Making Data Compressible

One question we all asking, I believe, "can we make data (more) compressible?"
I have been searching for the answer myself and so far I only have these two not-so-useful ways to do just that.

Like I described in Analysis on compressed data, to make data more compressible is to make them having an un-balance ASCII number, these two not-so-useful ways are shuffling and flattening. But what do I mean by not-so-useful ways? These methods can change the data content but even though it can be compressed, the result is not up-to-the-expectation (not good), they are:

Shuffling (without additional byte)
Shuffling will change the data content without the need for additional byte. It is the simplest method and similar to the simplest data encryption. It is simply changing the content in random hoping that the result will be un-balance. But I found the result always more balance (worst, decompressible).

Flattening or Eliminator (with additional byte for flag)
This one will eliminate one or more ASCII value from the content and use additional bytes (Flag) to make it able to decode it back to the original content. The result can be compressed using some of the compression method, but because of the additional bytes, the result is not smaller than the original size without this method.

Since there are plenty of ways to change the data content, I believe these two ways can be improved and when I or any of you people found a better way with a better compression ratio we will stop this research.

Shuffling