About this brotli compression checker
Brotli is a compression format developed by Google that shrinks text files such as HTML, CSS and JavaScript more than gzip does, usually by 15 to 20 percent. Browsers ask for it with the Accept-Encoding header, and the server replies with Content-Encoding: br when it supports it. Smaller files mean faster pages and lower bandwidth bills, which makes it a standard item in speed audits.
This tool sends three requests from our server to the page you name, asking for Brotli, then gzip, then no compression at all, and compares the answers. You see the encoding used each time, the number of bytes received, and the estimated saving of Brotli against an uncompressed response when both sizes could be measured in full. Some servers skip compression for very small files or for types that do not compress, so test a real HTML page.
If Brotli is missing, enable it in your web server, your CDN or your hosting panel; many CDNs offer it with one switch. The address you type is sent to our API at /api/brotli-checker.php. Redirects are followed up to three times and are checked first, private addresses are refused, and the app does not store searches.
Which controls should I check first?
The Brotli Compression Checker form labels its controls as Website address. Use those labels to keep the source and settings separate. A sample argument that selects a mode is not additional source text. Choose the setting matching your intended operation before copying a result, and change it explicitly when comparing another mode. The labels describe what this form accepts; the worked examples below demonstrate only the stated supplied values.
How should I interpret this specific operation?
Compression observations depend on the resource, request headers and response. An already compressed image may be treated differently from a text document. A reported byte saving compares the representations the check actually obtained; it is not a prediction of total site performance. Preserve the content-encoding and relevant sizes rather than copying only a percentage. A server's support for one response also does not prove that every cache layer serves the same encoding to every client.
Worked helper values
These values come from the reproducible calculation examples supplied with this tool. Each row identifies the complete supplied arguments and the returned value. A visible ↵ denotes a line break and ⇥ denotes a tab in text examples. Settings appear alongside source data so that a different mode is not mistaken for an identical input. These are helper calculations, not observations of a live device, fetched target, image decoder, browser preview or downloaded file. Use the stated method above to relate the returned fields to the actual operation.
| Supplied values | Calculated return value |
|---|---|
| brotli-checker; a b.com | /api/brotli-checker.php?q=a%20b.com |
| example.com ; x | example.com |
Whose network view does this result represent?
A server-side check observes the target from the checking service's network location. Your browser, a mobile connection and another resolver can see different answers because of caching, geolocation, routing, filtering or load balancing. Keep the exact queried target and the time of your check when comparing observations. A timeout means this attempt did not obtain a result within its limit; it does not prove the target is unavailable everywhere. To investigate a discrepancy, compare the same target and protocol from the relevant client environment as a separate step.
How should I prepare a target?
Use a public target whose spelling and scope you can explain. A hostname, an IP address and a complete URL carry different information. A URL includes a scheme, path and possibly a query string, while a hostname identifies a host without those request details. Follow this page's stated input convention and omit credentials or confidential query values from demonstration targets. Correct an invalid-target message before retrying. The address is sent to the checking service for processing, so this workflow is different from a local parser that merely splits the supplied text into components.
What can I infer from missing fields?
A missing field can reflect the target, the available response or the check's scope. It should not be replaced with a guessed value. Read any status or explanation beside the output and distinguish an empty answer from an error. Compare the same type of evidence across repeated checks rather than treating every field as an independent verdict. A registration record, resolver result and HTTP response describe different layers and can disagree without any of them being fabricated. Follow a discrepancy to the layer it concerns before changing DNS, application settings or certificate configuration.
How can I document a reproducible observation?
Record the exact input, result fields that matter, and any timeout or error text. If the target is a URL, preserve the path and protocol because checking the homepage can differ from checking a specific resource. When a value appears to change, repeat a narrowly scoped check and compare the complete response, including intermediate context if the tool exposes it. Avoid attributing a difference to a named provider or person based only on one field. These outputs are observations from a limited check, not a record of ownership or an exhaustive audit of every configuration on the target.
Is Brotli better than gzip?
For text files it usually produces smaller output at a similar speed. Both are supported by all current browsers over HTTPS.
Why does the tool say no compression?
Some servers only compress files above a size, or only for certain content types. Try the main HTML page rather than an image or a tiny file.
Why do I need HTTPS for Brotli?
Browsers only request Brotli on secure connections, so servers often enable it for HTTPS only.
How accurate is the saving percentage?
It compares the full sizes received. If a response was longer than 1.5 MB the percentage is not shown.