What is Brotli?

Brotli: smaller text than gzip, when configured correctly

Brotli (Content-Encoding: br) is Google’s open-source compression algorithm designed for web text. On typical HTML, CSS, and minified JavaScript, Brotli achieves roughly 15–25% smaller payloads than Gzip at similar decompression speeds in browsers—translating directly into faster downloads for users and crawlers.

Brotli is not a replacement for sound architecture. It will not fix unminified 3 MB bundles or unoptimized hero images. It will squeeze already-clean text assets further—low-hanging Technical SEO fruit when paired with code minification and proper cache-control.

This page covers Brotli specifically. For negotiation basics and MIME-type rules, see Compression. For legacy fallback configuration, see Gzip.

Brotli vs Gzip: decision matrix

Always ship both encodings for static assets. Brotli-first negotiation, gzip safety net.

Why Brotli beats gzip on text

Brotli includes a built-in static dictionary of common web tokens (<div, function, https://)—repeated strings in minified JS compress aggressively. Gzip’s smaller window still performs well but lacks those web-tuned literals.

Example illustrative savings on a 120 KB minified bundle:

Exact ratios vary; measure your assets.

Precompression workflow (recommended)

Build step:

# Illustrative — tools: brotli CLI, vite-plugin-compression
brotli -q 11 -o app.a1b2c3.js.br app.a1b2c3.js
gzip -k -9 app.a1b2c3.js   # gzip sibling for fallback

Deploy all three: app.js, app.js.br, app.js.gz.

nginx brotli_static on; gzip_static on; serves correct file with encoding header based on Accept-Encoding.

Benefits:

• Zero per-request compression CPU.
• Maximum ratio (quality 11) without slowing TTFB.
• Stable bytes for browser caching (immutable hashed names).

Dynamic Brotli for HTML

SSR and edge rendering may compress HTML on the fly:

brotli on;
brotli_comp_level 5;
brotli_types text/html text/css application/javascript application/json image/svg+xml;

Higher levels increase CPU—watch origin load during traffic spikes. CDN dynamic compression offloads this.

Response headers

HTTP/2 200
Content-Type: application/javascript
Content-Encoding: br
Vary: Accept-Encoding
Cache-Control: public, max-age=31536000, immutable

Test:

curl -sI -H 'Accept-Encoding: br' https://cdn.example.com/app.js | grep -E 'encoding|content-type'

Brotli and Core Web Vitals

Smaller transfers improve:

• LCP — critical CSS/JS/HTML arrive sooner.
• INP — less time blocked on network before script parse (especially on slow 4G).
• TTFB follow-up — not TTFB itself, but total page load window shrinks.

Lab Lighthouse may show modest gains; field CrUX improvements matter for ranking signals.

Crawl efficiency

Googlebot fetches HTML, JS, and XML sitemaps. Brotli-shrunk sitemaps and JS reduce per-URL bandwidth—useful on million-URL catalogs where crawl rate is constrained by host response time and size.

CDN defaults

Cloudflare, Fastly, Akama, and Vercel often enable Brotli automatically for eligible MIME types. Verify:

• HTML from origin is br when Accept-Encoding includes br.
• Custom error pages compress too.
• Vary: Accept-Encoding present on cached objects.

Pitfalls unique to Brotli

Brotli vs Zstd

Zstd (zstd) is emerging in some CDNs with excellent speed/ratio tradeoffs. Adoption is narrower than Brotli today. Monitor your CDN; Brotli + gzip remains the 2026 mainstream pair.

Relationship to image optimization

Brotli does not replace AVIF/WebP. A 2 MB PNG stays ~2 MB with br. Optimize images at codec layer; Brotli handles text stack.

Implementation checklist

□ CI emits .br for JS, CSS, SVG, optional HTML shells
□ Origin/CDN brotli_static + gzip_static enabled
□ brotli_types / gzip_types include application/javascript
□ Vary: Accept-Encoding on all compressed cacheable responses
□ curl tests for br and gzip paths after deploy
□ Lighthouse transfer sizes down vs pre-Brotli baseline

When Brotli matters most

High impact:

• JS-heavy client-side rendering sites.
• Large design-system CSS bundles.
• Huge XML sitemaps and RSS feeds.
• Global audiences on edge rendering CDNs.

Lower impact:

• Tiny mostly-static HTML brochure sites already under 50 KB—still enable, but measure before over-optimizing infra.

How Crawlox helps with Brotli context

Crawlox identifies pages with disproportionate downloaded weight during crawls. If text assets remain large despite Brotli at CDN, the issue is usually unminified sources or double-stacked libraries—not algorithm choice. Fix minification and bundle splitting first; then verify Content-Encoding: br in production to capture remaining byte wins for crawlers and users alike.