobile first.
HTTP Requests – Every file the browser fetches before the page appears. More requests = more round trips = slower load.
Page Weight (KB) – Total download size. Critical on mobile connections.
The Real-World Data: How Each Builder Actually Performs
The Admin Bar Study – 150+ Live Agency Sites
This is the most credible dataset available because it uses real production sites, not demo pages. Kyle Van Deusen tested over 150 agency websites from The Admin Bar’s directory and recorded their Lighthouse performance scores alongside which builder was used.
| Builder | Sites Tested | Median Score | Mean Score | Min | Max |
|---|---|---|---|---|---|
| GenerateBlocks | 14 | 90 | 88 | 60 | 100 |
| Oxygen | 5 | 82 | 78 | 56 | 87 |
| Bricks | 20 | 77 | 76 | 38 | 97 |
| Beaver Builder | 16 | 70 | 69 | 55 | 91 |
| Gutenberg | 6 | 69 | 70 | 56 | 89 |
| Kadence | 11 | 76 | 69 | 34 | 82 |
| Elementor | 30 | 66 | 67 | 49 | 88 |
| Divi | 25 | 62 | 64 | 50 | 98 |
Source: theadminbar.com/wordpress-page-builder-performance-in-the-real-world/ (March 2025)
Key finding from the study author: “If you’re using Divi or Elementor, expect an uphill battle. These builders consistently landed at the bottom, meaning if you’re using them, you’re starting at a disadvantage.”
GenerateBlocks had both the highest median (90) and highest mean (88) – meaning even without obsessive optimization, GenerateBlocks sites consistently score in the “Good” range.
WP Rocket Controlled Test – Divi vs Elementor
WP Rocket built identical test pages in both builders and ran PageSpeed Insights ten times each, averaging results. No optimization plugins applied.
| Metric | Divi | Elementor Pro |
|---|---|---|
| Mobile PageSpeed Score | 64/100 | 75/100 |
| LCP (mobile) | 5.8 seconds | 5.4 seconds |
| Page Size | 874 KB | 940 KB |
| HTTP Requests | 36 | 15 |
| CLS | 0.196 | 0.000 |
Source: wp-rocket.me/blog/divi-vs-elementor-performance-speed/ (December 2025)
Both builders have LCP scores in the “Poor” range (above 4 seconds). Google’s “Good” threshold is under 2.5 seconds. Neither passes without optimization, and even with WP Rocket applied, the structural overhead of both builders limits how fast they can get.
WP Bullet Real Migration – Elementor to GenerateBlocks
A real site migrated from Elementor Pro to GenerateBlocks on the same server (Vultr 1GB VPS, nginx, PHP 7.4), no CDN, no page caching. Same content, same design intent.
| Metric | Elementor Pro | GenerateBlocks | Change |
|---|---|---|---|
| Google PageSpeed Score | 49 | 88 | +39 points |
| Page Size | 512 KB | 331 KB | -35% |
| HTTP Requests | 54 | 34 | -37% |
| First Paint | 1,400 ms | 499 ms | -64% |
| After optimization | — | 95–99 | — |
Source: wp-bullet.com/wordpress-page-builder-performance-case-study-elementor-vs-generateblocks-benchmarks
A 39-point jump in PageSpeed – from 49 to 88 – just from switching the builder. No new hosting, no content changes, same server.
Why the Architecture Gap Exists
Traditional page builders (Elementor, Divi, WP Bakery, Beaver Builder) were built before the block editor existed. They created their own frontend rendering systems – custom JavaScript, custom CSS frameworks, proprietary data storage – that run parallel to WordPress’s native systems.
That parallel infrastructure is the source of the overhead. It doesn’t disappear when you install a caching plugin.
Block-based builders (Kadence Blocks, GenerateBlocks) extend WordPress’s native Gutenberg editor rather than replacing it. Blocks render server-side into standard HTML. CSS loads conditionally – only blocks actually present on a page load their CSS on that page. There is no JavaScript framework required to paint the layout.
This is why the WP Bullet migration showed a 64% reduction in First Paint time: GenerateBlocks doesn’t require JavaScript execution before the page renders. Elementor does.
Consistency: Which Builders Are Most Predictable?
The Admin Bar study also measured “delta” – the gap between a builder’s highest and lowest scores. A large delta means performance depends heavily on how the developer used the tool. A small delta means consistent results regardless of implementation.
Ranking from most consistent to least:
- Oxygen (delta: 31)
- Gutenberg (delta: 33)
- Beaver Builder (delta: 36)
- Elementor (delta: 39)
- GenerateBlocks (delta: 40)
- Divi (delta: 48)
- Kadence (delta: 48)
- Bricks (delta: 59)
GenerateBlocks has a slightly higher delta than Elementor – but its floor is 60 vs. Elementor’s floor of 49, and its median is 90 vs. Elementor’s 66. The variance exists but from a much higher baseline.
What “Needs Improvement” Actually Costs Your Business
A Lighthouse score in the 60s is not just an abstract number. It corresponds to real differences:
The Optimization Ceiling Problem
Caching plugins help. WP Rocket’s own test showed dramatic improvements – Elementor went from 75 to 99, Divi from 64 to 100 – in their controlled test environment. But those are controlled tests on minimal demo pages.
On real production sites with real content, real images, third-party scripts, and standard hosting, the gains are more modest. The Admin Bar’s real-world study shows the practical outcome: agencies who’ve presumably installed optimization plugins are still seeing median Elementor scores of 66 and Divi scores of 62.
The builder determines your practical ceiling in real-world conditions. Optimization helps you approach that ceiling. It can’t raise it.
Bottom Line
The data from three independent sources tells the same story:
- GenerateBlocks delivers the highest and most consistent Lighthouse scores in real-world conditions. Median: 90.
- Elementor and Divi consistently land at the bottom of real-world performance rankings. Median: 66 and 62 respectively.
- Switching builders – not optimizing – is what produces the largest performance gains. A real Elementor-to-GenerateBlocks migration showed a 39-point PageSpeed improvement on the same server.
If your site is built on a heavy builder and performance matters to your business, the data points toward one conclusion: the builder needs to change, not just the settings.
WP Boosters specializes in rebuilding WordPress sites from slow, bloated builder installs to lightweight, high-performance builds using GeneratePress, GenerateBlocks, Kadence, and Kadence Blocks – with a guaranteed 90+ Lighthouse score. Projects start at $2,500. Get a free quote