Think of the the send as being similar to a wet/dry. If you put a reverb on a single track as 10% wet, then you have 10% less dry of a signal (because it's now only at 90% dry instead of 100%). Additionally, if you want to adjust the amount or "wetness" of the reverb later, you're going to affect the dry signal as well (doubling the reverb to 20% would also chop off another 10% of the dry signal).
With a send you keep your original track 100% dry all the time, then adjust the reverb on the send to 100% wet and you can adjust the mix by turning the send track up or down in volume. By adjusting the send/reverb track up or down in volume you can adjust the wet signal without affecting the dry signal. This is all very basic send stuff though. The real power is in selecting a reverb, delay, or other effect and routing multiple tracks to it. This way you can have the same effect across multiple tracks without duplicating the effect you're using on each track and killing your computer processor. If you want your drums to be dryer than your keys, no problem. Just send less of a signal, maybe -20, on the drums and send the keys over at -5. It's super handy.
Taking it even another step further, you can process your effect chain without it messing with the raw instrument sound. A common example would be putting some eq on your reverb send so you can thin it out and leave more room in your mix without having to cut frequencies from the dry part of the signal. Most reverb's have this built in so you could do it without a send, but you can be more creative and try putting a phaser, flanger, weird panning, etc. on your reverb send to really mess with things.
Hope that helps, there are a lot of potential ways to improve your effect chains with sends. This is just grazing the surface.