I've never actually done it, but I've seen it done a lot over the years.
The technique is very simple, you feed +5V through a resistor to the A2D input, so it now reads 5V - no button pressed. You then have a series chain of resistors from this point, with switches down to ground. So when you press a switch it creates a potential divider with the upper feed resistor. By carefully selecting the resistor values you can find which key was pressed by measuring the voltage.
It's usual to have one key which shorts directly to ground, so that key would read 0V, no key 5V, and all others somewhere inbetween.
How many buttons are you wanting?.