Compression comes in two flavors: lossy, and lossless. Lossy compression can take file sizes down significantly, but some data will be lost (hence the name). Lossless compression on the other hand keeps all original data intact, but usually can't take file sizes down as much.

Due to the nature of SVGs being vector-based[^1] XML-defined[^2] graphics, unless you manually optimize them you can't get much better compression than removing the whitespace and comments in the files (if any).

If you want to maintain acceptable visual quality and are ok with them being quite small, you can rasterize them into a PNG, for example. Online tools such as CloudConvert or command-line tools such as ImageMagick can help you with this.

tl;dr: unless you need to use SVGs, convert them into tiny raster graphics.

[^1]: i.e. the image is defined as a series of curves and shapes instead of pixels. This generally results in less information per image ∴ smaller file sizes.

[^2]: i.e. it's written in plaintext, which can't be compressed without overhead.