Gzip Compression Ratio Benchmarks by File Type

Compression Benchmarks by File Type

Different file types compress at very different rates. This reference provides real-world benchmarks to help you estimate compressed transfer sizes.

Text-Based Web Assets

File Type	Typical Ratio (Gzip)	Typical Ratio (Brotli)	Notes
HTML	70–80%	75–85%	Tag repetition helps greatly
CSS	80–88%	83–90%	Highly repetitive syntax
JavaScript	65–75%	70–80%	Minified JS compresses less
JSON	75–90%	80–92%	Key repetition in arrays
XML	75–85%	80–88%	Tag names repeat heavily
SVG	70–80%	75–85%	XML-based, compresses well
Plain text	60–70%	65–75%	Less structured patterns
CSV	70–80%	75–85%	Header names repeat in some formats
YAML	65–75%	70–80%	Indentation helps compression

Already-Compressed Formats

File Type	Gzip Ratio	Worth Compressing?
JPEG	0–2%	No
PNG	0–1%	No (already DEFLATE)
WebP	0–1%	No
WOFF2	0–1%	No (already Brotli)
WOFF	2–5%	Marginal
MP4	0–1%	No
ZIP	0–1%	No

What Drives Compression Ratios

The key factors that determine how well content compresses:

1. Repetition density: More repeated patterns = better compression
2. Alphabet size: Fewer unique characters = better Huffman coding
3. Pattern proximity: Patterns within 32 KB window compress best
4. Content entropy: Random/encrypted data does not compress

Practical Application

When estimating transfer sizes for performance budgets:

• CSS budget: Multiply your raw CSS size by 0.15 (85% compression)
• JS budget: Multiply your minified JS by 0.30 (70% compression)
• HTML budget: Multiply your rendered HTML by 0.25 (75% compression)
• JSON budget: Multiply your JSON by 0.15 (85% compression)