HTML compression

Compression on the HTML document: brotli, gzip, or none.

Field data PhoneDesktopAll Scope All sites Q2 2026 edition · Phone field outcomes
Metric LCP INP CLS
1

At a glance the headline numbers for HTML compression

Compression on the HTML document: brotli, gzip, or none.

8
Categories
185,477 sites
46.1%
Top share
gzip
0.00
Top CLS
Median for gzip

27.9% of HTML documents are Brotli-compressed. 21.0% ship raw.

The State of Web Vitals · Q2 2026 · 189,915 sites · phone field datacorewebvitals.io/state-of-cwv
2

HTML compression who uses what, and how stable each group is

Median CLS
0
0.05
0.10
0.15
0.20
0.25
gzip0.0046% of sites
br0.0028% of sites
none0.0021% of sites
zstd0.005% of sites
deflate0.000% of sites
dcb0.000% of sites
identity0.000% of sites
application/json0.860% of sites
VariantShare of sitesMedian
gzip 85,419 sites
46%
0.00
br 51,784 sites
28%
0.00
none 38,901 sites
21%
0.00
zstd 9,278 sites
5%
0.00
deflate 86 sites
0%
0.00
dcb 6 sites
0%
0.00
identity 2 sites
0%
0.00
application/json 1 sites
0%
0.86

Gzip sites post the best median CLS (0.00). Application/json sites trail at 0.86. Correlation, not causation. computed

The State of Web Vitals · Q2 2026 · 189,915 sites · phone field datacorewebvitals.io/state-of-cwv
3

Passing CLS per HTML compression which group passes the CLS most often

VariantSitesPassing CLSCLS
Dcb 0% 6 100% 0.00
Identity 0% 2 100% 0.00
Br 27.9% 51,784 90% 0.00
Gzip 46.1% 85,419 89% 0.00
None 21% 38,901 88% 0.00
Zstd 5% 9,278 86% 0.00
Deflate 0% 86 83% 0.00
Application/json 0% 1 0.86
Good Needs Improvement Poor Sorted best-passing first · median colored by its own rating · pass = good CLS (0.1 at p75) · one value per site

No variant stands out: pass rates sit between 86% and 90%. computed

The State of Web Vitals · Q2 2026 · 189,915 sites · phone field datacorewebvitals.io/state-of-cwv
4

Why this matters for the Core Web Vitals, and where to start fixing it

The HTML document is the first thing the browser reads, and the sooner the markup arrives, the sooner the preload scanner can discover everything else: the stylesheets, the fonts, the LCP image. Brotli and gzip shrink HTML to a fraction of its raw size. None means the visitor downloads every byte of markup before the page can even start discovering resources.

Brotli compresses tighter than gzip and every major CDN offers it. This is a switch, not a project.

Related signals HTTP protocol mix → Connection hints → Cache strategy mix → Uncompressed text responses → Chrome field data from 189,915 sites, representing millions of real page loads · How we measured