Google: Raw and Encoded Bracket URLs Are Different URLs

Google Search Advocate John Mueller settled a long-running technical SEO question on Bluesky: a URL that uses raw square brackets and the same URL with percent-encoded brackets (%5B and %5D) are not the same URL.

He said he would have assumed they were equivalent, and that he may have given people the wrong answer in the past. The practical takeaway: Google will crawl both versions. If it finds the same content, it will de-duplicate and pick one. Your job is to make sure that choice goes your way.

Why encoded brackets are treated separately

The difference traces back to RFC 3986, the internet standard for URI syntax. Brackets and plus signs are in the reserved character set. Under Section 2.2, replacing a reserved character with its percent-encoded form makes a URI that is not equivalent.

Mueller explained that raw brackets in a query string are technically invalid by that standard, because brackets are reserved for wrapping raw IPv6 addresses in the hostname. One small exception: letter case inside the encoding does not matter, so %5B and %5b are treated as equivalent.

What Google does with both versions

When a server returns identical content for the raw and encoded forms, Google will crawl both. If the content is the same, possibly at checksum level, Google de-duplicates and picks one canonical version.

Browsers may encode brackets before sending what a user types in the address bar. But a request made from JavaScript or other code can send the brackets as they are. Mueller’s guidance for URLs is blunt:

always encode it, never show []

That matches Google’s URL structure documentation, which says characters defined as reserved by the standard must be percent encoded.

What existing bracket-based sites should do

Vanessa Fox, who earlier ran Google’s Webmaster Central, described a site that uses raw bracket notation throughout, still indexes fine, and has chosen to leave the structure alone. Mueller said he would not expect problems from that existing setup, but he would not set up a new site like this.

He also cautioned that other crawlers and fetchers may not be as exact as Google in handling these characters, based on issues Google encountered more than ten years ago.

If you already publish URLs with raw brackets, the safe move is consistency:

• Audit every internal link, canonical tag, sitemap entry and piece of structured data for raw [] versus %5B%5D.
• Choose the encoded form for all new URLs.
• Ensure existing links and metadata point to one version, not both.
• Test the crawlers and fetchers that drive your own analytics and tooling, not just Google.

The same logic applies to spaces, plus signs and %20, which are also reserved characters in the URI standard.

Why this matters for performance marketers

For e-commerce and D2C sites, query-string filters such as ?filter[]=123 are everywhere. If your ads, internal links and canonicals disagree on the bracket format, you are splitting signals across multiple URLs and asking Google to choose which one to keep.

Standardising the format is a small change that reduces index bloat, keeps ranking signals consolidated, and gives you control over the version in search. It is not about whether the URL works. It is about which URL owns the signal.

Source: Search Engine Journal