Render-blocking requests: fixes with or without a developer
Who adds each file on the list, what you can switch off without code, and what takes a developer.
What does Lighthouse mean by render-blocking requests?
Until every stylesheet and ordinary script in your page's head has arrived, your visitor looks at a white screen. Lighthouse lists those files under Render-blocking requests ("Requests are blocking the page’s initial render"), called Eliminate render-blocking resources before Lighthouse 13. Give the scripts defer, and load only the CSS each page uses, in a few files of about 15 kB compressed.
Just want the quick fix? Jump to WordPress, Shopify, your own templates or Next.js.
Which files does Lighthouse list under Render-blocking requests?
The report says: "Requests are blocking the page’s initial render, which may delay LCP. Deferring or inlining can move these network requests out of the critical path." The list holds the stylesheets and ordinary scripts (without defer or async) that your first paint waited for.

A shop's product page in Lighthouse 13.6, mobile, default simulated throttling.
- Transfer Size: the file as sent, compressed.
- Duration: Lighthouse's estimate for a slow phone, a new connection to another domain included.
- An empty Duration: the estimate was under 50 ms.
- Est savings: how much earlier Lighthouse estimates the page can paint.
The notice has no weight in the Performance score, but these files delay FCP (first paint) and LCP (largest element), and those two count. It passes when nothing is listed: one small file keeps it red if Lighthouse estimates a saving, orange if not.
Why does the page stay white until the last file arrives?
Your visitor taps a link and the screen stays white, then the page appears all at once.
The browser reads the HTML, collects the stylesheets and ordinary scripts in the head and paints nothing until they are in. Those steps are the critical rendering path. A script after a stylesheet waits for that stylesheet, and the browser stops reading the HTML until the script has run. So the first paint waits for the file that finishes last, and removing a small file that finished early gains almost nothing.
| In Chrome | Holds the first paint? |
|---|---|
<link rel="stylesheet">in the head | Yes |
@import ina stylesheet | Yes, and it starts late |
<script src>in the head | Yes |
defer, type="module" | No |
async | No, but it can run before the paint |
media="print" | No, and it downloads anyway |
How much later does your visitor see the page?
I measured a test page: a hiking shop's product page with the head WordPress prints today for WooCommerce, a form plugin and a block plugin. Every number below is from that page, the middle of five loads on a throttled phone.

Before

After
1.7 s into loading, before and after trimming the CSS.
Deferring the scripts alone gains almost nothing, because the CSS finishes later. With the page's unused plugin CSS off too, the first paint is about 85 ms earlier. Those two are the settings you can switch yourself, and they give no more here: a stylesheet that another stylesheet loads with @import now finishes last. The rest takes a developer: with that gone and the fonts on the shop's own domain the gain is about 240 ms, and trimming the CSS to what the page uses brings it to about 650 ms. LCP can't come before the first paint, and with less CSS the product photo arrives earlier. The finished page looks the same in every row.
| Each row adds to the one above | First paint | LCP |
|---|---|---|
| Before | 2.13 s | 2.31 s |
| Scripts deferred | 2.12 s | 2.31 s |
| Unused plugin CSS off | 2.04 s | 2.20 s |
@import gone, fontson its own domain | 1.89 s | 2.19 s |
| All the CSS it uses in one 9 KiB file | 1.48 s | 1.94 s |
Do render-blocking requests matter on a fast server?
According to the State of Web Vitals, 92.9% of sites have at least one render-blocking request.

Sites with a good FCP on phones, by render-blocking requests and TTFB at p75. Data: State of Web Vitals (corewebvitals.io)
With a server that answers in under 600 ms (time to first byte, TTFB) the share with a good first paint (FCP) hardly changes with the number of requests, and the wait from first byte to first paint is about 150 ms longer at 21 or more. Just past a second of TTFB, sites with 21 or more are 7 to 14 points below those with two or fewer, and those are already 14 to 30 points below sites under 600 ms. With a TTFB that high, fix the server first. That is a correlation.
You can leave the notice alone when:
- Lighthouse lists one or two small stylesheets of your own, with little or no Est savings;
- what's listed is a consent tool or a platform file (what to do instead).
A file behind an @import is a chain, and a chain gets fixed: Network dependency tree.
How do you fix Render-blocking requests?
The URL in the list tells you who owns each file. Start with the CSS, because it usually finishes last: take the largest file your page doesn't use, then any file on another domain (Google Fonts go on your own domain).
WordPress
- See who loads the file.
/wp-content/plugins/<name>/is that plugin,/wp-content/themes/your theme,/wp-includes/WordPress. - Switch a plugin's CSS off where the page doesn't use it. Contact Form 7 loads on pages without a form, WooCommerce outside the shop. Asset CleanUp (free) and the Script Manager in Perfmatters (paid) switch a file off per page or post type. Use Test Mode (Asset CleanUp) or Testing mode (Perfmatters) first, so only you see it, and check the page, form and cart.
- Defer the scripts with one setting. Load JavaScript deferred in WP Rocket, which also handles inline
jQuery(...)calls. Defer JavaScript in Perfmatters, with Include jQuery tested separately. Load JS Deferred, set to Deferred, in LiteSpeed Cache. Then click through the menu, gallery and cart. Delay JavaScript also clears the notice, and the script then runs on the visitor's first tap or scroll. - Combine what's left. WP Core Web Vitals, my own plugin, puts your stylesheets into a few files of 12 to 14 kB compressed, and defers the scripts, inline ones too. LiteSpeed Cache's CSS Combine makes one file. Let one tool combine. WP Rocket no longer has this setting.
- Leave Remove Unused CSS and critical CSS off. WP Rocket recommends Remove Unused CSS for this notice. It puts the CSS it finds per URL inside the HTML, so every page view downloads it again. Critical CSS puts the first screen's styles in the HTML and loads the rest after the first paint, so the page styles itself twice and often shifts.
- Where this stops. The theme's own big stylesheet, a script written into the theme and an
@importin a theme file take a WordPress developer. Send the notice's URLs and the percentage unused from the Coverage tab in Chrome DevTools. Unloading and deferring on a normal theme is half a day to a day; trimming a theme's own stylesheet is several days. Trimming gave most of the gain on the hiking shop's test page. - What undoes it. A new plugin often loads its CSS on every page, and a plugin update can bring back a file you switched off: check the list after each one. Edits in a parent theme disappear with its next update, so code goes in a child theme.
In code, the handle is the id WordPress prints without -css (contact-form-7-css):
add_action( 'wp_enqueue_scripts', function () {
if ( ! is_page( 'contact' ) ) {
wp_dequeue_style( 'contact-form-7' );
}
}, 20 ); WooCommerce's stylesheets have their own filter, woocommerce_enqueue_styles. Your own scripts get defer from WordPress:
wp_enqueue_script( 'acme-gallery', get_stylesheet_directory_uri() . '/js/gallery.min.js', array( 'jquery' ), '2.3.0', array( 'strategy' => 'defer' ) ); WordPress drops that defer without a warning when a script that depends on yours still blocks or an inline script comes after it, so check the tag in the page source. WordPress registers jQuery without a strategy, and defer added by hand breaks every inline jQuery(...) call: defer jQuery in WordPress. For a child theme whose style.css starts with @import, the code is on Network dependency tree.
Shopify
/cdn/shop/t/ in the URL is your theme, /extensions/ an app, with its name in the path. Without code: in Online Store, Edit theme, App embeds, switch off the embeds you don't use, because each one's stylesheet goes in the head. The first paint moves only if that file finishes last. An app you uninstalled can leave code behind: search theme.liquid and the snippets for its name, or ask the app's developer.
Scripts written into the theme and heavy CSS take a Shopify developer: send the notice's URLs. Deferring is half a day to a day on a normal theme; trimming its own stylesheet is several days. script_tag takes defer directly:
{{ 'gallery.js' | asset_url | script_tag: defer: true }} A new app can add an embed: check the list. A theme update arrives as a draft copy: your code edits stay unless they conflict, and then the theme's card says so.
Your own templates
Aim for one small stylesheet and every script deferred:
<link rel="stylesheet" href="/css/main.min.css">
<script src="/js/main.min.js" defer></script> - Trim. The Coverage tab in DevTools shows what the page never uses. According to the State of Web Vitals, the median home page with blocking CSS leaves 88% of its CSS unused.
- About 15 kB per file, compressed. A new connection delivers about 14 kB in its first round trip, so 15 kB is the budget.
- Combine the tiny ones up to that budget: every 1 kB file is its own request.
- One big file isn't the goal. Untrimmed, one file holds the same bytes.
- No inlining, whatever the notice's description says. The HTML gets bigger, everything waits for it and it usually isn't cached.
- No "rest of the CSS" loaded later.
media="print"only suits CSS that is off screen on every page at every width, which is rare. fetchpriority="low": on four stylesheets of the test page the first paint moved from 2.13 to 2.12 s and Lighthouse still listed them, so skip it.@import. Removing it on the test page gained about 110 ms with the scripts deferred. Network dependency tree shows more: two in a row, the last to another domain.- 103 Early Hints starts your stylesheet while the server still builds the HTML.
Next.js
Global CSS in the root layout goes to every route, a CSS Module only to the routes that import it, so keep the global file small. Load a script somebody added by hand with next/script, which defaults to afterInteractive:
import Script from 'next/script'
<Script src="https://example.com/widget.js" /> Files you can't change
- A consent tool with auto-blocking (Cookiebot, OneTrust, Usercentrics) has to run before the scripts it stops, so it blocks by design. Manual mode is the alternative; settle the legal side first.
- An A/B testing snippet (Optimizely) blocks so visitors don't see the original page first. Remove it when no test runs.
- Shopify's
content_for_headeraddedaccelerated-checkout-backwards-compat.csson all five stores I checked. Leave it.
Is the first paint earlier now?
Run Lighthouse three times and compare the middle results. The hiking shop's test page went from 13 files to one, score 97 to 100. That stylesheet stays listed, in red with 150 ms of Est savings, and you can stop there.

Before

After
Lighthouse 13.6 on mobile, simulated throttling: estimates, so they differ from the measured table.
- The real wait: record the page loading in the DevTools Performance panel. The Insights sidebar shows the measured Duration per file. The bar that ends last before the FCP marker is what your paint waited for.
- Empty list, page still blank: an A/B tool hiding the body with
opacity: 0, a page built by JavaScript, or a slow server.
Field data shows your real visitors. In CoreDash, compare FCP with TTFB per URL on mobile and look at LCP render delay. Search Console takes up to 28 days to show the change.

CoreDash, FCP per URL at p75.
The file that finishes last may belong to a theme, builder or app you can't change, or the wait may be in your server. Then book a Core Web Vitals power hour: we record your page, find what your first paint waits for and settle who changes it.

Need a faster site?
I'm Arjen, and I've been fixing slow sites for 17 years. If you're running into issues you can't fix yourself, send me your URL and let's have a chat.
Send me your URL
Why isn't our fix working?
Sometimes the CDN still serves the old HTML, or the fix went into the desktop template and mobile uses another one. Send me your URL and the problem, and I reply with a few times. I check your field data first, then we spend one hour on your site, and within a day you have my notes in writing. €350 excl. VAT.
Send me your URLNo developer to build the fix? I do the work myself, on a copy of your site.