While Ando Media has been releasing Internet Radio numbers since 2005, the company has repeatedly made changes to how it gathers and tabulates its publicly reported data.
Because of the changes, we really only have seven months of somewhat comparable numbers. Despite the limited data, we’re already starting to see that the differences between broadcast and Internet are more than just the way the services are delivered.
We noted the latest change to Ando Media’s methods the other day. In the past the company counted listening spans of any length. Now Ando Media only counts listening spans less than twenty-four hours.
The latest change ought to have a negative impact on TSL, time spent listening--while providing more realistic numbers.
We looked at the top 10 services at the beginning of the year and compared their February TSL under the old rule, to their March TSL under the new rule.
Out of the top 10 services in Ando Media’s ranker, seven lost TSL in March. As the graph above shows, Pandora and Clear Channel lost the most TSL, 11% each.
Citadel, Radio One, and AccuRadio actually gained TSL despite the rule change.
The top 10 services started the year with a combined TSL of 18.7 hours, and by April the group’s TSL had fallen to 16.4 hours, a 12% decline.
Pandora stands out because when it first appeared in the Ando Media ranker in September of last year, it had the distinction of having the lowest TSL of any service in the ranker. In the seven months since, it has continued with that distinction, actually losing TSL every month since.
If you’ve followed the service’s growth, you know that registered users have skyrocketed. Total sessions have nearly tripled, and average sessions have more than doubled.
But what about that TSL?
Broadcast program directors understand that with most formats there is an inverse relationship between cume and TSL. Generally speaking, formats that attract large audiences (like Top 40) generally have lower TSL, while smaller cuming stations have longer TSL.
Is Pandora’s meteoric growth in users behind the declining TSL?
Were Pandora a top ranked mass-appeal broadcast station, we would expect low TSL. However, the service shares more characteristics with a small audience niche broadcast station with above-average TSL.
Pandora is a service that provides individually customized stations, in their words:
The result is a much more personalized radio experience stations that play music you'll love - and nothing else.
A service that creates personalized radio for each user ought to have high TSL. Consequently, the service should be able to rapidly grow users and see no decrease in TSL, especially given that its TSL is considerably smaller than the streams of broadcast groups.
A possible explanation is the growth of mobile. Mobile user are the fastest growing segment of Pandora’s user base. It stands to reason that TSL would decline as the proportion of mobile users increases.
While Pandora’s TSL was already declining before mobile users were counted, it doesn’t preclude mobile having a negative impact.
If mobile users spend less time with Pandora than computer-bound users, then the service’s TSL should continue to decline as smartphone penetration increases. That could become problematic for a service selling advertising.
Right now with everyone focused on Pandora’s rapid growth, the continuing decline in TSL hasn’t gotten much attention.
New well-marketed stations have always enjoyed a honeymoon period during which rapid growth concealed fundamental flaws. Over time the flaws invariably caught up with the station.
We suspect this is one rule that the Internet will not disprove.
HDRadioFarce, our "beef" is with those who have declared Pandora the online winner and successor to broadcast radio. While it is a spectacular marketing success (which we have noted in our posts), it remains to be seen whether the product is as transformational as its boosters claim. We think it is too early to tell.
Any of a number of services may ultimately prevail. Right now Pandora has all the attention, so we've written more about it, but we've also written about Jelli, and we intend to write about other services.
As far as fearing any of the services might be a threat to radio, we have cautioned broadcasters about the threat many times. No hidden agenda there. We believe market forces will do their job, and listeners will ultimately decide who wins.
Posted by: Richard Harker | June 24, 2010 at 11:04 AM
James, thanks for your observations. We do a similar calculation of "hours tuned" modeled after Canada's BBM. With the published numbers, it is probably the best way to compare broadcast to streams. We'll have a post on the subject soon.
Regarding Pandora's already low and declining TSL, all streams are measured the same way, so the fact that Pandora has lower TSL than all the broadcast streams is curious, and the fact that it is declining relative to the broadcast streams is even more curious. As we point out, we have a trend of only seven months, so things may change.
Posted by: Richard Harker | June 24, 2010 at 10:30 AM
If Pandora was doing it's job as hyped, the TSL ought to be greater than terrestrial radio.
In the end a jukebox is and always will be nothing more than a jukebox and there are thousands available on the Internet.
I have never understood what was meant to make Pandora so different, in fact there is no difference between it and all the rest of the jukebox services "redefining radio".
But from what I see Pandora is creating hype to gain investment, and we all know where that leads.
The most interesting thing about the whole Pandora story is how the "experts" have swallowed the bait.
Posted by: Terry Purvis | June 24, 2010 at 06:18 AM
You all seem to have a real beef with Pandora. Perhaps, it is because Pandora, Last.fm, Jango, Slacker, etc are such a threat to terrestrial and HD Radio? Your link is definately off my blog, now. Put yourselves on-par with Paragon and Jacobs.
Posted by: HDRadioFarce | June 24, 2010 at 12:32 AM
I looked at some of the numbers and it appears that the average session lasts around 45 minutes.
Take that time, and multiply say .75 by the number of session starts, and you get the true number of hours listened, I think I saw it go up slightly to .81 hour recently but the session starts went down slightly.
So the total time spent listening is somewhere in the neighborhood of 150 million hours based on the recent data.
Various factors can account for the low TSL, wireless smartphone listening is one, and it also shows how the TSL per session may not be the way to look at it. You get up in the morning and maybe have a station on for a few minutes, then go in and shower and shave, then have breakfast listening, then you turn that one off and go into the car, listen the however long it takes to go to work.
You may listen lobnger at work if the content filtering doesn't block the stream, then repeat the process again going home.
Someone said that the average listener, not session, listened 11 hours a month.
So what we really need is how many unique users were there, and how much time they actually listened. May be hard to come by.
Posted by: James Anderson | June 23, 2010 at 06:48 PM