Note 1: I’ve been chipping away at this post for about eight twelve months off and on. I finally managed to bring it in for a landing yesterday night. File it under #slowblogging and #longreads.
Sam Lefebvre’s recent cover story on Pandora in the East Bay Express overlaps some of what I discuss here, and it’s well worth reading. That said, as our respective focuses aren’t the same in every regard, I’m hopeful that this post will serve as a useful compliment to Lefebrvre’s piece.
Note 2: This post was updated on December 8, 2014 with a bit of new content based on some comments I received when it first went up. I expect there will probably be more substantive updates as time goes by, and if that happens, I’ll let you know right here.
Note 3: If you’d rather read this post off-line, feel free to download a copy of it in pdf, MOBI, or EPUB format. Apologies in advance for any formatting anomalies. I’m not yet a wizard at doing ebook format conversions.
***
Of all the dedicated digital music streaming services to come on-line since 2000, none has drawn the ire of the music industry quite like Pandora.
On its face, this may seem peculiar. After all, Pandora is a legal, royalty paying service. It has passed 250 million registered users in the U.S. Moreover, despite loud protestations to the contrary, there’s a credible argument to be made that, on a per listener basis, Pandora actually pays more royalties per spin to both songwriters and master rights-holders than does terrestrial radio.
So shouldn’t Pandora be a win for record labels and music publishers? Aren’t they being irrational in hating Pandora more than all the other services?
Succinctly, no. Pandora has been more disruptive of established music industry practices than any other major legal streaming music service. So the music industry has some very real reasons–both financial and aesthetic–to hate Pandora.
This post is a deep dive into five of those reasons. I’ve tried to keep my thoughts in plain English. But fair warning: I do get into some technical legal stuff towards the end. You can’t grasp the full picture without it.
1. It doesn’t matter whether Pandora is great at music discovery, because the sort of music discovery Pandora provides is actually a relatively small part of traditional major label music promotion.
In the music industry, consumers are sometimes divided into two general categories: (1) lean-forward listeners and (2) lean-back listeners.
Lean-forward listeners are the super music fans. They like to be actively engaged in their listening experience. They are more likely to read music magazines, blogs, etc. They like being the person in their group who turns other people on to new songs and obscure oldies. When something new gets released, they want it ASAP. Often, they are younger. But there are plenty of older folks in this cohort too.
In the 1980s and 1990s, lean-forward listeners were the folks who read fanzines, went to lots of rock shows, made cassette mix tapes for themselves and their friends, and spent hours shopping for obscure records. These days, they’d be the people who can happily spend a few hours putting together an iTunes or Spotify playlist.
Although lean-forward listeners are a smaller group than lean-back listeners, they are still very important, because they are typically the early adopters. Without them, it would be harder to put new music out in the world, because lean-forward listeners are the ones most willing to receive it.
Lean-back listeners encompass most other music consumers. They are less engaged with music. While they like having music around, they aren’t interested in spending hours making mix tapes or constructing playlists. They also tend to be less adventurous listeners. It takes more time and exposure for lean-back listeners to embrace new music, and they are less likely to embrace music that is different from the music they already like.
Historically, lean-back listeners have been heavy users of old-fashioned terrestrial radio. They have willingly listened to advertisements each hour in exchange for a curated stream of free music (on average 11 songs an hour). This soundtrack has been their background while driving, cooking, studying, etc. And from time to time, when a song really catches their fancy, they might just be motivated to make one of their infrequent trips to the record store to buy it (or more likely in 2014, they might download the track or the album from iTunes).
Pandora has been particularly popular with the lean-back listener. By all accounts, music discovery is a major reason why. At its best, Pandora is like a skilled, album-oriented, FM radio DJ from the 1970s, mixing familiar music with other music that is unfamiliar but stylistically related. By listening to a Pandora station, people theoretically listen to music they would not have heard otherwise.
In some cases, Pandora listeners evidently like this unfamiliar music enough to take some action about it (e.g., go out and buy the CD or download the mp3s). According to a recent study by the Country Music Association, these Pandora listeners may well be more likely than terrestrial radio listeners to take some action on a new song they hear on the service. People don’t have to fiddle around with Pandora very much to get a usable stream of music, which many people appreciate. And since Pandora has a free ad-supported version, the cost of entry is low for consumers. Collectively, these attributes make Pandora a paradigmatic example of a lean-back listening experience.
When Pandora defends itself against attacks by record labels, it typically cites its music discovery value as one of its primary virtues, arguing that it is like terrestrial radio (or maybe even better): a complementary tool for promoting and selling recorded music to the general public. Therefore, it’s a win/win for everybody.
To the average consumer, this may seem like a good argument. It may even be a compelling argument for certain grass-roots, independent artists. For example, cellist Zoë Keating recently indicated on the blog Hypebot that Pandora has provided her with a significant promotional benefit (“People discover me there….[i]n my reckoning, for me, Pandora = valuable promo.”).
But major labels see things differently. From their perspective, music discovery of the sort provided by Pandora may be one aspect of music promotion, but it is by no means the only aspect (or even the most important aspect).
Here’s why: Throughout most of their existence, major labels have made their best money bundling groups of recorded songs together into albums. Typically, they have used a strategy that some have called “Mixed-Leader” bundling: a so-called “leader product” is made available as a single and then also combined into a larger compilation with 6–12 non-leader products (the “album”).
Labels settled on this approach because the marginal cost of an album isn’t that much higher than selling one or two songs at a time, and it is usually easier to sell 100 people something for $10 than it is to sell 1000 people something for $1. Moreover, if the $10 thing has 10 songs, including one or two that many people want, it makes it easier to up-sell them a compilation (i.e., you can have the two songs you want for $2.25 or you can get those two songs plus 10 more for $10).
Albums sold by major labels are generally either new releases or so-called back-catalog items (i.e., previous releases that continue to sell well enough that it is worthwhile to keep them in print).
Most of a label’s new releases fail, but a successful blockbuster generates the most income for the label in the shortest amount of time. Conversely, individual back-catalog items typically generate less money per quarter than a popular new release, but collectively back-catalog items can be a steady and persistent source of income with profit margins that keep getting better as upfront production and marketing costs are amortized over time.
Some blockbusters also end up being really strong catalog sellers. Other blockbusters sell a lot in the first year or so and then burn out. Conversely, some new releases don’t initially sell huge numbers but continue to sell steadily over many years, amassing very respectable sales. In the realm of rock and roll music, the Ramones and the Velvet Underground are two examples of that slow and steady phenomenon.
If a record company can be compared to a specialized investment bank, then new releases are the high risk/high yield growth stocks in its portfolio. Back-catalog items, on the other hand, are the lower risk municipal bonds with a steady and predictable return. But since not all blockbusters are strong catalog sellers, labels sometimes release records without blockbuster potential, in hopes that they will end up as strong catalog sellers.
Labels may also view certain releases as an investment in developing an artist who seems to have the long-term potential to deliver a blockbuster, accepting short-term losses in hopes of building a strong relationship between listeners and the artist and slowly achieving better sales over time. Indeed, a slowly growing base of passionate fans has served as the foundation of many strong selling back-catalog items. Whether it’s a release with slow sales but loads of critical acclaim or a word-of-mouth champion, some of the most beloved records of the last 40 years fit into this category.
Prior to the advent of digital music and file sharing, physical warehouse and retail space constrained the size of a label’s back catalog. Because there was only so much room in the warehouse, labels couldn’t stock everything, and catalog items that sold poorly were simply taken out of print.
This limited the universe of music that the average consumer could buy through mainstream retail outlets, and it mostly confined out-of-print items to the realm of collectors and super-fans (a relatively small subset of people within the already small lean-forward listener group). In a sense, it was a form of institutionalized forgetting, and it helped keep consumers focused on the “new.”
From a promotional standpoint, the assumption was that consumers already knew about back-catalog items or would discover them via music magazines like Mojo or listening to terrestrial radio stations with backward looking formats (oldies, classic rock, classic jazz, classical, etc.). Aside from occasional re-issue initiatives, which would get a little extra promotional attention, record labels focused most of their promotional efforts at any given time on a small universe of new releases.
Pandora makes it difficult to orient promotion in this way, because labels have very little input into most of the factors that put a particular song into a Pandora user’s music stream. Instead, the content of the Pandora music stream is initiated on the demand side, when a user types in an artist, song, album, or genre. From there, the vaunted Pandora algorithm takes over and creates the station’s playlist. Then, it’s the luck of the draw what songs that user will hear.
If the Pandora user has typed in the right combo of songs or artists, maybe that user will hear a label’s new release, but it’s much harder to determine how promotional dollars affect the outcome of that transaction.
As Peter Drucker has famously stated, “What gets measured gets managed.” Or to put that a different way, if you can’t measure something, it’s much more difficult to manage it.
All this means that Pandora presents a huge promotional problem for major labels, because a central goal of traditional, major label, music promotion is to focus demand on specific releases in a relatively compressed period of time. To do this, one must move beyond the lean-forward listener market and break through to the much larger lean-back listener market. It is nearly impossible to achieve blockbuster sales of a release without getting lean-back listeners on-board.
To the extent that labels invest large sums of money funding artist recordings and promoting them, they need focused demand on at least some of these recordings each year to recoup all of the money they spend, make a profit, and live to fight another day.
Typically, a few blockbusters subsidize the many releases that never recoup their costs (or in the case of slow-but-steady sellers in the back catalog, releases that take a number of years to achieve profitability). It was that way in the old days, and it remains that way today. The harder it is to focus demand, the harder it is to achieve a blockbuster, recoup costs, and remain solvent (let alone profitable).
So even though the overarching project of Zoë Keating and a major label are nominally the same (find people to buy their bundles of recorded music), the differing scale of their respective projects leads them to very different conclusions about the value of Pandora as a promotional tool.
For an artist like Keating, the sheer size of Pandora and its user-base can have good promotional value, because Pandora has a flatter hierarchy than mainstream terrestrial radio. As a result, there’s a better chance that Keating’s music may end up in somebody’s Pandora station sandwiched between 3 or 4 tracks that are more familiar to the listener.
Even though Keating has very little control of where and when her music gets played on Pandora, she still benefits from the mere fact that, relative to the scale of her project, a lot of people are hearing her music who otherwise would not have. If there was no Pandora, somebody like Keating would have almost no chance at all of reaching lean-back listeners, because it’s one of her only points of access to this group.
But if you are a major record label in the blockbuster business, none of these benefits are worth much. Unlike Zoë Keating, labels already have plenty of platform access in places where they know their music will be broadcast to a potential audience of lean-back listeners.
And when the goal is to scale things up as quickly as possible into millions of sales, you can’t just wait around and hope that eventually people will find a new release on Pandora. Indeed, Pandora mostly just steals people’s attention away from other media platforms that have already shown their value in helping labels reach lean-back listeners and achieve blockbuster scale.
Chief among these other media is terrestrial radio.
All media have a push and a pull aspect: Users either affirmatively pull the content of the medium to themselves or they elect to give up control and have content pushed to them without further user intervention. But the blend between push and pull varies from medium to medium.
In terrestrial radio, the pull happens when the user tunes into the station (for what it’s worth, the same thing is more or less true with digital satellite radio services like Sirius). After this choice is made, it’s all push until the user chooses to change the station or turn off the radio.
On Pandora, the pull happens in three ways: (1) when the user navigates to the Pandora website or opens its mobile app and uses Pandora’s tools to build a station or a super-station that shuffles through more than one of the user’s existing stations; (2) when the user chooses to listen to one of its stations, superstations, or one of Pandora’s genre-based stations; and (3) when a user hits the thumbs down button or skips a song.
Compared to Pandora, terrestrial radio listeners cede most content control to the broadcaster. To the extent they possess a pull filter at all, it is relatively coarse and non-interactive. When you change the station, there’s no guaranty what you’re going to find on the next station. Will it be a commercial? A song you like even less? You never know.
The more you change stations, the more it interrupts any sort of musical flow. Otherwise, changing the station has only a very indirect and delayed effect on what any station broadcasts, because programming decisions are based on a statistical ratings survey (i.e., an aggregated picture of what all listeners, on average, are doing).
So while many people do regularly station jump, it’s often a pretty frustrating and low value proposition. As a result, plenty of people don’t jump around very much. Instead, they stay on one station, suffering through the commercials and songs they don’t like. Radio programmers do their best to keep this group of people as large as possible, which usually means trying to offend as few of them as possible.
At the end of the day, this is the core value of terrestrial radio to both advertisers and labels: Even though plenty of people jump from station to station, on average, a large and statistically predictable number of people are always listening to each terrestrial radio station’s broadcast at any given time.
Pandora, of course, doesn’t work this way. By comparison, it seems to offer listeners an enhanced value proposition (but see #3 below). Even though it is considered “non-interactive” streaming radio for regulatory purposes, it gives users much more fine-grained pull filters.
A user can’t tell Pandora to play a certain song on demand. That’s one reason it is considered “non-interactive” under the law. But each user of Pandora’s free version does get 6 skips per hour per station, and up to 24 total skips per day across all stations. And if a user has some knowledge of artists, music genres, etc., he or she can pretty predictably build a mix that eliminates a number of the more unpopular aspects of mainstream commercial radio stations.
Moreover, by clicking the thumbs down button when songs come up that the user doesn’t like, the Pandora station is further fine-tuned over time. In other words, when a Pandora listener engages one of Pandora’s pull filters, it has an immediate, discernable impact on what happens to that listener’s music stream.
I don’t have data on how often the average Pandora listener uses Pandora’s skip function, but the average Spotify listener apparently skips 14.65 songs per hour. So it stands to reason that many users of Pandora’s free service routinely skip 6 songs per hour, and that they would skip even more if it was allowed.
By its very nature, then, Pandora makes it harder for large record companies to predictably focus demand, especially on lean-back listeners. It replaces the numerous smaller bundles of music sold by record labels with a new, singular, mega-bundle encompassing millions of songs.
Instead of buying ten songs to get the two that they like, consumers buy (or are given) access to a few million songs to get to the thousands of songs they might be interested in. From there, it’s a bit of a crap-shoot exactly what songs consumers will hear, because it’s in the hands of the algorithm.
Consequently, the more people use Pandora as their primary means of listening to recorded music, the harder it is for labels to predictably reach lean-back listeners with new releases, focus demand, achieve blockbusters, and make consistent money.
2. Hookers and blow don’t mean much to the Pandora algorithm.
Prior to the advent of digital music, large major labels had three significant advantages over all competitors: (1) they controlled distribution of vinyl records, cassettes, and compact discs, which gave them privileged access to brick and mortar retailers, and insured that their releases were available for purchase everywhere; (2) they had privileged access to terrestrial radio, the single most powerful tool for focusing demand on particular releases (especially among lean-back listeners); and (3) they had privileged access to television and the press, two other very important demand-focusing tools.
All of the institutions I’ve just mentioned are political systems filled with human decision-makers, just like Congress, a state legislature, or a city council. Much of traditional music promotion is the political process of influencing these decision-makers.
As in the legislative realm, all manner of strategies have been deployed in the service of this goal. It could be as simple as building a relationship with a record store owner or radio station programmer in hopes that this personal connection will eventually gain a release some extra attention. Or it could be a more nefarious quid pro quo (e.g., an outright cash payment to a radio programmer or the provision of indirect benefits like free meals, drinks, or hookers and blow).
Regardless of what gets exchanged, these promotional activities are all premised on one idea: If you can influence the right people to get behind your release, it will significantly amplify the likelihood of your release reaching lots of people.
Historically, this has been a relatively efficient process, because the universe of “right people” hasn’t been so large that a well-funded label can’t touch most of these people when it needs to. Moreover, if people at your company are engaging in this sort of lobbying all the time, they develop strong relationships and privileged access to all the right people.
So it hasn’t been that big of a mystery whose back needs to be scratched. That doesn’t mean a label can’t botch the lobbying campaign, but there typically hasn’t been much mystery about which parties the lobbying campaign needed to be directed towards.
The politics of this system have also been relatively decentralized. Even in this era of media consolidation, there remain a lot of different people that one could potentially influence to change the course of promoting a release on radio, television, and in the press. If a couple of the right people can be convinced to get behind a release, they may be sufficient kindling to light a bigger fire for it and convince other people to get behind it. Satellite radio services like Sirius, while similar to Pandora in some respects, have this human aspect too.
Even though many of these institutions have become very numbers and metrics driven, there are still humans making programming decisions in these institutions. As a result, when all else fails, if you have enough money to spend, there’s probably somebody who will grant you the access you seek (whether that’s radio spins, a cover story in a popular magazine, a television appearance, getting booked on a good tour, etc.).
By comparison, Pandora seems relatively insusceptible to the sort of politicking in which large labels have historically been adept. Who exactly would you try to influence at Pandora to ensure your release gets heard by more people? A Pandora executive? The people who categorize the music? One of the software developers?
Could getting somebody to change the algorithm really increase the odds that your song will be heard by more people? Maybe.
Could a label pay Pandora to give a release privileged placement in different streams? Well, recent news is that this is possible in a manner of speaking. Apparently, Pandora recently did a direct licensing deal with digital rights organization Merlin on behalf of 120,000 independent artists, labels, and distributors. Under the deal, Merlin accepted a lower royalty rate from Pandora than the compulsory royalty rate being requested SoundExchange.
In exchange for taking this lower rate, Pandora agreed to “steer” more music administered by Merlin into Pandora users’ streams. But the more of this music that it “steers” into listener’s streams, the lower the royalty rate it has to pay to Merlin. As the Trichordist blog has pointed out, the deal may also have some other downsides for artists, providing Merlin’s member labels with the legal means to access the artist’s share of the streaming royalties typically administered by SoundExchange (more info on these royalties and what SoundExchange does in #4 below).
But even with this deal in place, Pandora remains mostly a black box: Songs are submitted, vetted by Pandora employees, and then they are put in the hands of the Pandora algorithm.
Whether it’s a major label release or one of Zoë Keating’s albums, they each get coded with metadata and run through the Pandora algorithm. Even if songs subject to the Merlin deal get an algorithmic boost as a result of that deal, it’s still mostly in the hands of the users. If a user keys in a set of variables that meshes with the metadata around Zoë Keating’s song, it will play on the user’s station. The same is true of the major label new release.
But there’s not a clear A+B = C type of relationship, where a label applies resources to a song (e.g., via a “radio promotion” campaign) and as a result it gets a minimum number of radio spins in certain markets (i.e., a pretty solid guaranty that a certain number of people will at least hear the song a few times).
Even with preferences of the sort built into the Merlin deal, Pandora still doesn’t consistently and predictably expose users to a slate of favored major label new releases. Instead, users act on Pandora and then get exposed to whatever songs Pandora serves up. The process isn’t exactly random. But from the standpoint of a label promoting a release, it’s way more random than the label would like it to be.
The end result is this: Pandora is a very different animal than terrestrial radio, despite its protestations to the contrary. To a large extent, Pandora does not exist to meet the needs of large record labels trying to promote new releases, because it operates within a different incentive framework than terrestrial radio.
Unlike Pandora, terrestrial radio pays no performance royalty on sound recording copyrights (it only pays a performance royalty on the underlying musical composition copyright of each record it plays). From a logical standpoint, this framework makes very little sense. In most countries around the world, broadcasters pay a performance royalty for both the sound recording copyright and the musical composition copyright. But not in the United States.
The historical justification for this carve-out is that record companies receive promotional value in lieu of cash compensation when their records are played on terrestrial radio. So as a matter of policy, sound recording copyright owners aren’t granted a performance royalty (that’s a fancy way of saying that this is the compromise to which everybody in the legislature was willing to agree).
To the extent record companies can demonstrate that terrestrial radio stations are not providing promotional value on their releases, they are better positioned to argue that Congress should institute a sound recording performance royalty on terrestrial radio plays.
Terrestrial radio stations don’t want to pay an additional royalty. So it’s in their interest to help large labels promote their releases at least enough that there is some “promotional value” they can point to when major labels periodically lobby for a terrestrial radio performance royalty on sound recordings (See e.g. the #irespectmusic campaign, which seems to be the most recent industry push for sound recording royalties on terrestrial radio spins).
In addition, terrestrial radio stations have gradually learned that limiting variety serves their economic interests as well. Most terrestrial stations make their money from selling advertising. The more people who consistently listen to a station, the more a station can charge for this advertising.
Radio advertising rates get set based on ratings determined by Nielsen/Arbitron, the company that tracks radio listenership numbers. Around six years ago, Nielsen/Arbitron, started using a device called the Portable People Meter to track what stations their survey subjects are listening to. And as the Wall Street Journal recently explained, these devices yield much more granular data than the listening logs that survey subjects had previously filled out.
This new ratings data confirmed what many radio programmers already believed: Average listeners are much more likely to change the station when they hear an unfamiliar song. As a result, terrestrial broadcasters have increasingly narrowed the number of tracks that they keep in regular rotation, and in the process expanded their listenership.
The top ten songs of 2013 were played close to twice as much as the top ten songs of 2003. For example, in 2013, Robin Thicke’s song “Blurred Lines” was the most played song on terrestrial radio. It received 749,633 plays. Conversely, the most played song of 2003, “When I’m Gone” by 3 Doors Down, received 442,160 plays.
As a non-interactive streaming radio service, Pandora is required to pay a royalty to the sound recording copyright owner each time it plays one of its sound recordings (usually the sound recording copyright holder is going to be a record label in this scenario). But unlike interactive streaming services (e.g., Spotify, Rdio, Rhapsody, etc.), Pandora does not have to negotiate its license terms and royalty rates directly with the major labels.
Instead, Pandora licenses the music it plays pursuant to a statutory licensing scheme and an antitrust consent decree (more on these things in #4 below). So it isn’t beholden to large record labels in the same way as an interactive streaming service. Nevertheless, it still provides a user experience that is more similar to these interactive services than to what terrestrial radio offers. Therefore, to the extent it negotiates with big labels and publishers at all, it does so on a very different terrain.
The end result of these different rules is that the record company is not king on Pandora. The user is king, and Pandora’s primary concern is the user experience. Pandora will only make money if it can achieve a very high adoption rate among the general public. At present, it cares about adoption above all else, irrespective of whatever lip service it pays to its value as a promotional tool for record labels.
Because Pandora encompasses a multiplicity of terrestrial radio formats, including formats geared toward highlighting new releases and formats focused on playing familiar and unfamiliar back-catalog items, the distinction between an unfamiliar song and a “new release” is, at best, blurry on Pandora. This makes Pandora particularly infertile ground for major label record companies looking to direct large masses of people toward particular new releases. New releases are no longer just competing against other new releases with a selected slate of old standbys mixed in between. Now, they compete for attention with the entire history of recorded music.
If no new recorded music was ever released again, one wonders whether Pandora’s business model would be adversely affected at all. Many members of the listening public (especially folks over 35), are busy, don’t care that much about music, and don’t have a lot of time and energy to spend on it (the very definition of lean-back listeners). The universe of music these folks want to hear tends to be relatively small and familiar, and they simply don’t have much mindshare available for new releases.
Recent academic research supports this idea. According to Dr. Elizabeth Margulis, Associate Professor and Director of the Music Cognition Lab at the University of Arkansas, 90 percent of the music people listen to is music they have heard before.
There’s also a good chance that the musical style of new releases won’t mesh with the tastes of these listeners either. But that’s no problem for Pandora, because it already has millions of older songs available for streaming, plus a stream of new material from independent labels and musicians who are unaffiliated with a label of any kind.
Pandora can easily add a little bit of variety to a user’s station, without ever touching any newly released music. Once a piece of music has been coded with the appropriate metadata, Pandora’s algorithm can match it to other music that it should fit with reasonably well. In fact, one could listen 24 hours a day for a lifetime and not hear all of the songs Pandora already has available. With a backlog that big and a large cadre of listeners that unadventurous, the value of newly released music is diminished.
But even if many users don’t actually want that much musical variety in their radio streams, I think most users still value the possibility of variety that the Pandora service provides, especially in an era where breadth of taste is often viewed as an indicator of higher social status (See Carl Wilson’s book, Let’s Talk About Love, for more on that idea).
Most folks don’t want to admit to themselves that they like the tried and true above all else. Instead, they like to nurture a picture of themselves as aesthetically adventurous. They also seem to like the idea that their listening experience is custom-tailored to their specific likes and dislikes. So even if their Pandora stations lack variety, they like having the ability to determine for themselves what that specific lack of variety is going to be.
This is very different from most traditional terrestrial radio stations, where a mass of listeners explicitly choose to synchronously listen to a mix of music that has been curated by third-party experts. (If Pandora is the mePod, then traditional terrestrial radio is the wePod.)
As between the Pandora mePod and the terrestrial radio wePod, large record labels will take the wePod every time, even if it yields less income per spin per listener, because terrestrial radio formats and practices align more harmoniously with the music industry’s traditional bundling practices (i.e., each radio spin potentially touches many more people and competes with far fewer songs for listener attention)
Rather than receiving cash compensation in the form of sound recording performance royalties, major labels exchange their recordings for bundles of pre-screened potential music buyers (many of them lean-back listeners). The net result is an environment where it’s easier to gain promotional traction, focus consumer demand on a smaller universe of releases, and increase the likelihood of a blockbuster release.
3. Pandora is like the Pepsi Challenge of music listening, and the Pepsi Challenge is a deeply flawed method of choosing your favorite anything.
Most of us have had the experience at one time or another of hearing a new song on the radio and not liking it the first few times it is played. Then, after a few weeks of hearing it a couple of times a day, we have a change of heart. All of a sudden, we realize “wow, I actually do like this quite a bit.”
Not everyone behaves this way. Some music listeners assimilate new music very quickly, judge it up or down, and move on. These folks have an uncanny ability to know on the first listen or two that they’re hearing something worthy of their attention. But most people aren’t like that. They need to hear a new song multiple times before they form a solid opinion, just like sometimes one hates the taste of a certain food the first time one tries it but then later has a change of heart. As I said before, this is particularly true of lean-back music listeners.
I didn’t like the first sip of Laphroaig Scotch I tried. That’s for sure. But since my dad had bought me a bottle of it for Christmas and it’s an expensive bottle, I had some incentive to stick with it past that first sip. I’m glad I did. The more I drank it, the more my palette adjusted, and the better it got.
On the other hand, some things taste great at first, but the more you eat them (or drink them, or listen to them) the less appealing they become. Malcolm Gladwell spent a chapter in his book Blink talking about this phenomenon in the context of Pepsi/Coke taste tests.
Pepsi typically won the sip test, where people had to choose which sip they liked most. But when people were asked to drink more than a sip over a period of days or weeks (a so-called “home use test”), Coke won, because it was slightly less sweet and it wore better for many people over the course of drinking an entire can or case.
In my experience, the terrestrial radio user experience is much more like a home use test than a sip test (same goes for digital satellite radio). The terrestrial radio listener gives up most of the control. As a result, such a listener is much more likely to get multiple exposures to new songs (i.e. something more akin to a home use test).
Sure, the listener can change the station, and certainly plenty of listeners do. Terrestrial radio programmers spend a lot of time worrying about people changing the station, and they work hard to program in such a way that listeners do this as little as possible. (This is one of the reasons a lot of people think mainstream terrestrial radio sucks: programmers are constantly chasing that lowest common denominator).
But regardless of how one feels about the quality of a lot of terrestrial radio, I’m guessing that program directors succeed more than they fail and more people don’t twist the dial constantly than do. Or at least a significant number of people put on a station and stick with it for a while, because it’s a pain to constantly change the station.
As a result, terrestrial radio listeners are going to hear whatever songs their station plays during the hours they spend listening. And each time a song plays, all of the people then listening are going to hear it at the same time. So each spin is going to touch a lot of people at the same time (that bundling phenomenon I discussed in #1 above).
If an unfamiliar song comes on and some listeners don’t like it, they’re still going to hear it. And if they listen consistently to that station and this unfamiliar song is played regularly over a period of weeks, at least some of these people may well change their mind about it, just like many sip testers changed their mind about Coke when they had more extended exposure to it.
This is a big part of what music promotion is about: Convincing radio program directors and DJs to play a song enough that it has a chance to break through a music listener’s resistance to the unfamiliar, change her or his mind about a song, and maybe even begin the process of slowly building a long-term relationship between the listener and the artist. Throughout the last century, terrestrial radio has proven an efficient means of achieving these goals, and it has had a symbiotic relationship with large record labels in these endeavors.
A big reason this has worked is because average listeners have very clear expectations about what they get when they tune into a particular station. And they understand the role that repetition plays in different terrestrial radio formats (i.e., they know that terrestrial radio is often more than a sip test and they willingly sign up for that). This is why commercial terrestrial radio has been particularly popular with lean-back listeners.
Upon tuning into a particular terrestrial radio station, one learns pretty quickly what the premise is, in terms of its programming philosophy. What level of variety does the station provide? Are a steady flow of new releases a part of this variety? Is there variety within the older releases that get played? All of these factors are a part of a station’s programming premise, and by tuning in, one opts into that premise.
If one tunes into a non-commercial station like WFMU out in New Jersey, one expects a very eclectic mix of music. Indeed, much of the time, this mix is too eclectic for most people.
If one tunes into a non-commercial station like KEXP in Seattle or KCRW in Los Angeles, one expects a good variety of music, but one also knows that for much of the day the programming is not anything goes. There is a format. New releases get slotted in. Some songs are in heavier rotation than other songs, etc. Over a period of months, things shift, songs enter the programming universe and exit from it. A lot of thought and artistry goes into balancing the various elements.
If one tunes into a local classic rock oldies station, like KJR in Seattle, one expects to hear a bunch of old familiar hits from the ’60s, ’70s, and ’80s. But even on a station like that, which is not playing new releases, the programming still has a contour. If one listens a lot to an oldies station over a period of months, one realizes that the playlist is not static. It moves and changes too. Some songs are there all the time, but others slide into rotation for a while and then slide back out. Even within the universe of oldies, there is an effort to incorporate some variety into the programming.
Pandora’s programming premise is at once more general and more specific than any one terrestrial radio station format. It seeks to aggregate all radio stations and radio formats into one meta-format: the perfect balance of variety and familiarity made just for you and to your specifications. To continue with the bundling idea, Pandora is a bundler of radio formats as much as music, and its bundle is so large that it might be more accurate to call it a meta-bundle.
When Pandora users enter the Pandora listening experience, they expect that they will discover new music in the context of music that they already like. Unfortunately, in my experience, Pandora often fails to deliver on this premise. It manages familiarity pretty well, but it’s not nearly as good at managing either new release variety or variety within a group of familiar songs.
Pandora’s interactivity undermines new release variety, because it gives people the tools to avoid new music while telling themselves that they are embracing variety. It’s much more of a sip test than a home use test. If somebody has a bad reaction to a song on their first listen and they hit thumbs down, they are unlikely to hear that song again. As a result, listeners aren’t given an opportunity (or forced) to live with unfamiliar songs and perhaps change their minds about them. They have the option to take a sip, dismiss something, and never encounter it again.
As I already touched on above, the one-to-one, personalized nature of Pandora also makes it much less efficient to touch a large group of listeners in coordinated fashion over a relatively short period of time. Labels have no idea when one of their tracks is going to be heard in a stream, and unlike terrestrial radio, each spin doesn’t reach thousands of people simultaneously. In many cases, it just touches one person at a time. And if that person hits the thumbs down, the opportunity to market that song (and that artist) to them on Pandora is most likely gone for good.
I’m sure many people see this personalization as one of Pandora’s biggest virtues. But at least as far as traditional music promotion techniques are concerned, it makes music promotion a lot more difficult, because labels have less guaranteed ability to touch large masses of people with a single spin, and they also get less opportunities to expose the average listener to multiple spins of the same song. Consequently, I can see why record companies find this to be an unfriendly medium for promoting and selling music and musicians to the public.
Philosophically, I’m not so sure how fond of it I am either. For I wonder how well that sort of thumbs up/thumbs down sip test mentality serves our cultural life over the long run. If 77% of graduating high school seniors use Pandora for music listening at least occasionally and 54% use it at least once a week, one wonders, moving forward, how this will effective peoples’ relationship to music and their consumption of it. Will services like Pandora undermine the ability of artists to build the sort of connections with listeners that built long careers and a strong back catalog in the 20th century?
Call me elitist, but I worry about the consequences of living in a sip test world. Maybe we sometimes need to be saved from ourselves. I’ve had so many great experiences when somebody or something challenged me to push past my initial reaction and give something new a chance. To the extent that services like Pandora predominate and help us avoid doing that, I’m not sure it’s a win.
Call me lazy too, but I also find the constant musical sip test to be overwhelming after a while. As often as not, I appreciate a curated experience where I don’t have to sip test every aspect of my life. Institutions like record labels (and their promotional arms) play an important role in that curation process. They are filled with people who are passionate and knowledgeable about music (most music business people are lean-forward listeners). The best of these people also have the empathic ability to consistently predict what average people are going to like before these average people even realize themselves that they are going to like it.
To the extent that Pandora undermines the human/political aspect of curation, the power and significance of these empathic people is diminished. So it’s no surprise that many people in the music business have a visceral emotional reaction to Pandora, because Pandora has these passionate, lean-forward listeners fighting for survival.
Many music industry opinion-leaders now find themselves living a dystopic fable, in which they are John Henry with his hammer and Pandora, with its algorithm, is the steam engine. It may be efficient, but it lacks soul and seems to appeal mostly to people who don’t actually care about music or new music very much (a very large demographic as it turns out). In short, music industry people really dislike the aesthetics of the landscape Pandora creates.
Personally, while I’m passionate about music, I’m not anti-algorithm. I guess I’ve got a lean-back listener streak. I grew up listening to the radio, and I often prefer having somebody else drive the playlist. I was excited about Pandora when it came out, and I used it quite a bit for a while. But ultimately, I concluded that Pandora’s algorithm just isn’t that good.
This is why I rarely use Pandora anymore. I found it particularly bad in the realm of variety within familiarity. Too often, when listening to one of my Pandora stations, I’d hear the same “familiar” songs over and over again. And this isn’t just a snotty lean-forward listener complaint. I’ve heard the same complaint from people who I think would self-identify as lean-back listeners.
Pandora did a reasonably good job choosing songs that were familiar to me, but it did a poor job of dynamically shifting the universe of familiar songs that it plays on the same station over time. If it chose a Neil Young song for my station, it tended to play that same Neil Young song each time I played the station.
Of course, because it has a higher level of interactivity than terrestrial radio, I can press thumbs down and erase songs from the mix (or defer them for a period of time). But if I wanted to futz that much with a playlist, I’d use a fully interactive service like Spotify and make my own customized list. The point of radio is that it is a lean-back experience, where someone or something else is doing the driving for you. Personally, I don’t want my music listening experience to be one long sip test.
Perhaps the folks at Pandora believe that this sort of repetition is the same as the sort of repetitions one finds on terrestrial radio (i.e., the positive repeats that encourage people to try new things, live with them, and learn to love them and the artists who made them, even if they disliked them at first). Or perhaps they don’t include more variety within an artist’s catalog because it makes it more difficult for them to comply with the statutory license terms for non-interactive streaming radio (which I’ll talk more about in #4 below). Or perhaps they’ve determined that most people actually like those kinds of repeats.
Whatever the reason for these poorly nuanced repeats, I am much more annoyed by them than I am by repeats on terrestrial radio. They violate the premise of variety within a zone of familiarity that is central to Pandora’s overarching programming philosophy.
If KEXP plays Neil Young every day (which they probably don’t), they certainly don’t play the same Neil Young song every day. The DJs know that “Sugar Mountain” was played last week, so if they want to play a Neil song today, they choose something different (or perhaps they decide that it’s too soon to play another Neil Young song).
On the other hand, KEXP may play a new release in heavy rotation 10 or more times in a week. But unless the listener reaction to a particular song is incredibly strong, it will start exiting heavy rotation within a month, and from then on, you’ll only hear it intermittently (if at all).
In my experience, Pandora’s algorithm isn’t smart enough to make those sorts of distinctions within variety, and personally, I find that very annoying. Annoyed is not an emotion that any music service or radio station wants its listeners to feel. If something is annoying, people usually try to avoid it. Human radio programmers understand this clearly, and much of their art involves finding the right balance to avoid this reaction from their particular listening audience.
On Pandora, if repeats are annoying, the user hits thumbs down on the offending song, which is a cool option to have. But pressing thumbs down also takes energy, and the more energy one spends trying to reign in poorly nuanced repeats of familiar songs, the less bandwidth one has for processing and absorbing unfamiliar songs (and developing relationships with new artists). As result, new release repeats start to blend together with the repeats of familiar songs. Over time, all repeats start to feel annoying, regardless of their character, and the more annoying they seem, the less valuable they become as tools of promotion.
4. It’s difficult to negotiate a better deal with a statutory royalty rate.
In the early 1970s when I was around 7 years old, my family lived in San Miguel de Allende, Mexico for a few months. Sometimes, I’d go with my dad to the outdoor markets. This was my first introduction to bargaining. My dad would ask how much an item cost. The person at the market would name a price. My dad would counter with a lower price. Eventually, they’d either come to an agreement and a deal would be made or we’d walk away.
My dad was a Jewish guy who grew up in Philadelphia in the 1930s and 1940s, when bargaining was still the norm on many commercial transactions. As we were walking away from the market one day, he explained something important to me about negotiation: “If you aren’t prepared to walk away, you’ll never get the best price.” In other words, the one who needs it least typically has most of the bargaining leverage.
Major labels and large music publishers understand this adage all too well. Through the years, the bargaining leverage has mostly been on their side in the commercial transactions they undertake. Any transaction around recorded music involves the rights to two distinct copyrights: (1) the copyright to sound recording; and (2) the copyright to the underlying musical composition.
Typically, record labels control the sound recording copyright and music publishers control the musical composition copyright. And historically, in most recorded music transactions, the transaction can’t happen unless both these parties give their permission. So if either one doesn’t like the looks of a deal, they can vote with their feet and walk away from it.
But with non-interactive streaming radio services like Pandora, the major labels and publishers can’t just walk away, and here’s why:
Pandora is allowed to use the labels’ sound recordings pursuant to license terms that are mandated by the government (i.e., a statutory license). SoundExchange, a non-profit performance rights organization (“PRO”), is authorized by the government to collect the statutory non-interactive streaming license royalties and distribute them to the various stakeholders.
The same is true of music publishers. Most songwriters and music publishers are affiliated with a PRO like Ascap[1] or BMI. Typically, each PRO negotiates and enforces blanket performance license agreements on behalf of its members (although see below). These agreements allow a licensee to perform all of the works in the PRO’s catalog. By banding together in a PRO, songwriters and publishers have historically been able to secure better negotiating leverage and more thorough enforcement of performance licenses than any individual songwriter or most music publishers could accomplish on their own.
When a large group of potential competitors band together for shared advantage, it is likely to raise antitrust concerns, and both Ascap and BMI have been subject to antitrust litigation with the Department of Justice (“DOJ”). As a result, for many decades, both Ascap and BMI have operated subject to antitrust consent decrees with the DOJ. In the case of Ascap, its consent decree includes the following requirements:
- Ascap members may grant only non-exclusive rights to Ascap, and they must retain the right to individually license their works;
- Ascap must grant non-exclusive licenses to licensees either for a period of time or on a per program basis;
- Ascap cannot insist on a blanket license;
- Ascap must offer a genuine economic choice between a blanket license and a per program license; and
- If a putative licensee and Ascap disagree on fees, the putative licensee may request that the Federal District Court for the Southern District of New York (SDNY) determine a reasonable fee, with Ascap having the burden of proving the reasonableness of the fee.
From a practical standpoint, this means that there is also de facto government regulation of negotiations about musical composition performance rights (more on this in #5 below).
Therefore, unlike interactive streaming services (e.g., Spotify, Rhapsody, etc), which must negotiate licenses directly with the record labels, Pandora doesn’t need to negotiate directly with each label about the sound recordings in its catalog. It just uses the statutory license framework for that and the anti-trust consent decree process for the performance rights on the underlying musical compositions.
To the extent that there has been any negotiation at all on the statutory rates, it’s mostly been in the context of court litigation, administrative law proceedings, and government lobbying. Just to put a cherry on top, Pandora has pushed hard in Congress to get the statutory streaming royalty rates reduced (rates that record labels argue are already too low).
So right out of the chute, there is no straightforward way for the music industry to get a better deal. And as you might imagine, record labels really don’t like that dynamic (forgetting, of course, that their own business of selling records was built in no small part on the back of the statutory mechanical license in the decades after the 1909 Copyright Act created it—a rate that remained unchanged for many decades). But it gets worse from there, because of how the existing statutory licensing deal works:
Non-interactive streaming services like Pandora pay all the digital performance royalties they owe for using sound recordings directly to SoundExchange. From there, SoundExchange is mandated by law to distribute each dollar of streaming royalties it collects as follows:
- 50% goes to the owner of the sound recording copyright (if there is a record company involved, this will typically be the record company);
- 45% goes to the featured artist (if there is a record label involved, this will typically be the solo artist or members of the group featured on the recording);
- 5% goes to non-featured musicians (aka session musicians) through payments that SoundExchange makes to the American Federation of Musicians Trust Fund; and
- 5% goes to non-featured vocalists (aka session singers) through payments that SoundExchange makes to the American Federation of Radio and Television Artists trust fund.
If you understand anything about how a record contract works, you’ll immediately see that the statutory streaming royalty framework is quite different than a traditional transaction in the music industry.
For example, when a store sells a CD, it pays 100% of the wholesale price to the distributor (which in the case of major labels is typically an arm of the label). From there, the label controls that money. If an artist has any unrecouped advances from the label, all of the artist’s share of the CD sale goes to recoup those advances and the label pays the artist nothing until that money is paid back.
But under the statutory streaming royalty framework, 45% of every dollar that Pandora pays to SoundExchange gets directly distributed to the featured artist. Even if the featured artist’s royalty account is unrecouped, the record label never gets to touch this money, because it goes straight in the artist’s pocket.
Deals that labels do with interactive streaming services like Spotify typically don’t work this way. They work more like a physical product transaction. The interactive streaming money first flows through the label and is applied to unrecouped advances before the artist receives any of it.
Moreover, in a number of cases, labels have taken an equity stake in interactive streaming services like Spotify or Beats as a condition of licensing their content. So if the interactive service is later acquired by a third-party, the label receives a windfall, and the label is not necessarily required to share this money with its artists. It all comes down to what the artist’s contract says, and most artist contracts don’t obligate the label to share this pool of money.
This is exactly what has happened with Apple’s recent $3 billion acquisition of Beats Music. Those labels with an equity stake will get a share of the $3 billion, none of which is likely trickle down to their artists.
This is how labels like it. As much as possible, they want all the cash to flow through their coffers first before the artist touches it. Job one for a label is recouping the investment it makes in any artist it signs. The more streams of income that flow through the label, the better the odds are that the costs can be recouped. And at the end of the day, it’s much easier to avoid paying money than it is to collect money you are owed.
So labels hate the statutory royalty deal for at least two reasons. First, they think that the gross per stream royalty is too low, but their hands are mostly tied to make it higher. Second, the net portion of each statutory royalty dollar that can be applied toward recouping artist advances is half of what they get from most other exploitations of the artist’s master tapes (e.g., record sales, digital downloads, interactive streaming, movie and television licensing).
The more music consumption happens on non-interactive services like Pandora, the more gross income is not subject to the traditional deal. Consequently, labels are justifiably concerned about their future if more and more consumption happens on Pandora.
5. Pandora is revenge of the nerds: So uncool.
Historically, the most valued currency in the music industry has been “cool”. Conversely, the most valued currency in the tech industry has been “smarts.” For a very long time, cool has usually trumped smart out in the world at large. But since the dawn of the Internet, smart has increasingly trumped cool. It’s trumped cool so often that it has created its own sense of what cool is. Indeed, a decade into the 21st century it sometimes seems like smart and cool have merged together.
That’s why Pandora’s rise has such a coup de grace, revenge of the nerds, super-mega-fuck-you aspect for the music industry.
For decades, people in the music industry were cool and mostly in control of their business. They were among the opinion leaders of the cultural zeitgeist and usually got to call the shots and get their way when it mattered.
As the rules and regulations governing digital streaming were being debated and crafted in Washington during the 1990s, the record industry and music publishers lobbied hard to limit so-called interactive streaming services, because they were worried that this sort of streaming was too much like the record listening experience and would therefore cannibalize the sale of physical product.
But this position on streaming put the record business at odds with the other media cool kids in the broadcast industry. Broadcasters could see a lot of potential in digital streaming technology and they wanted the latitude to pursue it. As a result, streaming couldn’t be completely killed. Instead, a compromise had to be struck, and a distinction was drawn in the law between interactive and non-interactive streaming services.
Rules were put in place governing what constitutes a non-interactive streaming service, and the statutory non-interactive streaming royalty I discussed above in #4 was put into place.
On its website, SoundExchange summarizes the attributes that a service must have to be considered non-interactive, including the following:
- not allowing the listener a reasonable foreknowledge of the transmission of a specific sound recording at a specific date and time (e.g., by the use of a published advanced program, playlist, announcement, etc.);
- not creating transmissions as part of an “archived program” (a) less than 5 hours in duration, or (b) available for a period exceeding 2 weeks;
- not creating transmissions as part of a continuous program less than 3 hours in duration;
- no more than 4 tracks by the same featured artist (or from a compilation album) may be transmitted to the same listener within a 3 hour period (and no more than 3 of those tracks may be transmitted consecutively); and
- no more than 3 tracks from the same album may be transmitted to the same listener within a 3 hour period (and no more than 2 of those tracks may be transmitted consecutively).
At the point of impact, it seemed like the record industry had accomplished the goal of protecting its core brick-and-mortar business from the threat of interactive streaming services. But as with all legislation, the possibility of unintended consequences was always lurking in the background.
The science writer Steven Johnson has described this space as “the adjacent possible,” a term coined by the theoretical biologist Stuart Kauffman. It is “a kind of shadow future, hovering on the edges of the present state of things, a map of all the ways in which the present can reinvent itself,” and the “strange and beautiful truth about the adjacent possible is that its boundaries grow as you explore them. Each new combination opens up the possibility of other new combinations.”
As skilled hackers always do, people at companies like Pandora dug into the existing music industry ecosystem, probing it for weaknesses and opportunities. In the case of Pandora, they looked at the intersection of the broadcast industry and the recorded music industry, asking the following question:
“What do music consumers want and what are the adjacent possibilities of digitized music and networked computers to give it to them in a more user-friendly, efficient, pleasing, and automated fashion?”
Note that this question is focused on the audience. It’s not concerned with the people who make and sell music.
Their answer was a service that used an algorithm to provide the mix of musical variety and familiarity that I discussed above.
Next, they asked a second question:
“How can we deliver this service to a mass audience with the least amount of interference from various third-party rights holders, who may try to exert control over the nature of the service or demand more compensation for their content than we are able to pay, thereby killing it before it has a chance to prove its viability?”
After digging into the adjacent possibilities of the relatively new regulatory framework for digital streaming, the hackers at Pandora concluded that it offered them a great opportunity. They could engineer their service to take advantage of the new statutory license for streaming radio, make an end run around the record labels, and mostly control their own destiny without label input or interference.
The result? Even in the areas where Pandora hasn’t completely controlled its own destiny, it has still forced record labels and music publishers to do battle on ground that is much more favorable to Pandora’s interests than the alternative.
Rather than sitting in a private conference room where either party can walk away at any time and for any reason, Pandora has forced record labels and music publishers to negotiate in arenas where the fate of labels and publishers is mostly in the hands of decision-makers other than the labels and the publishers (e.g. In the legislature, administrative law proceedings, and in the courts of law and public opinion).
A recent example of this dynamic is Pandora’s conflict with Ascap over performance royalty rates for the period from 2013–2015. During the course of these negotiations, certain major publishers (e.g., Sony Music) withdrew digital streaming rights from Ascap’s catalog, opting instead to negotiate directly with Pandora. The goal of this strategy was to gain better leverage for both the publishers and for Ascap.
But when private negotiations between Ascap and Pandora did not yield agreement on a royalty rate, Pandora utilized the antitrust consent decree discussed in in #4 above and brought a rate court suit against Ascap in the Federal District Court for the Southern District of New York.
There, Pandora sought a reduction in its current royalty rate from 1.85% of its revenue to the rate that terrestrial radio was currently paying, 1.7% of revenue. Ascap, on the other hand, sought to raise Pandora’s royalty rate to 2.5% of revenue for 2013, and 3% of revenue for 2014 and 2015.
In the end, the court’s decision was something of a draw. Rather than changing the royalty rate, the court decided to keep the current 1.85% rate in place. That being said, the decision did apparently save Pandora $21 million in royalties compared to what Ascap was asking for. In addition, the court indicated that the Ascap consent decree did not allow member publishers to withdraw specific classes of rights from the Ascap catalog. If a publisher wants Ascap to administer its catalog, Ascap must administer all classes of rights. Otherwise, it can’t administer any of that publisher’s rights.
The implications of this second part of the decision are still being sorted out. At least for now, major publishers like Sony and Universal won’t be withdrawing their compositions from the PROs. But if they eventually end up withdrawing them (as they have threatened to do), it may well imperil the continued existence of the PROs, because the big publishers generate a significant portion of each PRO’s revenue
Big-name artists who control their own publishing may also choose to leave the PROs. For example, Pharell Williams recently announced that he was leaving Ascap for Global Music Rights, the rights management start-up founded by music industry power-broker Irving Azoff, whose roster also includes members of Journey, Foreigner, Fleetwood Mac, Soundgarden, and the estates of Lennon and Ira Gershwin. Recently, Azoff indicated publicly that he was prepared to take 42 of his clients, away from the YouTube ecosystem, including its new Music Key service. And there’s been speculation that it won’t be long before Pandora hears from him too.
If things continue to trend in this direction, it will bring significant change to the music publishing landscape. And while it may not negatively affect the large publishers and big-name songwriters, who have the resources to administer and enforce performance licenses themselves, the collapse of major PROs could be very damaging to individual songwriters and smaller publishers.
For example, the Ascap consent decree requires Ascap to directly pay songwriters their share of any money that it collects (very similar to the way SoundExchange works, as I described in #4 above). This is irrespective of any contract that a songwriter may have entered into. This means that Ascap writer payments always go directly to the songwriter and aren’t used to recoup advances of publishing contracts, etc. BMI is required to handle things similarly.
In a world without PROs, these payments probably would no longer be handled this way. Instead, it would be whatever a songwriter’s contract said, and the songwriter’s share of performance income would likely be subject to recoupment and the less transparent accounting practices of private publishers. If that’s the end result of all this, it will be a huge disruption of a framework that has been in place for many decades. It will probably also be bad for many songwriters.
While the courtroom and the legislative battles continue to rage, the struggle to unseat Pandora in the marketplace has also continued. Everyone was sure that services like Spotify or Apple’s new iTunes radio product would be the final nail in Pandora’s coffin. Didn’t happen. More recently, Apple has purchased Beats Audio in an effort to beef up its streaming music presence. And while it’s still entirely possible that music licensing costs will eventually kill Pandora as a stand-alone company, Pandora is still here, at least for now. Its service continues to resonate with the public, and its user-base continues to grow.
As things stand, it’s hard to argue that Pandora hasn’t managed to out-hack the music industry, and in many respects, outmaneuver it both politically and in the legal system ever since.
That being said, there have also been a few recent signs that perhaps the music industry is starting to adapt to the new rules of engagement. For example, in the wake of the Ascap/Pandora decision (and one imagines lobbying from the music publishing industry), the DOJ recently announced that it will review the antitrust consent decrees under which both Ascap and BMI have operated since 1941.
As part of this process, the DOJ has requested public comments from stakeholders, so it can evaluate whether the existing decrees continue to protect competition. Perhaps the outcome of that process will be a set of rules that allow publishers more flexibility about what rights they choose to grant to the PROs. Or perhaps it will lead to the end of the consent decrees and a more wide-open, unregulated market.
The recently proposed Songwriters Equity Act of 2014 (“SEA”) is another attempt by the music industry to change the rules under which it engages with companies like Pandora. According to an Ascap fact sheet, the SEA would amend sections 114 and 115 of the Copyright Act, allowing the Copyright Royalty Board, which sets certain statutory royalty rates, to do the following: (a) consider other royalty rates as evidence when establishing digital royalty rates; and (b) utilize a standard for determining compulsory mechanical license royalty rates that better considers market conditions.
(Update: After this post went up, Casey Rae, CEO of the Future Music Coalition, reminded me of the following in a Facebook comment: “The Songwriter Equity Act’s call for allowing evidence from sound recording statutory rate setting proceedings is the exact opposite of what the publishers pushed for (and won) back in 1995.”
That’s a great example of the difficulty Pandora has presents to incumbent labels and publishers trying to defend their existing business model in the legislature. The future moves very quickly, and it is full of unintended consequences, which are hard to predict. The legislative process, by contrast, moves very slowly and does not quickly adapt to unintended consequences. As a result, it’s almost always a step or two behind the future.)
The music industry has also recently prevailed in a couple of court cases against Sirius XM, relating to whether Sirius XM was required to pay public performance royalties on sound recordings created prior to 1972, the year that Congress formally established a federal sound recording copyright.
Sirius/XM satellite radio had argued that it was not required to pay statutory royalties on pre-1972 sound recordings, because those tracks weren’t explicitly included in the federal copyright provisions that established the statutory royalty. Instead, state copyright law should apply (in this case California’s copyright statute).
Sirius/XM had then maintained that section 980(a)(2) of the California copyright statute did not explicitly include the exclusive right to control public performances of a recording. Therefore, the California copyright statute did not grant owners of pre-1972 sound recordings the right to sue for infringement based on public performance. As a result, users like Sirius/XM and Pandora did not need a license to perform these tracks.
Ultimately, the judge in the case found for the plaintiffs, Flo and Eddie, members of 1960s band the Turtles, concluding that Section 980(a)(2) of the California copyright statue did grant copyright holders an exclusive public performance right.
The Sirius/XM decisions will undoubtedly be appealed. But if the rulings are upheld, it will be a significant victory for the music industry. And it could have a major of effect on services like Pandora, most of which were using more or less the same arguments as Sirius/XM to justify their refusal to pay royalties on pre-1972 sound recordings.
Perhaps these decisions about California copyright law will mark the moment when the old-school music industry turns the tide and gains additional negotiating leverage vis a vis services like Pandora and Sirius/XM, as labels and artists force these services to do battle over pre-1972 masters on state by state basis. That being said, if the recent past is any guide, like them or hate them, it would be foolish to underestimate Pandora and its digital music brethren, for as Stuart Kauffman said in a 2012 lecture: “We seem to evolve into the adjacent possible.”
Whatever happens, it will be interesting to see how things play out.
Afterword 1: What about YouTube? Doesn’t the music industry hate them at least as much as Pandora?
Yes. The music industry may actually hate YouTube even more than it hates Pandora. But YouTube is different than Pandora. It is a platform for user shared audiovisual content. As such, it has facilitated unauthorized distribution and playback of all kinds of content, including music. But even though YouTube is now readying a paid music service, it has always been more than a music service. Indeed, I doubt YouTube’s creators intended it to be a music service when they rolled it out.
It was YouTube’s users that turned YouTube into a de facto music service, as they grasped its adjacent possibilities and realized that it could be used as an interactive music streaming platform. YouTube certainly helped this along at various points (e.g., by adding the playlist feature). But YouTube remains different than dedicated music streaming services like Pandora and Spotify. It’s more like a gray market version of a P2P sharing platform in terms of how people use it.
On the one hand, it’s perceived as being more legitimate and approachable to mainstream consumers than P2P platforms like Napster and BitTorrent (having the backing/protection of Google has helped mainstream it). It also has a lot more legitimate user-generated content, whether that’s cat videos, wedding ceremonies, or whatever.
On the other hand, there’s some significant overlap between YouTube and P2P, in terms of the volume of unauthorized content available there and the Whack-A-Mole game content owners must play to get this content taken down. So while YouTube has had a major role to play in disrupting the music business, it’s a different role than a service like Pandora.
That’s why I haven’t focused on YouTube in this blog post. Even without YouTube, Pandora would still be a huge problem for the music business, because of the way that it has leveraged the PRO antitrust consent decree and the statutory royalty scheme around non-interactive digital streaming. Pandora might be even more popular and successful right now if YouTube had not evolved into a music streaming service, because there would likely be fewer interactive streaming players competing with Pandora right now for market dominance in the legal music streaming space.
It’s the existence of P2P and YouTube that has provided interactive streaming services with one of their best selling points to labels: “You should license us your content, because even if you don’t like the money we are offering, it’s significantly better than the money you can make from P2P or YouTube, and over the long-haul it will get better if we succeed in the market, because we are cutting you in on a share of our revenue.”
As explained above, Pandora hasn’t had to have that sort of conversation with labels, because it hasn’t needed to do direct deals with them. So labels get little or no say with Pandora about deal terms, and they have much less influence over what gets played there. It’s the worst of both worlds.
Afterword 2: What about musicians? How should they feel about services like Pandora?
If all the disruption wrought by services like Pandora is undermining the power of big labels, where does that leave musicians? Shouldn’t the average musician benefit from that? Isn’t a flatter hierarchy with better platform access for more people a big win for most folks? What about getting paid directly 45% of every non-interactive streaming dollar (or 100% of every non-interactive streaming dollar if you control your own masters)?
The answer to those questions depends on who you are. For insurgent, independent musicians like Zoe Keating the answer can be “yes,” because they really don’t have much to lose. Pandora’s flatter hierarchy gives them a better chance of being heard than does commercial terrestrial radio, where an independent musician has almost zero chance of getting airplay. Moreover, any money independent musicians make from streaming royalties is money they probably wouldn’t have otherwise earned. That said, any royalties they earn are still dwarfed by the promotional value of Pandora.
Pandora has also recently unveiled the AMP platform, which allows artists to view Pandora listener data about their songs. This sort of information could definitely be useful in helping an artist understand where he or she is most popular. In turn, this could lead an artist to play a show in a market that would have otherwise been ignored or help an artist to determine where to focus promotional energy.
Notwithstanding those potential upsides, it’s also possible that the AMP platform won’t end up being that useful for many artists, that Pandora spins won’t yield much promotional value, and that all they’ll do is decrease the likelihood that a listener will purchase an artist’s album, thereby depriving the artist of sales that have a much higher profit margin than even numerous Pandora spins.
Here’s a quick example to illustrate what I’m getting at:
Let’s say that it’s 2012 and an artist sells her album at shows for $12. It’s a pretty bare bones album that didn’t cost that much to record. She wrote all the songs. She makes $4 of profit on each sale (so that’s the total net revenue from both the masters and the publishing, and we’re assuming that the other $8 of each sale goes to cd replication, recording expenses, studio musicians, etc.).
In 2012, a Pandora spin was worth $0.0011 in digital performance royalties (i.e, the money I described in #4 above, which Pandora would typically pay SoundExchange). According to Ascap, a Pandora spin is also worth around $0.00008 in performance royalties on the music publishing, which go to the songwriters. That said, some songwriters have indicated that their publishing royalties on Pandora spins are lower than that.
But if we accept the numbers above as reasonably accurate, then the aggregate royalties that Pandora pays on a single spin would be around $0.00118. Or to put it a different way, Pandora pays around $1.18 in royalties for every 1000 Pandora streams of a given track. Therefore, to match the net income from a single album sale, our hypothetical indie artist above would need approximately 3,390 Pandora spins.
As you can see, you need a lot of Pandora spins to make the same money you’d make from one album sale. If you are an artist who is currently very popular in the marketplace (e.g., Macklemore and Ryan Lewis), this may not be a problem for you. Indeed, you may well benefit financially from Pandora, because you are generating the exceptionally large number of spins necessary to yield sizable royalty payments. You are also probably getting a good promotional benefit from Pandora.
But for most other folks, including many incumbent musicians and songwriters who had success in the pre-digital music business, Pandora is more likely hurting their earnings.
In the old days, selling 1000 albums was kind of an initial benchmark for a local band. If a band sold 1000, nobody was quitting their day job, but it was an indication that the band might have a shot at building something bigger. If we use the numbers from my example above, a band would need to accrue 3,390,000 spins on Pandora to make the same money it would make selling 1000 albums with a $4 profit on each one. Even if the profit was only $2 on each album, a band would still need to accrue 1,695,000 spins on Pandora to make the same money.
The scale and geographic reach of Pandora’s user-base is obviously much larger than that of the average local music market in which the indie band was trying to sell its 1000 records. But even so, accruing 3.6 million of anything is a challenge, especially when you have very little input or control over whether or when your music will be played.
Most folks will just have to hope that the Pandora algorithm is kind to them. Or in the alternative, they’ll need to reach listeners by some other means first and convince them to go to Pandora and build stations from their songs.
It gets back to that bundling idea from #1 above. What would you rather do, convince 1000 people to pay you $12 or hope that significantly more people (maybe even millions) will find your music on Pandora (a process that will be more or less random if you are an unknown band)?
“But wait,” you may be saying, “these two paths aren’t mutually exclusive, are they?”
No, they’re not. But it’s naive to think that the existence of services like Pandora haven’t changed consumer expectations.
When somebody buys an album, they are in essence pre-paying for the right to listen to those songs forever. But they are also taking on a higher level of financial risk upfront. If they don’t end up liking the album, they don’t usually get their money back. The band gets to keep it.
Conversely, on Pandora, you pay nothing (or a recurring monthly fee of less than $10), so there is very little risk involved in checking out new stuff, but when you haven’t paid upfront, you are also less likely to spend extra time with something you don’t immediately like, because there is no sunk cost.
The sip test is in full effect.
As a result, when that band is out trying to convince those 1000 people to spent $12 on its album, it faces a much more difficult environment today than the one that bands faced 20 years ago, because consumers no longer have to pre-pay to get the music.
For incumbents, who crossed the threshold into mass success back in the day, Pandora also creates a related but slightly different problem. Most of these artists are no longer ascending and their best sales are behind them. But in the pre-digital era, their past successes helped them to build a relatively high equity value in their back catalog. As a result, many artists continued to make a good income stream from back-catalog sales.
Along the same lines, if an artist was also a songwriter, this person might continue to make a pretty good stream of income from terrestrial radio play as well. One reason this was true is because of the way that PROs distribute the big pot of money they collect from terrestrial radio plays. All songs in a PRO catalog are not created equal. Unlike Pandora, which distributes on a census basis (hard count of actual plays), PROs like Ascap and BMI distribute performance money from terrestrial radio plays based on a survey (i.e., using a statistical model).
It’s what Bandago CEO Sharky Laguanahas has termed a parimutuel royalty system: all the money collected goes into a big pool and then gets distributed to artists based on their share of overall terrestrial radio plays as determined by the survey. PROs also collect money for the use of music in bars, restaurants, clubs, and other venues. This money also gets distributed according to that survey.
If your song is played occasionally on the terrestrial radio and it never comes up in the survey, you won’t receive any royalties for those spins (or any uses in a bars, clubs, etc.). Conversely, on Pandora, the same spins do generate a royalty, although it will be very small.
The parimutuel distribution methodology favors those songs that have been played the most on the radio. In some cases, songs with a prior track record of radio success continue accruing income, even when they aren’t being played on the radio as regularly anymore.
Moreover, as noted above, a much smaller universe of songs get played on terrestrial radio than on Pandora, and these songs tend to get played over and and over again. Consequently, if one of your songs enters that universe of terrestrial radio popularity, there’s a good chance you’ll continue to make money from it, even as its volume of spins drops off, because your song is one of the big winners in the parimutuel pool.
The survey approach isn’t exactly a winner-takes-all scenario, but it does tends to benefit some PRO members more than others. For these winners, it’s like interest compounding or appreciation in the value of a piece of prime real estate. Initial success begets income growth, while increasing the statistical odds that a writer will continue to receive more income over time.
Back in the day, most musicians and songwriters never made it into this cohort (and even fewer folks make it in there today). But for the select few who did fight their way over that hump, the mailbox money was a nice reward for years of hard work (I’m guessing that this has never been more than 5% of aspiring songwriters and the number is probably considerably smaller than that).
In the Pandora world, unless something significant changes, recorded music is unlikely to have the same sort of meaningful, long-term equity value for artists that it had in the period from 1945-2004. Sure, musicians like Macklemore and Ryan Lewis are raking in some pretty good cash right now, but it will be interesting to see, over the next 25 years, whether the inflation-adjusted earnings of their masters and publishing ever matches those of a band that was “successful” in the glory days of the pre-digital music business (e.g., Frank Sinatra, Neil Diamond, Paul, Simon, Carole King, Bob Dylan, Fleetwood Mac, Bob Seger, Guns and Roses, Pearl Jam, Nirvana, etc.). Barring something unforeseen, I suspect that the answer will be “no.”
Indeed, it’s very much an open question right now whether the post-digital, Pandora world can sustain successful, long-term careers for artists, rather than just isolated episodes of success (which are nice but insufficient to sustain an artist over the long haul).
Episodic success has always been more prevalent in the music business than long-term success. But it’s the tangible examples of long-term success that inspire many people to follow their dreams and shoot for the stars. If it becomes clear that this sort of success is even more of a pipedream today than it was in the 20th century, I wonder if this will reduce the number of people who choose to follow their dream? Indeed, some have argued that this is already happening.
In any case, if you are a person currently trying to follow your dream, it’s important to consider how your dream may be intertwined with the fate of the large record labels currently struggling to adapt to Pandora reality.
Don’t get me wrong, there are plenty of negative things to say about these institutions. But even if you aren’t interested in signing to one, their continued existence may nevertheless affect your ability to realize your goals.
Moving a project forward requires an eco-system and the availability of capital. Whether you like them or hate them, major labels are an important part of the current music business eco-system. To the extent that they falter, this eco-system may be undermined, and that may not be a win for you.
Sure, it’s a lot cheaper in 2014 than it was in 1974 or 1984 to record your music and distribute it all over the place. Therefore, the ability to raise the money you need to do these things is no longer the blocking issue for most people. Petty much anyone with $500-$1000 and some skill can make a pretty good sounding recording and distribute it on-line. You don’t need a major label deal to do that.
But the ultimate goal remains the same: If you are going to have a long-term career, where you can viably make even a meager living, you will need to somehow rise above the pack and focus significant consumer demand on your project. Back in the day, you wouldn’t have even been given the opportunity to make a record until you had distinguished yourself from your peers. Today, just about anyone can record a record for an affordable price. So the problem has shifted.
Since the onset of a stable market for recorded music at the start of the 20th century, musicians have been a lot like silicon valley start-ups are today. Not everybody admitted it out loud, but most were seeking to be acquired by a bigger company (i.e., signed to a label). Some iconoclasts went their own way and stayed independent. Some managed to make a go of it.
Especially after the onset of affordable multi-track recording equipment, the independent route became a more viable option. But for every band like Superchunk, whose members managed to start their own label and build it into a successful business (Merge Records), far more musicians either failed in their efforts to start an indie label or opted to be acquired (“signed”) to a large label if that option presented itself.
If the major-label sector of the eco-system is unhealthy, as it has been for a while now, that has a way of trickling into the eco-system as a whole. Less money is available to fund acquisitions (i.e., to sign new artists). Less money is available for development, and what money there is gets focused in more risk averse fashion.
That doesn’t mean that nobody will be able to break through and succeed in this situation. Indeed, it may create opportunities for insurgent artists at the grass roots level. But as a musician, it’s important to have a clear picture of how your goals may be intertwined with those of the large labels, even if you despise them and don’t aspire to getting signed by one.
On the other hand, it’s also important to be clear about when your interests may not be aligned with those of a large label or music publisher. Major labels and big publishers see things through the prism of their problems and needs (and to a certain extent, their artists and songwriters do too). They are the incumbents. They have benefited from their incumbency. And while these players have been slowly losing many of these benefits, they’re not passively accepting that outcome. They’re fighting like mad to preserve as many benefits as they can.
Therefore, the bottom line questions for a musician must always be these:
“Is their fight my fight? They claim that they are fighting for me, but are they really fighting for me? They claim to know best, but do they really know best?”
The only way you can answer these questions intelligently is to educate yourself enough about the issues so that you can understand how your interests relate to the whole (and how your short-term interests and your long-term interests might not always be the same).
That’s the trade-off of 2014. It’s easier than ever for people to produce and distribute their music. But if you want to make money from it and build a career, you can’t just be an artist. You also need to be an entrepreneur. You can’t say “I just want to make music. I’ll let other people deal with the business details.” You need to dig into the details yourself.
I hope this blog post has made a small contribution to helping you engage in that process.
What’s your take? Was this deep dive helpful? Am I full of it? Hit me up on Facebook or Twitter and share your thoughts.
- I opted in this piece to follow the NYT‘s style of referring to the American Society of Composers, Authors and Publishers as “Ascap” rather than “ASCAP.” Why would an ASCAP member do that? Well, upon reflection, I decided that the NYT was on to something with its approach. At least to my eye, “Ascap” is less visually jarring on the page than “ASCAP,” especially when it’s used repeatedly. So I went that way. One of the joys of blogging is that I sometimes get to follow the Jake London Manual of Style. But please don’t hold me to “Ascap” over “ASCAP.” I might change my mind in the future.↩
One Reply to “Five Reasons The Music Industry Hates Pandora The Most”
Comments are closed.