Right now you have graphs for CPU, Bandwidth, Disk, but that's only graphs and that's not quite enough.
I think you can provide some sort of "monitoring as a service" for droplets.
This can be done agent-less, by providing some standartized way (for graphs it can be just graphite-compatible input on your side). Or you can release optional agent.
Also there are some checks (TCP checks) that can be done server-side, and with them you could check for service availability (e.x. HTTP checks, SSH checks, mysql checks, etc - just to test if TCP connection can be established).