To do this, you'd need each station to submit more than their playlist. You'd have to make available the daily music logs to truly expose what turns listeners on, or off.
Technically speaking, most of these playlists ARE freely available - you just have to throw an internet 'net' over them to gather the data. Visit ZM, Niu FM, Mai FM, Flava - most stations that have an active online presence usually have a 'now playing' ticker. It would just take a bit of code to subscribe to this information, collate and generate a dynamic html type document. By 24 hours, you'd have a good idea of high rotates, the back library, and WHEN.
To truly track a song, you may have to measure it by having surveys crunched to 5 minute blocks (rather than 15 at present), and graph the numbers that way.
Perhaps a more powerful way of measuring the success of a song (and no, not based on sales), is using something like Audioscrobbler that records playlists data. Combine this information with listening data from Research International, and you have oblivious listening trends and a way to measure the true appeal of such logged song.
Further to this, the net that Audioscrobbler has cast could work on an international scale. It collects data from all stations generating a playlist. The true popularity of a song could be measured more accurately between Frankfurt and Sydney, for example.
Take Tam, for example. If a graph of listening habits like this represented the overall popularity/appeal of a song gathered from all Auckland, NZ or Worldwide radio - it would not only be a true representation of good programming, but loyal listenership too (especially if the 'hottest' song was also her favourite).