So normalize increases the volume of the track such that at the loudest moment during the normalized range, every available bit is used (100% normalization). There are grades down from this. Everything gets louder in equal proportion, the same as when you increase the volume on the track. I use normalization for gain staging into my FX usually. It gives me an easy way to keep different songs hitting the compressors on the tracks in a similar way. The signal strength into the compressors has everything to do with what will get compressed.
If you want the track to appear to be the same volume from beginning to end, you could use a volume envelope and adjust the volume of quieter passages up and the louder passages down. Doing this impacts all parts of the peaks and valleys of the wave form in equal proportions.
Compression enables you to only impact the louder moments, effectively enabling you to lower the volume only on a range of the volume spectrum rather than equally across the entire spectrum. This can make it so that it's easier to hear the quiet moments after the guitar is struck, for example. The work of a compressor cannot be mimicked with volume fader automation. It is possible to mimic the attack and release part of the compressor (reduce here, increase there), but when you reduce, you reduce all parts of the spectrum equally. It's not possible to only reduce the loudest 20% to 1/2 it's volume increase with a volume fader. When you reduce the volume, all parts of the spectrum are reduced equally.
Compressors come in a great variety of choices are usually used in conjunction with volume envelopes.
Read up on the compressors you have in Sonar. Hit F1 for help. See if you don't get some ideas of things to try.