We have been watching the recent developments in Internet radio audience measurement.  Here are the stakes:
ARBITRON   is trying to develop new business and generate more revenue
PANDORA wants to have audience data to sell advertising and capture a portion of ad revenue going to radio
ANDO/TRITON wants ARBITRON to stay out of Internet audio measurement so it will not have a giant competitor
CLEAR CHANNEL et al don’t want Pandora to be able to compete for radio advertising.
What all parties need to consider is what advertisers want.  I have to chuckle at some of the points raised recently by Arbitron.  Does a moving people meter really mean someone is listening to radio?  If an audio source is present near a people meter did the listener actually initiate it?  Was it really their choice?  I have always believed that server side data is better than an estimate.  PPM methodology is an estimate based on a representative sample (many argue the sample is too small).  With server side data you have actual data – use it.  Ando Media currently uses server side data to prepare its Webcast Metrics Internet radio top 20 ranker.  Arbitron has announced that they are going to utilize server side data as well although many of the details are still forthcoming.  Arbitron is being held hostage by its existing terrestrial radio clients over measuring Internet radio only stations such as Pandora.  Of course terrestrial radio companies represent the bulk of Arbitron’s current revenue.
Audience data is utilized in Internet radio as a predictor of the number of ad impressions that can be generated.  Audience measurement data is not used to calculate ad impressions. When a listener starts a stream then most ad servers, including Ando, will then insert ads prior to the stream playing ( preroll) or then begin inserting audio on receiving an ad cue.  You do not have to be listening for any set period of time.  By its very nature, in Internet radio each impression is counted when a listener actually hears the ad.  Do advertisers care that different people hear different ads?  You bet but not why ARBITRON raises this issue.
Maybe what we need is an impression measurement validation firm.  In any event it’s going to take some time to sort out Internet radio measurement and it’s unfortunate that there is not a larger advertiser voice?  Some big dogs are in the fight.