This is the second in a series of “All Known Thought” mini-white papers that NPR President & CEO Vivian Schiller has asked me to write for internal and public radio system use. I’m posting it here with permission. The intro to this series was posted nearly a month ago, so I’m hoping that the next ones will follow every couple of weeks.
________________________
“The human brain must continue to frame the problems for the electronic machine to solve.” – David Sarnoff
David Sarnoff, founder of NBC and long-time head of RCA, was no Luddite. However, the ironic quote above notwithstanding, he did famously battle with Edwin Armstrong and Philo Farnsworth, each blessed with notable problem-solving human brains. Armstrong was the inventor of FM radio and Farnsworth was the inventor of the all-electronic television (yes, there was an earlier semi-mechanical version). In the first case, Sarnoff was protecting his company’s investments in AM radio and television, and in the second, he was protecting a competing television technology favored by RCA. Sarnoff’s masterstroke was to successfully lobby the FCC to move the original FM band from 42-50 to 88-108 MHz in 1945, obsolescing the half million FM receivers that were said to then exist and making AM safe from FM competition for another two or three decades.
Now we are entering an era when a newer technology, Internet Protocol (IP), is making it possible to not only transmit programming directly to listeners, but to do it across the globe as easily as across town. Wired and wireless – not just to computers but to vehicles, and bicyclists’ and joggers’ ear buds as well. This is becoming a lot like … radio. Our early instincts are, like Sarnoff’s, to protect our franchise.
This AKT will attempt to gauge the magnitude and viability of this new medium and see what it suggests about strategies going forward. All of these will be approximations, of course. We will cover both wired and wireless manifestations, but with an emphasis on the latter.
How is IP radio different?
The first difference was stated already and goes to the important problem of how to market your radio service in this environment. With a broadcast station your signal has limited geographical reach; but with IP radio, it’s global, and the scary part is, so is everyone else’s. But how does this play out in actual use? I can share two pre-smartphone data points, but remember, anecdotes aren’t a substitute for good research. In 2006, we at Northwest Public Radio (NWPR) tallied the geographical location of users to our live streams based on IP address. It showed one-third within our network’s coverage area, one-third in the Pacific Northwest but not in our coverage area, and one-third elsewhere on the globe. Later that year, I replicated this for WUSF in Tampa and found largely the same ratio. What explains this spread? Given that neither NWPR nor WUSF were promoting their services outside their coverage areas, perhaps it’s just the natural migration of listeners from the coverage area enhanced a bit with random people who stumble upon you during a search or aggregator usage.
A second difference is that your station on IP radio shows up outside your physical domain and doesn’t give listeners a predictable place (88-92 MHz in most cases) to find it. You have a marketing job if you want people to find your URL or your mobile application (app herein) from among the dozens of standalone public radio station apps or in the lists of stations that the many streaming aggregator apps are offering. Even with four decades in public radio behind me, I am bewildered by the huge array of call letters in aggregator apps (and in receivers like Livio’s NPR Radio). Call letter brands count locally and among your expatriates, but aren’t likely to be effective for discovery purposes.
There are also several economic differences. IP bandwidth is not free. Broadcasters don’t really own spectrum, though there are substantial property rights in that regard, and for most purposes we treat it as ownership. We almost always provide content for free (are there over-the-air pay TV channels left?). However, in the IP world, both the provider and the listener rent bandwidth based on how much of it they think they’ll use. Now, rather than just two parties, we have the two original parties, plus two internet providers plus the proprietors of the interconnecting “cloud.”
In broadcasting, barriers to entry are high and licenses are relatively scarce. At least on the commercial side, if one couldn’t make a station work economically, the “greater fool theory of broadcasting” (there’s always another buyer who thinks they can make it work) would usually kick in when the station sold. On the IP side, the entry barriers are low – no broadcast license, low capitalization for a basic entry (such as radioparadise.com), entrants are highly automated, and there’s usually no “greater fool” to bail you out of a failed entry.
To list another obvious difference, IP radio is inherently two-way and, for now, broadcasting is one-way (in a later AKT, I will write about how this could change). This enables the recipient to tell the source, “yes, I liked that song; play more like it.” This capability characterizes “pure plays” – on-line streaming services not retransmitting over-the-air broadcasters. Pandora is the most successful of these by far, but also, Slacker, Last.FM (owned by CBS Radio), Spotify (available now in Europe) and others are competing in this space. Stitcher is doing this for spoken word programming. Ando Media’s Webcast Metrics research shows Pandora beating the combined usage of the next two largest rankers, CBS Radio and Clear Channel, since February [source: www.andomedia.com/news.aspx].
What’s the current state of IP radio listening?
Mark Ramsey Research and VIP Research recently conducted a survey of 2,000 radio listeners in 22 markets and asked the following interesting question [source: http://tinyurl.com/23e4ssz]:
If tomorrow you could get Internet access from the dashboard of your car and you could listen to thousands of radio stations from all over the world through an Internet receiver on your dash as easy to use as your radio, would you…
a. Listen less to my local radio stations as I explore new ones online
b. Listen just as much to my local radio stations no matter what’s online
The response was that 34 percent would “listen less.” Another data point only, of course, but it suggests that the wireless impact will be significant. IP radio doesn’t need to take all of your listeners to be impactful in an era of small or non-existent margins, it only has to skim the cream off your listening. Arbitron and Edison Research’s The Infinite Dial 2009 [source: http://tinyurl.com/d84w22] provides some mixed data. It found that 27 percent of Americans listened to online radio in the last month and 14 percent of MP3 player users are spending less time with over-the-air radio; however, online radio users spend more time with radio overall, not less.
Here are some more data, triggered by a Bridge Ratings (bridgeratings.com) analysis this month of Ando’s data [source above] which reported that the share of “average active sessions” (AAS) went from about 35 to 49 percent “pure play” (Pandora, et al.) and from 65 to 51 percent for terrestrial broadcast streams from November 2009 to June 2010. That led to looking through a year (August 2009 to July 2010) of Ando data to see what was going on. It reports each month the top 20 rankers among its clients. The chart below [click for larger image] is a summary.
There were two pure-plays that were reported each month from November 2009 through July 2010, and 11 terrestrial broadcasters. The blue line on the chart is 97-98 percent Pandora; the rest is AccuRadio. Pandora alone more than doubled in size from August 2009 to July 2010 with mobile use growing substantially. The yellow line represents 11 broadcast groups, which seem to be holding their own over the 12 months. This line is about 75 percent Clear Channel Radio listening. Nearly all the variability in the line is due to the religious broadcaster (Educational Media Foundation). CBS Radio (magenta line) is distinguished from Clear Channel for three reasons: some of its streaming is via AOL, which is not reported by Ando; it owns pure-play Last.fm and it’s unclear whether that listening is rolled up here; and because CBS Radio has been shedding stations outside the major markets since 2006.
Pandora is now to its product category what Jello and Kleenex are to gelatin and tissue.
Most of this listening was for wired IP radio throughout the period, though wireless is coming on fast, especially for Pandora. Although most, if not all, the streamers listed have some sort of mobile presence through iPhone and Android apps, it’s still early in the mobile rollout, with all the limitations mentioned earlier.
It’s important not to imply an equivalency between Ando’s AAS and Arbitron radio AQH. Ando uses a 1-minute standard to record and doesn’t collect demographics, very likely leading to a higher number than radio AQH, which uses a much longer time standard and a 12+ demographic. That said, the top 20 Ando rankers – pure plays plus broadcast – have an AAS equivalent of less than two and a half percent of the national radio 12+ AQH in the same time period, which is less than half of one percent of the 12+ population. Separately, for the few stations whose streaming shows up in Arbitron data, it is similarly much lower than broadcast AQH. Radio listening still dominates.
As mobile apps become more widely used and integrated into vehicle sound systems where Arbitron meters can measure listening, wireless gets more bandwidth and better bandwidth management techniques are implemented – all of which will happen – broadcast station streaming should move off its apparent plateau. But there is a lot we can do to improve streaming performance (see later strategies discussion).
What is the state of IP radio technology?
We hear talk about building IP devices into the dashboard of a car, and also about tethering smartphones to the dashboard to use the internet capabilities of those phones. An early example of the latter is the Ford/Microsoft Sync collaboration on a dozen Ford models. This seems more viable, since consumers can use their existing smartphone subscription.
In truth, anyone with a smartphone and $5 for an audio cable or $30 for an FM modulator can already use their phone as an IP radio assuming one is willing to use the phone as a tuning device. Android and iPhone have mobile apps that access individual stations or, as in the case of aggregators such the NPR News app, access large lists of stations. Using the phone as a “tuner” is really not so difficult – and soon it will be even easier – when whatever you’re streaming on your phone as you step into the car automatically is played via your car’s sound system. You should also presume that location services will give you a list of local stations.
In other developments, the FCC opened up some spectrum for wireless IP last month and we will definitely see more of that over the next few years. Wireless devices will operate at faster speeds. Compression and transmission algorithms will continue to improve.
With the most common IP distribution protocol (“unicast”), the more successful you are in reaching listeners, the more bandwidth you will need to rent. This has led many to conclude that IP radio isn’t really scalable. Unicast requires a separate stream between a source and each recipient. Think of it as a bush – lots of branches emanating from the roots. But there’s an alternative protocol called “multicast,” which has been around for awhile but is new to wireless networks. Multicast is a confusing term for broadcasters because we use it to mean multiple programs over one channel. IP multicast, however, means one program data stream to multiple locations, branching along the way. Think of it more like a tree with a skinny trunk than a bush –branching happens far from the roots.
IP multicast protocols are now being added to wireless networks for multimedia transfers. For example, Qualcomm’s MediaFLO technology, although a troubled business, provides TV services (for now) to its own subscribers (FLO TV), and to AT&T and Sprint smartphones. The unicast scaling objection is obsolete. Other technologies that mitigate the limitations of unicast are so-called “edge strategies,” content delivery networks that move content to strategically placed hubs nearer the end user.
When multimedia streaming usage grows beyond the essentially hobby level it’s at today, the technology advances described above will be there to manage the load. The greater bandwidth needs driven by video will prevail and bring audio along for the ride on your mobile device.
The impact of wireless pricing
The recent announcement by many wireless carriers of the end of “all you can eat” data plans has many wondering if this will slow or end multimedia distribution on mobile devices. Tiered pricing and data limits are set-back in the short-term, but it’s really the way that the carriers have to price scarcity. Pricing is an elastic way of dealing with the near-term shortage of spectrum in a competitive environment. If you have data service from a wireless internet service provider (WISP) or satellite provider, you know that their restrictions are pretty severe.
When new bandwidth and greater use of IP multicast or other one-to-many technologies come on line at 4G+ speeds, as they surely will, competitive factors will force pricing back down, and it’s likely we will see the reappearance of all-you-can eat pricing. In the medium and long term, don’t consider pricing, like technology, to be an obstacle to the growth of IP radio.
Strategies for making it work for us
Where is public radio in the Ando Media top 20 rankers? In total, we don't know. The 12 months examined include only three instances: WXPN broke into the top 20 in October 2009 with an AAS of 1,800, and WNYC did in June and July 2010 with an AAS of 4,000. There’s probably not a market in the country where the public radio station isn’t consistently in the top 20, so this seems to be some sort of index of underachievement. Maybe or maybe not. Public radio isn’t showing in the top 20 because (1) we don’t aggregate streaming metrics as do the broadcast groups who are listed and (2) frankly, too many of us seem to be doing all we can do to make it hard for our listeners to find our streams.
In the course of my work I visit a couple of station web sites a week and – tough love warning – one often really has to work to find the “listen now” button. For joint licensees, radio shows up as a tab among three or four TV tabs, so you need to get past that hurdle. Click the tab and then scan around the page for the listen button. Oh, there it is in a list in the lower right. Then you get a choice of maybe two or three program services and two or three different streaming technologies. If you don’t have time to call up your 18 year old for help, you might give up.
Contrast that with the web sites of the Clear Channel stations or other broadcast groups. Radio is not hiding behind a television page. The “listen now” button is prominent (often top left), and you don’t get a range of confusing choices for streaming technology. We’re radio stations, so make it easy to find radio.
Audiences are built through discovery and maintained through relationships. Public radio does relationships well and social media tools are helping that. On the radio, discovery often happens with the twist of a knob. Discovery also happens with good marketing, but most stations don’t have enough time or money to do marketing right, which is exacerbated by the international reach of IP radio. In your market, you can push your stream on air. The emerging multi-platform world does need a discovery strategy, though, and best practices say it should be through what we might call a “be-everywhere” strategy – “distributed distribution.”
Cross-linking online improves discovery both directly and, by improving search ranking, indirectly. Many university stations, for example, will find it helpful to cross-link on university web sites and apps. Individual station apps for the iPhone or Android are good (though listings ca n be bewildering), but do consider building and joining collective apps. For example, KPLU might have its own multi-platform app, and also be listed on a jazz station app, a Seattle-Tacoma radio app, a public radio of the Northwest app, or on an app for Lute Athletics (just trying to see if you’re paying attention on the last one). Build partnerships with non-profit organizations and bloggers in your community and get them to carry links to your streams and content archives.
If public radio is going to remain competitive, now is not too soon to act strategically. There will be more on this topic in the later “AKT” concerning developments in radio.
__________________
Update: For more on this topic, I’d suggest you check out Skip Pizzi’s valuable paper, The Mobile Internet: A Replacement for Radio? It was written for the Station Resource Group and is available on its web site. Link. Also check out "Fear and Loathing about mobile streaming" on Stephen Hill's Spatial Relations blog. Link.
--Dennis Haarsager
Dennis,
This paper does valuable work extinguishing some of the lingering objections to fully embracing digital operations, particularly wrt to mobile bandwidth issues and automobile operation of IP "radios."
On a bad day I see reactions to these 'issues' as more foot-dragging by public media personnel who are just not comfortable entering a new medium with a substantial learning curve, new paradigms and new technologies to be mastered.
Comments like Steve Rubin's are sadly typical -- he points to a "problem" that in 2010 is ironically the opposite of what he describes. Sure, some retrograde webcasters are still using inferior quality low bit rate Real Audio or MP3 streams that they encoded ten years ago. But the typical 64k AAC stream today actually exceeds the audio quality of FM. It has a full 15khz bandwidth, and a noise floor that is at least 30db quieter than FM under optimum conditions. 3G mobile data bandwidth easily handles 64k stereo streams of excellent quality. This is no longer an issue.
There's one point in your piece that requires comment because recent technical developments make it less of an issue. You say "With the most common IP distribution protocol (“unicast”), the more successful you are in reaching listeners, the more bandwidth you will need to rent."
This is true for normal hardware servers located on private networks or on dedicated Content Delivery Networks. But the latest generation of "cloud" CDNs like Amazon's new CloudFront do not require fixed bandwidth allocations ordered in advance. You pay only for what you actually use, and the services scale on-demand as required.
Cloud hosting is a fundamental increase in infrastructure efficiency which has already started to drive the cost of basic operation down even further. In a future paper you will be saying "don't consider pricing of bandwidth to be an obstacle."
Your final point that "now is not too soon to act strategically" is a fascinating example of the kind of polite, qualified rhetoric that has characterized public radio discussions about digital for the last ten years.
The time to act strategically was five years ago, when you and other public radio progressives were pushing for a truly strategic response to the challenges and opportunities inherent in global digital networks.
It did not happen then. It is happening now by default in some of NPR's digital initiatives, and particularly in the work that is being done by NPR, PRI, APM and PBS to build a "Public Media Platform" -- but in a way that is limited and will not be nearly comprehensive enough to change the way business is done in the system, or to improve the position of public radio vis-a-vis new competitors like Pandora and other content aggregators.
Individual players on the national or local levels of the system may have their own strategies, but I'm afraid that alluding to any kind of unified strategy for public radio or public media as a whole is wishful thinking. The diverse federation we have is handicapped by "channel conflict" and ill suited for comprehensive action, which would require a unified public offering and major changes to the underlying business methods.
Let's face it: as a network, we don't do strategy.
Posted by: Stephen Hill | Sunday, 24 October 2010 at 06:35
Public radio will also need to differentiate itself more clearly as Dennis suggests. We are already noticing a certain "digital fatigue" among some demographics, i.e., the embarrassment of riches syndrome caused by the proliferation of apps accentuating the divide between the early adapters and the rest. Much of this remains anecdotal, so more research is needed. But an overabundance of offerings may simply confuse those whose goal is community (such as public radio can provide) as opposed to those who are fascinated with the latest set of neat toys.
Posted by: Jeffrey Dvorkin | Saturday, 16 October 2010 at 08:06
What this AKT paper does not discuss: The audio quality of IP radio. Some of it is fairly poor -- far less than AM -- and this surely affects the adoption rate of IP radio. Listeners, either consciously or sub-consciously, make note of the difference and adjust their habits to meet their preference (some don't notice or care, while others expect good high fidelity). I suspect this issue will be solved in time as broadband gets broader. But for now, I would consider it an obstacle to the advancement of this medium.
Posted by: Steve Rubin | Friday, 15 October 2010 at 16:25