Not everyone is happy with the radio deal I mentioned in my last post. Techdirt points to Michael Robertson who tries to make the case that even with this new deal it's impossible for Internet Radio to survive. His last paragraph concludes...
The economics don't lie. Experienced radio man Bob Bellin says, "No start up webcasting business can give up more than 25% of their revenue and make a profit." Bill Goldsmith of Radio Paradise says that long term viability requires a payment of no more than 10-12% of total revenues for both publishers and record label royalties. The new rates, while lower, are nowhere close to being sustainable. While I've used Pandora as a reference point to illustrate the situation, the same could be written about Slacker, Last.FM, or any webcasting company. Net radio is going to die a slow death in the United States.
Wow. The CEO of Pandora seemed pretty happy for a guy whose company is going to die a slow death. But skepticism is only worthwhile if I investigate so I decided to look at the numbers.
To start we need to understand what we're talking about here. To get there let me quote Rob Pegoraro of the Washington Post who does a nice job of summarizing the new rates. Here it is...
* Large Webcasters--defined as those who rake in more than $1.25 million in annual revenues--will pay either 25 percent of those revenues or a per-performance rate in the neighborhood of a tenth of a cent, whichever is greater. This deal runs through 2015.
* Small Webcasters--those raking in less than $1.25 million a year in revenues and fewer than 8 to 10 million listener-hours a month of broadcasts, depending on the year--can choose between paying a percentage of their revenues (12 percent of the first $250,000, then 14 percent of everything on top of that) or their expenses (7 percent). These options run through 2014.
And since I've decided to use Pandora as my example case I turned to a 2007 article from PCMAG.com which gives that average listener rate...
Pandora.com is a site that streams music to listeners, based upon their preferences. On average, each Pandora listener there tunes into to Pandora.com around 18 hours a week, listening to an average of 15 stations each. This was a Pandora.com-positive group, to say the least.
Now that we have all those numbers lets take a look at how they come together. Please remember this is a vast oversimplification. I'm just trying to prove the song costs aren't enough to sink the company so I'm leaving out all the other business related factors such as IT staff, programmers, sales, etc...
With that said I am assuming an average song length of 3.5 minutes. Combine that with 18 hours a week of listening per Pandora user you get about 309 songs a week. If Pandora is then paying $.0014 per song you get $.43 per user per week for Royalties.
Now Internet ad rates are usually measured in CPM (Cost Per Thousand(roman numeral M)) so lets look at a thousand users. We already know the royalties cost will be about $430 per week and that's on the 309 songs we spoke about. But Pandora shows an ad about every 3 songs (I'm told, I use Pandora One) meaning that's 103 ads being shown putting their CPM to cover royalties at about $4.17 per Ad ($430 per 1000 users / 103 ads shown to all 1000 users = $4.17 per ad). To put those numbers in perspective understand sites like TechCrunch and ReadWriteWeb were charging around $36 per CPM before the downturn and Internet Radio tends to have higher rates (plus Pandora has double the unique visitors of Techcrunch and over four times that of ReadWriteWeb).
But what about bandwidth? Bandwidth is an expense other types of Radio don't have to deal with so just to be fair let's look at it. Pandora streams music at 128Kbps (to non-Pandora One users) or 16KB per second. That makes the total footprint of an average user around 1,036,800 KB per week (16KB per second * 60 seconds per minute * 60 minutes per hour * 18 hours per week). Now at work I have a 45 mbps fiber line and we pay about $1,000 per week for. It delivers 5,625 KB per second at a rate of $.0017 per second. Multiply that by 184 (1,036,800 KB per week / 5,625 KB per second)and you get around $.31 per user per week for Pandora or $310 per 1,000 users
(Keep in mind these numbers are way high. An organization buying in bulk could get much better prices for bandwidth not to mention content delivery networks that could knock the price down further)
Again that's on about 309 songs per week making a cost of about $1 per ad or $5.17 total CPM needed to pay for the actual songs being served (Royalties and Bandwidth).
As I said up front, this isn't meant to represent the money it takes to run Pandora. For that you'd need to look at all the maintenance costs, employee costs, sales staff, etc... But what I am trying to do is to debunk the idea that no startup radio station can survive under this royalties pricing structure. It's simply not true. The truth is royalties aren't THAT big a piece of the pie anymore. Certainly not enough to mean the difference between life and death for a service like Pandora.
Which is why both sides seem happy with this new agreement.
Debunking The Other Numbers... It didn't fit in the framework of the above post but I wanted to address one other section of Mr. Robertson's post. He says...
As an example, Pandora estimates they will do 1 billion hours of streaming in 2009 which requires $17 million in royalties. They estimate they will do $40 million in revenue which equates a payment to the major labels of 42.5% of their total revenues. No business can operate with such an enormous financial obligation.
Has anyone ever worked in a retail business that sells any kind of physical goods? If so you'd know the manufacturer easily takes more than 42.5% of the revenue from products. I worked in a computer store as a kid and we'd regularly only get 5% to 10% off hardware like Video Cards, Sound Cards, etc... To back that up consider Grocery stores which have notoriously thin margins (profit margins that hoover around 3%).
Also note the 1 billion hours number suggests a much lower average time per listener than what we used above. Pandora said they had 20 million registered users in Dec. 2008. 1 billion hours of streaming / 20 million users = 50 hours per YEAR. Compare that to the 936 hours assumption we made above (aka 18 hours a week) and the picture looks even better for Pandora.
Finally, it should be said that I know a lot of web companies that would kill for $23 million in revenues this year.
Addendum: Putting aside the Pandora example a lot is being said about the $25,000 fee being imposed on all Internet Radio stations. The argument against it says that fee will kill a lot of small Internet Radio stations. To that I'd point out two things...
1. The $25,000 fee is not an additional fee, it's due up front and then applied to the station's royalty payments
2. The $25,000 equates to only about 1,100 regular listeners. Let me do the math. $25,000 per year equals $2,083 per month which in turn equals 1,488,095 song plays (at .0014 per song). That in turn equals 343,406 song plays per week. Using the 18 hour per week and the 3.5 minute per song averages that number of songs equates to 1,112 regular listerners.
(Note that small stations don't even have to pay a per song fee but instead pay only a percentage of their revenue or their expenses)
So basically this $25,000 is more of a "barrier to entry fee" in that it's designed to keep the labels from doing a lot of work on companies who aren't serious enough to have a little funding. I don't necessarily agree with that strategy but it's a perfectly reasonable one and certainly not one designed to kill Internet Radio