Pandora: Damned with faint praise


If you were called average, would you feel pleased? Special? What if you called your significant other average? Would you expect your comment to be well received?

To be average is…nothing special.

So a recent blog post by a new-media pundit calling Pandora average, and then going to great lengths to prove it, caught our attention. Here’s what he wrote:

Don’t let ill-informed yahoos convince you that Pandora and its like are somehow less “sticky” than the average music radio experience.

For someone who has devoted a lot of space praising Pandora and portraying it as a serious threat to radio, reducing it to an average music radio experience is clearly damning with faint praise.

The problem is that the blogger inadvertently proved that Pandora isn’t average. Pandora is worse than average.

His conclusions were based on a comparison between Ando Media estimates for Pandora and local radio numbers provided by Arbitron “informed sources.” Read the entire post here.

The post may have been motivated by the fact that Pandora has the lowest TSL (time spent listening) of any service rated by Ando Media, a little less than an hour per session. The TSL of most broadcast group streams is two to three times as great. This is a subject we covered last fall here.

So rather than compare Pandora to the other Ando Media measured streams, the author chose to compare the service to Arbitron rated radio stations.

The unnamed source at Arbitron advised the author that the average person listens to a music station two hours a day. According to the author, the average Pandora user spends an hour with Pandora twice a day, or two hours a day.

In other words, the Pandora user spends as much time online as a radio listener listens! Wow! Pandora is just as sticky as radio!

What the author fails to understand is that Ando Media and Arbitron do not measure TSL in the same way. Ando counts any listening over a minute. Arbitron only counts quarter-hours of five minutes or more of listening. It does not credit minutes 1-4 that Ando does.

While that may seem like a trivial difference, when Ando switched from Arbitron’s five minute requirement to their new much more lenient one minute rule last year, the difference boosted TSL of streams we checked by an average of 167%, and as much as 250%!

The only reason Pandora has an hour of TSL is because Ando Media lowered the bar to arbitrarily boost its clients’ TSL. Were Pandora measured like broadcast radio stations are measured, its TSL would be considerably less than an hour a session.

This is in line with an estimate we made a few weeks ago here that an average session on the service is about eight songs. Some stickiness!

But let’s assume that the author made an honest mistake. Perhaps he just doesn’t understand the differences between the way Ando Media measures usage (by counting IP addresses) and the way Arbitron measures listenership.

For the sake of discussion, let’s accept his numbers and ponder the implications of Pandora users spending no more time with the service than the average music listener spends with a local radio station.

Think about it: Pandora is a carefully crafted service that creates customized playlists based on the unique music tastes of its users. Its goal is to create the perfect listening experience:

Pandora recognizes and responds to each individual’s tastes. The result is a much more personalized radio experience-stations that play music you’ll love-and nothing else.

How can it be that listeners spend as much or more time with local radio stations that program to millions of people? How can that be?

Pandora apologists have a lot of explaining to do.

–Glenda Shrader Bos & Richard Harker of Harker Research