Huon Wilson wrote an article titled 10 > 64, in QR codes, about using decimal encoding to reduce QR code density.

The article mentions that a QR code can contain segments in different “modes”, such as numeric, alphanumeric, binary and kanji. I didn’t know that! I had assumed that the entire QR code had to use the same mode.

Something I keep returning to is that if a QR code encodes a case-insensitive URL, you can reduce the density by storing it as uppercase, thereby enabling the use of the alphanumeric mode instead of the binary mode. Since the protocol part (https://) and domain name part (www.example.com) are always case-insensitive, it might be possible to encode some URLs more efficiently even if the rest of the URL is case-sensitive.