Executive Summary

9 - 3 1 0 - 0 2 6 R E V : O C T O B E R 2 , 2 0 0 9

________________________________________________________________________________________________________________ Senior Lecturer Robert C. Pozen and Alex Rosenfeld (MBA 2009) prepared this case with assistance from Global Research Group Senior Researcher David Lane. HBS cases are developed solely as the basis for class discussion. Cases are not intended to serve as endorsements, sources of primary data, or illustrations of effective or ineffective management. Copyright © 2009 President and Fellows of Harvard College. To order copies or request permission to reproduce materials, call 1-800-545-7685, write Harvard Business School Publishing, Boston, MA 02163, or go to www.hbsp.harvard.edu/educators. This publication may not be digitized, photocopied, or otherwise reproduced, posted, or transmitted, without the permission of Harvard Business School.

R O B E R T C . P O Z E N

A L E X R O S E N F E L D

Pandora: Royalties Kill the Web Radio Star? (A)

Let’s hope you never leave old friend Like all good things, on you we depend So stick around ’cause we might miss you

— Queen, “Radio Ga Ga”1

Joe Kennedy, president and CEO of Pandora, sat in front of his computer and closed out of the music player on his company’s website.a His office fell eerily silent, which felt appropriate given the news he had just received from Washington, D.C. The day was March 2, 2007, and the Copyright Royalty Board (CRB), the government body that determined rates for copyright statutory licenses, had just announced its decision to increase the sound recording performance royalties required of the web radio industry by 2.5 times over the next five years.

Since Kennedy had joined the company in 2004, Pandora had grown into one of the largest and most popular web radio broadcasters, or “webcasters.” In 2007, Pandora claimed 5.7 million registered users, a figure that had increased by over five-fold in just the past year.2 Advertisers were increasingly drawn to Pandora’s young audience. At the most recent board meeting, Kennedy had announced to Pandora investors that the company was projected to break even in the next two years.

The announcement from the CRB drastically changed this outlook, however. Five seemingly infinitesimal numbers expressed in fractions of a penny—the company’s sound recording performance royalties for 2006 through 2010—pushed profitability out of sight. This increase in sound recording performance royalties, which were the royalties Pandora was required to pay each time it broadcast a sound recording, made it more likely that the company would require additional funding on top of the tens of millions it had already raised, and brought up serious questions about the business’s ability to survive.

Kennedy and Tim Westergren, the company’s founder and chief strategy officer, firmly believed in the right of record companies and recording artists to be compensated for the use of their copyrighted recordings. Pandora had paid sound recording performance royalties of $3.0 million since its founding in 1999, in addition to the $0.2 million in musical works performance royalties that

a http://www.pandora.com

310-026 Pandora: Royalties Kill the Web Radio Star? (A)

2

Pandora had separately paid to songwriters and music publishers.3 But the new royalty rates seemed grossly unreasonable.

What additionally bothered Kennedy, Westergren, and others in the webcasting industry was the inconsistent application of the rates across competing radio services. While webcasters now faced steep rate hikes, over-the-air radio broadcasters still maintained their historical exemption from sound recording performance royalties, and satellite and digital cable radio paid rates significantly lower than those of webcasters.

Kennedy was incredibly proud of the loyalty and passion that Pandora had elicited among its millions of listeners. But would this passion be enough to save the company and web radio?

Web Radio

Webcasting, short for “web broadcasting,” involved the distribution of digitally encoded audio content over the Internet using streaming media technology. A descendant of over-the-air radio (alternatively referred to as “AM/FM,” “terrestrial,” or “traditional” radio) and satellite radio, webcasting began in the early 1990s. The first Internet-only commercial webcasters emerged in 1995. The revenue model established by these start-ups was, like that of traditional radio, largely advertising-based. However, unlike traditional radio, webcasters were not limited to audio commercials, but could also place display and text ads on their websites. In fact, many webcasters shunned audio commercials altogether in favor of providing a more listener-friendly experience. The low entry costs and potential revenue streams of this new business model brought hundreds of new entrants, large and small, in the years that followed. Increased broadband penetration in the early 2000s further boosted web radio’s growth. Between 2004 and 2006, average monthly listening hours for web radio increased at a compounded annual growth rate of 34%, and were expected to increase by nearly that much in 2007.4

Pandora emerged in the mid 2000s as part of a second generation of webcasters. These services were marked by a greater (and, in the case of services like Pandora, unlimited) number of channels, increased customization, and richer features. By 2007, major media and Internet companies owned many of the larger webcasters; the five major web radio companies besides Pandora included AOL, Live365, MTV, RealNetworks, and Yahoo. Exhibit 1 shows the relative market share of these companies between Q2’06 and Q1’07. The large commercial webcasters were members of the Digital Media Association (DiMA), the national association representing digital media companies. Other subgroups in web radio included small webcasters, non-commercial webcasters, and traditional radio simulcasters.

Webcasting was one of several online music services enabled by the Internet and broadband penetration. Other online music models included interactive services, including subscription on- demand services such as Rhapsody and ad-supported on-demand services such as Imeem, and à la carte download services such as iTunes. Webcasting differed from these other online music models by the limited control that web radio listeners had over the selection of specific songs played. Specifically, Pandora and other webcasters were designated non-interactive online music services, which under the provisions of the Digital Millennium Copyright Act of 1998 (DMCA) required that they abide by certain restrictions, including:

Interactivity. The service could not enable listeners to dictate the songs being played. The ability to select particular artists and skip songs was also limited.

Pandora: Royalties Kill the Web Radio Star? (A) 310-026

3

Transmission of copyright information. While the song was being played, the service was required to identify the song title, the album from which the song came, the featured artist, and (if feasible) the songwriter.

Performance complement. The service was required to limit performances, including the number of songs from an artist or album that could be played consecutively or in a set period of time.

Advance program schedules. The service could not publish an advance program schedule of when a specific artist, song, or album would be played.5 Webcasting challenged the historical relationship between traditional radio and the major record

companies. AM/FM radio was limited by the number of frequencies on the radio dial, and therefore could only broadcast so many songs to listeners in a given hour. This airtime was highly coveted by the major record companies, which competed to influence the playlists of radio stations. The result was limited music variety that made it particularly challenging for independent record labels and artists to promote their music.

By contrast, webcasters such as Pandora were able to broadcast diverse catalogues of music across an unlimited number of user-launched stations, while online retailers such as iTunes were able to carry a near universal inventory of digital songs and albums. In the midst of all these changes, the major record companies, while still retaining size and influence, grew increasingly marginalized.

Kennedy and Westergren viewed webcasting as a democratizing force in the new age of the music industry and the natural evolution of radio, which had progressed from AM/FM to satellite and digital cable, and now to the Internet. The primary opportunity for Pandora, they believed, was to gain listening share from other forms of radio. Web radio still paled in comparison to over-the-air radio. For example, only 19% of the United States population listened to web radio on a weekly basis as of January 2007, whereas over-the-air radio was essentially ubiquitous.6 However, the body of web radio listeners was growing rapidly. The “game changer,” highly anticipated by those in the industry, was the advent of mobile devices and wireless networks that would allow web radio to be streamed beyond the computer. As of March 2007, wireless network operators had not yet rolled out the infrastructure upgrades necessary to offer broadband data to most users, but they were scrambling to do so. The opportunity presented by mobile was especially exciting as it would allow webcasters to reach listeners in the car, where approximately 40% of aggregate radio listening took place.7

While online retailers like iTunes gained much of the early attention in the era of digital music, Westergren put things in historical perspective when speaking to the opportunity for companies like Pandora. “Broadcasting,” he pointed out, “has always been an industry multiple times the size of record sales.”8

Pandora Overview

Savage Beast and the Music Genome Project

Westergren, a musician and composer, teamed up with two friends in San Francisco to launch Pandora, initially called Savage Beast, in 1999. Under the guidance of Stanford University music professor and composer Nolan Gasser, the group began building a music discovery engine that they named the Music Genome Project.

Each song in the Music Genome Project library was required to have been dissected first by Pandora employees, who almost always had a background or training in music theory. These analysts spent 15-30 minutes codifying as many as 400 different attributes of what they called a song’s “musical DNA.” This DNA was used to match a song with its similarly coded “musical

310-026 Pandora: Royalties Kill the Web Radio Star? (A)

4

neighbors,” which ultimately allowed a listener to hear artists and songs based on an initial musical preference and mood. For example, a listener tuning into B.B. King’s “The Thrill is Gone” might subsequently hear Muddy Waters’ “Mean Old Frisco” due to its “similar blue roots, acoustic rhythm piano, major key tonality, an electric guitar solo, and a dynamic male vocalist.” Once a song was coded, it was placed in the project library. By March 2007, this library had grown to a collection of over 500,000 songs.9

Initially, Westergren licensed his music discovery concept to online and offline music stores. The Music Genome Project was the back-end to AOL’s and Barnes & Noble’s online music portals, and it also powered kiosks in record stores including Best Buy, Borders, and Tower Records. But license payments proved insufficient for building a profitable business. By 2004, the economic situation had become so bad that Westergren was forced to lay off workers.

Pandora.com

In late 2004, Pandora managed to pull together a Series B round of financing of $8 million. Concurrent with the investment, Westergren and the company’s board agreed that it was time to bring on a new CEO, allowing Westergren to assume the role of chief strategy officer. The company hired Joe Kennedy (HBS ’85), a consumer-oriented executive who had been the former president and COO of E-Loan and the VP of sales, service, and marketing for the Saturn division of General Motors. Upon reviewing the existing business model, Kennedy proposed that Pandora redefine its strategy to become a direct-to-consumer Internet radio service.

It was this idea that led to the rebirth of Pandora as a free and highly customized online radio destination based on the Music Genome Project. Listeners used Pandora by logging onto the company’s website and creating “stations” by entering favorite songs or artists into their user profile. Pandora then located the artist or song in its music library, often first playing a song by the artist and then streaming that song’s musical neighbors. While listening, users were able to give each song played a “thumbs up” or “thumbs down,” which allowed Pandora to further customize users’ listening experiences. (Exhibit 2 shows various Pandora screen shots.)

The reborn Pandora was launched in September 2005. With incredible viral growth and little marketing spend, the website’s use exploded, quickly gaining market share from its competitors (see Exhibit 1). Not only had the company obtained 5.7 million registered users by March 2007, but aggregate listening hours were increasing nearly 250% year-over-year.10 Pandora had also received more than 500 million instances of “thumbs up/thumbs down” feedback.

Between analyst input via the Music Genome Project and user input via the feedback system, Pandora amassed a massive database of information. Related to the underlying database taxonomy, the database-building processes, and the algorithms used on top of the database, the company had one patent issued and three applications outstanding as of March 2007.11

Pandora accepted music submissions from a variety of sources, including the musicians themselves, as inclusion on Pandora could have notable benefits. First and perhaps most importantly, musicians would be connected to listeners likely to appreciate their music. Second, musicians would receive compensation for performance of their works, as discussed below.

Business model

As of March 2007, Pandora earned revenue in three ways: advertising on its website, subscription fees from users who wanted to opt-out of advertising, and affiliate fees from iTunes and Amazon.com. Advertising revenue was by far the largest contributor, accounting for 93% of total

Pandora: Royalties Kill the Web Radio Star? (A) 310-026

5

revenue. This revenue was solely generated by display and text ads; Pandora did not run audio commercials.

Pandora offered advertisers several value propositions. Listeners actively interacted with music discovery features on Pandora’s website, giving advertisements a high level of visibility. More than 65% of these listeners were in the highly sought-after age range of 18-34. Pandora collected basic information from users when they registered (including age, gender, and zip code) and musical preference information after the music streaming began. This allowed Pandora to offer advertisers the opportunity to create targeted messages for different demographic groups. For example, a beer company expanding distribution to a new region could run display advertisements exclusively to males in their mid-20s in relevant zip codes with a partiality for alternative rock.

By March 2007, Pandora had the equivalent of 86 employees, including an 11-person advertising sales operation. About half of the company’s employees were music analysts. As the user base grew, revenues were beginning to catch up with run-rate expenses, which were largely comprised of advertising commissions, streaming costs, performance royalties, and employee salaries.12

Pandora had raised four rounds of financing from eight venture capital firms, and was targeting a Series E round in late 2007.13 It was important to Kennedy and Westergren that their investors shared their vision for the company.

Public Performance Royalties

Pandora and other webcasters were responsible for paying two types of royalties, each to separate parties in the music industry. These were musical works public performance royalties and sound recording public performance royalties. Other radio broadcasters were required to pay the same set of royalties, though sometimes in markedly inconsistent ways. (A schematic summarizing performance royalties in the radio broadcasting industry, based on the detailed discussion in the sections that follow, is included in Exhibit 3.)

Importantly, webcasters, as non-interactive online music services, also faced a different royalty framework than their online peers, such as interactive services and à la carte download services, described above. Interactive services and à la carte download services, for example, were often left to negotiate directly with copyright holders in order to use their songs and recordings.14 By contrast, Congressional rulemaking set the royalty rates for webcasters and other radio broadcasters.

Musical works performance royalties

Musical works performance royalties were paid to the songwriters, composers, and publishers of musical works. While songwriters were sometimes also the performing artists on a song’s recording, (e.g. Bob Dylan recording “Tangled Up in Blue”), this was not always the case (e.g. Sinead O’Connor recording Prince’s “Nothing Compares 2 U”). Publishers administered song rights on behalf of songwriters by finding users such as record companies or film producers, issuing licenses to these users, and collecting money. The income from these licenses was split, generally 50/50, between publishers and songwriters. Many of the largest publishers were owned by or affiliated with the major record companies, though there were also standalone publishers, including well-established songwriters who did their own publishing.15

In order to streamline the licensing process, songwriters and publishers affiliated with entities known as performing rights organizations (PROs). The PROs issued “blanket licenses” for entire catalogues to users such as radio, television, nightclubs, amusement parks, etc. They were

310-026 Pandora: Royalties Kill the Web Radio Star? (A)

6

subsequently responsible for collecting and distributing monies from these blanket licenses to songwriters and publishers. The major U.S. PROs were the American Society of Composers, Authors and Publishers (ASCAP, which claimed 330,000 affiliate members), Broadcast Music Incorporated (BMI, which claimed 375,000 affiliated members), and SESAC (formerly known as the Society of European Stage Authors & Composers). While the blanket licenses issued by the PROs were not statutory, the PROs were federally overseen due to monopolistic and restraint of trade concerns. Specifically, the PROs operated under consent agreements with the Justice Department and the Federal Communications Commission (FCC). Disputes over musical works performance royalties were heard by the rate court in the Southern District of New York.16

ASCAP and BMI, by far the largest of the PROs, offered their first licenses for online music in 1995.17, 18 Large commercial webcasters such as Pandora generally paid the PROs royalties based on a percentage-of-revenue calculation, subject to a minimum fee. As of March 2007, the webcasters paid 1.85% of their revenue to ASCAP and 1.75% of their revenue to BMI for the right to perform—to broadcast—the musical works in ASCAP’s and BMI’s catalogues. Musical works performance royalties expenses were therefore currently equal to about 3.6% of revenue for Pandora. The company was required to periodically provide performance statistics to the PROs so that the royalties could be appropriately allocated and distributed to songwriters and publishers.

Terrestrial, satellite, and digital cable radio broadcasters also had to license musical works from the PROs. While the specifics of the royalty rate calculation methodology varied for each party, all broadcasters, including web radio, ultimately and effectively paid comparable rates.

Sound recording performance royalties

Early history The practice of payola, in which record companies would pay radio stations to play their records (effectively a “negative royalty”), appears to have started in the big band era of the 1930s and 1940s. By the 1950s, the practice was so widespread that infuriated lawmakers passed, with limited effectiveness, anti-payola laws with heavy monetary punishment and potential jail time for violators. However, while the law made payola illegal to the detriment of broadcasters who had financially benefited from the practice, it appeased these same broadcasters by giving them an exemption on paying sound recording performance royalties to artists. Under this reasoning, the artists and record companies received free promotion for their material and the broadcasters received their “payola” in the free use of copyrighted recordings.19 This was further formalized in 1972, when Congress, in the process of amending federal law to protect sound recordings, made it clear that it had no intention to create a right for the artist or record company to be paid when their recordings were played on radio or television.20

While record executives could not dispute the powerful promotional benefit of radio, they couldn’t help but envy how large and profitable the radio industry became through the royalty- exempt use of their copyrighted recordings.

Digital Performance Rights and Sound Recording Act of 1995 The opportunity for record companies to reassert themselves came with the advent of satellite radio and digital cable in the 1990s. The Digital Performance Rights and Sound Recording Act of 1995 (DPRA) created, for the first time in the United States, a right for the artists and record companies to receive compensation when their records were performed. However, this right applied only to digital, subscription-based broadcasts—transmissions via satellite and digital cable.21

In enacting the DPRA, Congress sought to provide a new right and royalty (protecting and benefiting creators and copyright owners) and to promote efficient collective licensing processes (benefiting licensors and licensees), while constraining anticompetitive practices. After consulting

Pandora: Royalties Kill the Web Radio Star? (A) 310-026

7

with the Antitrust Division of the Department of Justice, Congress incorporated into the DPRA several provisions that furthered these objectives, including (i) a statutory license (rather than an exclusive right) to ensure the availability of blanket licenses to play music over new digital services, (ii) a limited antitrust exemption to promote efficient license negotiations, and (iii) the availability of a Copyright Arbitration Royalty Panel (CARP) as a backstop or safeguard to ensure that above-market royalties would not be imposed on licensees.

Integral to the safeguard provided by the CARP process were the standards to be used by the arbitrators to determine the appropriate rates and terms for the new statutory license. Congress adopted the standards set forth in the Copyright Act at 17 USC § 801(b) that balanced the interests of licensors, licensees, and the public.22 Known alternatively as the “801(b) standard,” “balanced standard,” or “four-factor standard,” these guidelines directed that “reasonable copyright royalty rates” fulfill four objectives:

• To maximize the availability of creative works to the public; • To afford the copyright owner a fair return for his creative work and the copyright user a fair

income under existing economic conditions; • To reflect the relative roles of the copyright owner and the copyright user in the product made

available to the public with respect to relative creative contribution, technological contribution, capital investment, cost, risk, and contribution to the opening of new markets for creative expression and media for their communication; and

• To minimize any disruptive impact on the structure of the industries involved and on generally prevailing industry practices.23

This standard was applied for satellite and digital cable radio services in the DPRA. Specifically, digital cable companies paid sound recording performance royalty rates of 7.25% of revenues which were initially set by an arbitration proceeding (and later locked in under a settlement through 2011).24 Satellite companies on the other hand initially paid sound recording performance royalties according to a private settlement which was reached when satellite radio operations began in the 2000s. A later decision of the government body charged with rate setting set royalties for satellite radio at 6.0% to 8.0% of revenues.25, 26

Digital Millennium Copyright Act of 1998 The emergence of Internet radio meant that royalty standards needed to be revisited, this time for webcasters. The Digital Millennium Copyright Act of 1998 (DMCA) did so. The DMCA first reestablished that non-subscription broadcasts licensed by the FCC, namely terrestrial radio and television, were exempt from sound recording performance royalties.27 Congress then addressed the issue of compulsory rights for webcasters. The recording industry, taking advantage of its size and power relative to the nascent web radio industry, lobbied hard with Congress, taking advantage of the more general fears that legislators had regarding digital music piracy, a key thrust of the DMCA. In response, Congress, this time without consulting the Department of Justice, created a new standard known as the “willing-buyer willing-seller” standard:28

In establishing rates and terms for transmissions by eligible non-subscription services and new subscription services, the [CARP] shall establish rates and terms that most clearly represent the rates and terms that would have been negotiated in the marketplace between a willing buyer and a willing seller. In determining such rates and terms, the [CARP] shall base its decision on economic, competitive and programming information presented by the parties, including:

310-026 Pandora: Royalties Kill the Web Radio Star? (A)

8

(i) whether use of the service may substitute for or may promote the sales of phonorecords or otherwise may interfere with or may enhance the sound recording copyright owner’s other streams of revenue from its sound recordings; and

(ii) the relative roles of the copyright owner and the transmitting entity in the copyrighted work and the service made available to the public with respect to relative creative contribution, technological contribution, capital investment, cost, and risk.

In establishing such rates and terms, the [CARP] may consider the rates and terms for comparable types of digital audio transmission services and comparable circumstances under voluntary license agreements.29

Reflecting on the royalty standards and rates that were set in place as of the DMCA, Kennedy couldn’t help but find humor in the “mess” that had resulted from decades of lobbying, arbitration, and legislation:

Radio’s delivered over four different pipes. You have traditional broadcasting, which is towers scattered all over the country, and satellite, which is really just a great big tower in the sky. AM/FM gets a complete exemption from royalties and satellite gets rates based on the balanced standard. Then you have digital cable and Internet, which is really just one cable running into your house, and you get the balanced standard for the end of the cable that runs through your set-top box and “willing-buyer willing-seller” for the other end that runs through your cable modem. It’s totally goofy. I mean, think about it.

2002 CARP arbitration The first CARP arbitration took place in 2002, setting web radio rates retroactive to 1998 and going forward to 2005. The CARP initially set rates at approximately $0.0015 per performance (i.e. per song per listener). The Copyright Office, which directly supervised the CARP, subsequently cut this rate in half to approximately $0.0008 as a result of a joint petition filed by interested parties.30 Pandora, which had not yet launched its web radio service, was not involved in this arbitration process. Generally, smaller webcasters (defined in the industry as webcasters with revenue under $1.2 million) did not have the money to participate, and only later began to negotiate adjusted rates based on a percentage of revenue under the Small Webcaster Settlement Act of 2002.31

The 2002 CARP arbitration was long and expensive, and it produced a result about which no party was satisfied, so Congress amended the rules in 2004. The royalty rates, which were originally to be revised every two years, would now be revised every five years. The CARP was replaced with a new government body called the Copyright Royalty Board (CRB), comprising three permanent judges who would hear cases related to various compulsory licenses.32 Notably, the Copyright Office would no longer supervise and review the statutory license royalty rate proceedings; full responsibility had been transferred to the CRB.33

2007 CRB decision On March 2, 2007, the CRB issued its first decision on a statutory license royalty rate, setting forth the rates, set forth below, that would apply to royalties for sound recording performances for the five-year period that had begun in 2006.34 (A “true-up” would take place to account for retroactive royalties.)

In the proceedings leading up to the decision, the webcasters and SoundExchange, the PRO that negotiated and collected royalties on behalf of artists and record companies, had submitted to the CRB their respective arguments and expert testimony regarding the appropriate benchmark agreements for determining the “willing-buyer and willing-seller” price in a hypothetical webcasting marketplace. The webcasters argued that the most appropriate benchmarks were the musical works performance royalties paid to ASCAP, BMI, and SESAC. These royalties were calculated as a

Pandora: Royalties Kill the Web Radio Star? (A) 310-026

9

percentage (approximately 3.6% in aggregate) of revenue. SoundExchange, on the other hand, maintained that the most appropriate benchmarks were the sound recording performance royalties found in the market for interactive online music services.

In response to these arguments, the CRB arrived at two verdicts. First, it decided in favor of a per- performance usage fee structure as opposed to a percentage-of-revenue model. Explaining its reasoning, the CRB cited the fact that, unlike a percentage-of-revenue metric, a per-performance metric would be directly tied to the nature of the right being licensed. It also cited measurement, definition, auditing, and enforcement challenges with the percentage-of-revenue model. Finally, it noted the desire to avoid a situation in which “copyright owners are forced to allow extensive use of their property without being adequately compensated due to factors unrelated to music use such as a dearth of managerial acumen at one or more of their [s]ervices.”35

Second, deciding in favor of SoundExchange, the CRB concluded that the most appropriate benchmarks were the sound recording performance royalties found in the market for interactive online music services. According to the CRB, the interactive market was a benchmark with characteristics reasonably comparable to non-interactive webcasting, including similar buyers and sellers, a similar set of rights to be licensed, and similar end consumers. The key difference between the markets was interactivity, which the CRB felt was appropriately accounted for by SoundExchange’s proposed adjustment to the benchmark to reflect the added value associated interactivity. The CRB also explained its reasoning for rejecting the music works marketplace as a benchmark, citing first the fact that the music works marketplace, relative to the sound recordings marketplace, involved a different set of sellers selling different rights, and secondly citing empirical evidence—rates for ringtones, digital downloads, music videos—that the CRB believed demonstrated that “sound recording rights are paid multiple times the amounts paid for musical works rights.”36

Adhering to these verdicts, the CRB announced the following rates for commercial webcasters, including simulcasters, for the five-year period that had begun in 2006:

2006: $0.0008 per performance

2007: $0.0011 per performance

2008: $0.0014 per performance

2009: $0.0018 per performance

2010: $0.0019 per performance

In a footnote, the CRB effectively disclaimed responsibility for the financial impact the newly set rates would have on the webcasters. In the view of its judges, the CRB had been tasked with applying the statutory standard and was not in a position to make “a policy decision.”37

Financial Impact

Despite the CRB’s disclaimer, Kennedy knew the newly announced sound recording performance royalty rates would have a serious financial impact on Pandora. Doing some back-of-the-envelope calculations, he pulled together an analysis, shown in Exhibit 4 and described below, to better understand the potential consequences for the company’s business model.

At the time, Pandora’s advertising sales team was achieving CPMs, or cost per thousand impressions, of approximately $10.00. Audience studies showed that Pandora listeners, on average,

310-026 Pandora: Royalties Kill the Web Radio Star? (A)

10

interacted with the website six to seven times per hour—for example, when they used the “thumbs up/thumbs down” feature or when they looked at the screen to see what artist was playing. The advertising sales team when mature would sell about 65% of advertising inventory. Based on these facts, Kennedy determined that each listener could eventually contribute $0.0423 of advertising revenue per hour. Average song length of 3:52 minutes meant there were about 15.5 song performances per listener per hour. The math therefore suggested that each performance contributed $0.0027 of advertising revenue.

Ignoring other streams of revenue, which were relatively minor for purposes of his analysis, Kennedy began to walk through the company’s variable expenses on a song performance basis. These variable expenses included advertising commissions, streaming costs, and musical works performance royalties. As laid out in his analysis, Pandora was achieving marginal profit of $0.0015 per performance before sound recording performance royalties. Including rounding, this translated to $0.0008 marginal profit per performance based on the current royalties rate of $0.0008.

However, based on the newly announced rate of $0.0011 for 2007, marginal profit contracted to $0.0004 per performance. Sound recording performance royalties, which would amount to 40% of revenue, would cut profit margins by 46%. Worse, that was just the 2007 rate, which would escalate at a compounded annual growth rate of 24% through 2010. The 2010 rate of $0.0019 yielded a projected marginal loss. In other words, Pandora would be losing money with each song performance! Of course, Kennedy hoped the company would achieve higher CPMs by 2010, but there was a limit to how much advertisers would pay, and he was uncertain whether the company could grow CPMs fast enough to overcome the potential losses.

Looking at these numbers made Kennedy even more perplexed about the CRB’s decision. Setting aside whether “willing-buyer willing-seller” was the appropriate standard to begin with, he wondered what circumstances would ever possibly compel Pandora or any other webcaster to be a “willing buyer” of sound recording performance royalties at a cost equaling nearly 70% of their revenue. The numbers were simply unsustainable and threatened to bankrupt the industry. Unfortunately, Kennedy knew that complaints about the CRB process would be of little comfort to the company’s investors.

Based on the company’s current fixed cost base, Kennedy had estimated that the company would be able to achieve breakeven at approximately 7.5 million users. By cutting marginal profit by 46%, the newly announced rate for 2007 raised that threshold to at least 13.8 million users, even higher if subsequent years were examined. Breakeven, which Kennedy had thought was right around the corner, was suddenly now distant. Pandora would require even more funding than had been anticipated.

How would he deliver this news to the company’s venture capital backers, who had already invested tens of millions of dollars over four rounds of fundraising? How would they react to the news that not only would profitability be delayed, but returns would be that much lower if and when profitability was achieved? The question, he feared, would not be whether to pull the plug, but when.

Interested Parties

Setting down his pen, Kennedy picked up his phone to call Westergren, who was in the middle of a week-long tour of town hall meetings with Pandora listeners across the country. “You’ve heard the news?” Kennedy asked.

Pandora: Royalties Kill the Web Radio Star? (A) 310-026

11

“I was on the bus when I got the message,” said Westergren. “The only thing I could think was, ‘We’re dead.’ I ended up missing my stop and arriving a half hour late to today’s event. What are we going to do?”

“Right now, I think the best thing is for us to take a step back and reassess the situation,” replied Kennedy. “For starters, let’s take another look at everyone involved—their representatives, their interests, and the cases we think we can make to them.”

Westergren agreed, and over the next hour, they mapped out the landscape.

Congress

With 26 years of practice in business, Kennedy spoke to the lesson he was taking from this experience: “I think the real story here is the massive mismatch of political influence. A new industry develops with entrepreneurs completely ill-equipped to win a lobbying fight with the RIAA, who’s been in Washington for decades. And so the RIAA moves smartly and powerfully to establish the rules while they still have a massive upper hand in congressional power.”

Westergren forwarded a news release from an industry publication that echoed Kennedy’s thoughts, an industry commentator in the article stating that, “The only thing I can think of that sets [web radio] apart from [other services like satellite] is that they can afford…to defend themselves against the predatory attacks of the major music conglomerates, and we cannot.”38

However, both Kennedy and Westergren knew from their conversations with Jonathan Potter, executive director of DiMA, that some Congressmen were beginning to sympathize with the plight of web radio relative to other radio platforms. Recording companies and artists, meanwhile, had proposed legislation that would require traditional radio broadcasters to finally begin paying sound recording performance royalties. Yet achieving “platform parity”—having all radio platforms subject to the same royalty standards and comparable royalty rates—would require a formidable lobbying and legislative push, especially given the resistance of influential AM/FM operators to any scenario that threatened their royalty exemption.

Webcasters and the Digital Media Association

DiMA represented the coalition of large commercial webcasters, which included Pandora, AOL, Live365, MTV, RealNetworks, and Yahoo. Small webcasters and non-commercial webcasters were represented separately. For small webcasters, the CRB’s decision was especially injurious, as their ability to use a percentage-of-revenue calculation, as had been established under the SWSA, was eliminated. Many of these small webcasters were mom-and-pop businesses on which their operators depended for their livelihoods. The large webcasters did not necessarily object to a different methodology for small webcasters, so Kennedy felt the respective parties could work together well.

There was some degree of suspicion, however, between “pure” webcasters and AM/FM simulcasters. AM/FM operators were hypersensitive to the talk among webcasters about platform parity. Kennedy joined others in noting that the National Association of Broadcasters (NAB), the industry association for terrestrial radio and television, had been slow to condemn the new rates: “Ninety eight percent of terrestrial radio’s business is generated from royalty-free over-the-air broadcasting. Some wondered whether they’d be willing to fund losses on the other 2% of their business if it cleared out this upstart competition for them.”

310-026 Pandora: Royalties Kill the Web Radio Star? (A)

12

SoundExchange, Record Companies, and Recording Artists

SoundExchange generated obvious benefits for its members by maximizing sound recording performance royalties. However, the three constituencies that SoundExchange represented—the major labels, independent labels, and artists—each had different interests. Kennedy and Westergren felt that these parties stood to lose if web radio went out of business.

For recording companies, Pandora represented a powerful promotional vehicle to drive music sales and legitimate music consumption, which was especially valuable in the era of digital music piracy. Westergren reeled off a few facts:

• Pandora was one of the top referral sites for music sales on iTunes and Amazon.com, online retail stores which represented a growing share of revenue for the industry.39

• A Nielsen/NetRatings study had concluded that Pandora listeners were three to five times more likely to have purchased music in the last 90 days than the average American.40

Kennedy and Westergren felt that Pandora represented an especially valuable opportunity to independent labels and artists, who were not affiliated with any of the four major recording companies.. More than 50% of song performances on Pandora came from independent musicians, compared to less than 10% on AM/FM radio. Seventy percent of the sound recordings in the Music Genome Project came from artists not affiliated with major labels. There were no prerequisites for inclusion in the library. If an analyst received an amateur, homemade recording and liked it, it would be added. Pandora, Westergren felt, represented a democratizing force for creativity. He had told friends, “I often wish I could start my band now instead of back in the early nineties when our only resources were a van, a staple gun, and a pile of flyers that we handed out or stapled to telephone poles.”41

Web radio listeners

In just one year, Pandora’s registered user base had grown from 1.1 million to 5.7 million, with the average listener tuning in for 2.5 hours each day.42 This represented, as Westergren put it, an “explosion of accessible music diversity” in which listeners could discover artists and songs that would otherwise be invisible to them.43 Tim read off a particularly touching e-mail he had recently received from an older Pandora user:

Let me tell you that you are a blessing in my life. I’m 77 years old and the music I like and grew up with just isn’t played much anymore. Sometimes tears come to my eyes when I hear certain songs [on Pandora]. They bring back so many memories. I don’t think I have heard any songs I haven’t liked. Thank you from the bottom of my heart. I send you arms full of appreciation.44

Tim had visited nearly 100 cities over the previous 18 months conducting town hall meetings with Pandora listeners. He knew this kind of enthusiasm and gratitude was not uncommon. But how could it be leveraged in a highly legalistic struggle overseen by an obscure government body?

The Campaign Begins

After wrapping up his call with Westergren, Kennedy turned to check his e-mail. He noticed a priority-flagged message from Jonathan Potter of DiMA, requesting a call in a half-hour’s time. Kennedy had become one of Potter’s closest confidantes in the web radio industry, not just because of Pandora’s meteoric rise, but because of Kennedy’s ability to think sharply and strategically about the

Pandora: Royalties Kill the Web Radio Star? (A) 310-026

13

issues at hand despite having initially come to web radio from outside the industry. Kennedy knew that Potter would be calling to solicit his thoughts on their options, and expected that DiMA would be leaning heavily on both him and Westergren to support its strategy.

DiMA had the ability to request a rehearing of the decision with the CRB. This seemed like a no- brainer.b The webcasters also had the ability to appeal the CRB’s determination to the United States Court of Appeals for the District of Columbia Circuit, an appeal which would need to be filed within 30 days after the publication of the determination in the Federal Register. The judicial review process, however, could take over a year, and the prospects of successfully getting a court to block the decision of the CRB were slim.45

In addition to a request for rehearing and a potential judicial appeal, Kennedy believed there might be other alternatives for the webcasters. Taking another look at the list of interested parties that he had crafted with Westergren, Kennedy felt confident that DiMA and Pandora could strengthen their negotiation position. But who were the appropriate parties to enlist and address? And how should various campaigns be sequenced?

The ringing phone interrupted Kennedy’s thoughts. It was Jonathan Potter. “Time to fight the good fight,” he said to himself, leaning over his desk to answer.

b On March 19, 2007, DiMA would in fact proceed with a formal request that the CRB reconsider the validity of its royalty rate determination. DiMA also asked that the CRB revise two technical aspects of its decision, one related to minimal royalties per channel and the other related to rate calculation methodology. On April 16, 2007, the CRB denied all motions for rehearing, granting leniency only on DiMA’s technical request related to rate calculation methodology.

310-026 Pandora: Royalties Kill the Web Radio Star? (A)

14

Exhibit 1 Relative Market Share of Large Commercial Webcasters, Q2’06-Q1’07

Market share by total unique visitors (% )

54% 53% 52%

50%

34% 32%

30%

27%

5% 8%

10%

16%

7% 7% 8% 8%

0%

10%

20%

30%

40%

50%

60%

Q2'06 Q3'06 Q4'06 Q1'07

Yahoo AOL Pandora Live365

Source: Created by casewriters based on data provided by comScore in December 2008.

Note: Data unavailable for MTV and RealNetworks.

Pandora: Royalties Kill the Web Radio Star? (A) 310-026

15

Exhibit 2 Pandora Screenshots

Menu for creating a new station

Music player

310-026 Pandora: Royalties Kill the Web Radio Star? (A)

16

Explanation for why song was selected and artist details

Links to purchase music at affiliate websites

Source: www.pandora.com, accessed May 2009.

Pandora: Royalties Kill the Web Radio Star? (A) 310-026

17

Exhibit 3a Musical Works Performance Royalties in the Radio Broadcasting Industry, 2007

B ro

ad ca

st er

s P

R O

s C

o p

yr ig

ht h

o ld

er s

AM/FM radio

Satellite and digtal

cable radio

Web- casters

ASCAP BMI

SESAC

3.0-4.0% of revenue

3.0-4.0% of revenue

3.0-4.0% of revenue

Note: Licenses not statutory, but federally overseen.

Publishers Songwriters Composers

Source: Casewriter.

310-026 Pandora: Royalties Kill the Web Radio Star? (A)

18

Exhibit 3b Sound Recording Performance Royalties in the Radio Broadcasting Industry, 2007

B ro

ad ca

st er

s P

R O

s C

o p

yr ig

ht h

o ld

er s

AM/FM radio

Satellite and digtal

cable radio

Web- casters

Sound Exchange

Record companies Artists

“W-B W-S standard” $ per performance (15.0-200.0% of revenue)

“Balanced standard”

Note: Licenses statutory, overseen by the Copyright Royalty Board (CRB).

“Broadcast exemption”

6.0-8.0% of revenue

Source: Casewriter.

Pandora: Royalties Kill the Web Radio Star? (A) 310-026

19

Exhibit 4 Joe Kennedy’s Back-of-the-Envelope Analysis

Revenue per performance

CPM $10.00 Cost per single impression 0.0100 Interactions (i.e. impressions) per listener per hour 6.5 % of advertising inventory sold 65.0% Interactions (i.e. impressions) per listener per hour with advertising 4.2 Advertising revenue per listener per hour $0.0423 Song performances per listener per hour 15.5 Advertising revenue per song performance $0.0027

Marginal profit per song performance

$ %

Advertising revenue per song performance $0.0027 100.0%

Variable expenses (% of advertising revenue): Advertising commissions $0.0005 20.0% Streaming costs 0.0005 20.0% Music works performance royalties 0.0001 3.5% Other variable expense 0.0000 1.0%

Total variable expenses $0.0012 44.5%

Marginal profit per performance before sound recording performance royalties $0.0015 55.5%

Sound recording performance royalties - Current $0.0008 27.5% Sound recording performance royalties - CRB proposal for 2007 0.0011 40.4% Sound recording performance royalties - CRB proposal for 2010 0.0019 69.8%

Marginal profit per performance - Current $0.0008 28.0% Marginal profit per performance - Pro forma for 2007 0.0004 15.1% Marginal profit per performance - Pro forma for 2010 (0.0004) (14.3%)

Breakeven

Estimated number of users required for breakeven - Current 7,500,000 Estimated number of users required for breakeven - Pro forma for 2007 13,884,661 Estimated number of users required for breakeven - Pro forma for 2010 (1) N/A

Note 1: Analysis implies that company is losing money with each additional user

Source: Joe Kennedy.

Note: Data and analysis are illustrative.

310-026 Pandora: Royalties Kill the Web Radio Star? (A)

20

Endnotes

1 Roger Meddows Taylor, “Radio Ga Ga,” song lyrics, Beechwood Music Corporation, BMI Work #1223436, http://www.bmi.com/, accessed July 29, 2009.

2 Joe Kennedy, “Pandora Case,” e-mail message to Alex Rosenfeld, January 19, 2009.

3 Ibid.

4 “AccuStream Research: Internet Music Radio in 4th Consecutive Year of Expansion,” AccuStream press release, May 16, 2006, on AccuStream website, www.accustreamresearch.com/news/20060516.html, accessed April 2009. “AccuStream Research: Internet Music Radio and Track Plays Chart 4.8 Billion Listening Hours in 2007,” AccuStream press release, February 21, 2008, on AccuStream website, www.accustreamresearch.com/ news/20080221.htm, accessed April 2009.

5 David D. Oxenford, “Internet Radio – The Basics of Music Royalty Obligations,” Davis Wright Tremaine LLP, www.dwt.com/LearningCenter/Advisories?find=56047, accessed April 2009; 17 USC § 114(d)(2)(C).

6 “Bridge Ratings Industry Update – Internet Radio,” Bridge Ratings press release, February 21, 2007, www.bridgeratings.com/press_02.21.07.Internet%20RadioUpd-.htm, accessed April 2009.

7 Joe Kennedy, “Pandora Case,” e-mail message to Alex Rosenfeld.

8 Tim Westergren, “Building and Sustaining Successful Enterprises,” MBA class discussion, November 18, 2008, Harvard Business School, Boston, MA.

9 Joe Kennedy, “Pandora Case,” e-mail message to Alex Rosenfeld.

10 Ibid.

11 Ibid.

12 Ibid.

13 Ibid.

14 Oxenford, Davis Wright Tremaine LLP, “Internet Radio.”

15 Donald S. Passman, All You Need to Know About the Music Business, Sixth Edition (New York: Simon & Shuster, 2006), pp. 206-211.

16 Joe Kennedy, interview by author, Oakland, CA, October 31, 2008.

17 “ASCAP’s New Media & Internet Licenses,” ASCAP, www.ascap.com/weblicense, accessed April 2009.

18 “BMI New Media: Webcasters,” BMI, www.bmi.com/newmedia/entry/C1168/pdf535475_1, accessed April 2009.

19 Doug Perlson, “Payola: Could an Old Idea Save Online Radio and the Music Industry?” Silicon Alley Insider, September 8, 2008, www.alleyinsider.com/2008/9/could-payola-save-online-radio, accessed April 2009.

20 Passman, The Music Business, pp. 292-294.

21 Ibid.

22 U.S. House Subcommittee on Courts, the Internet and Intellectual Property, Committee on the Judiciary; Internet Streaming of Radio Broadcasts, testimony of Jonathan Potter, Executive Director of the Digital Media Association, July 15, 2004, www.digmedia.org/docs/7-15-04%20H-JUD%20Written%20Testimony%20FINAL .DOC, accessed April 2009.

23 17 USC § 801(b).

Pandora: Royalties Kill the Web Radio Star? (A) 310-026

21

24 “Copyright Royalty Board Asks for Comment on Music Choice Royalty – Satellite Radio Is Next,” November 8, 2007, post on blog “Broadcast Law Blog,” www.broadcastlawblog.com, accessed April 2009.

25 “Satellite Radio Music Royalty Reconsideration Denied by Copyright Royalty Board – What a Difference a Standard Makes,” January 12, 2008, post on blog “Broadcast Law Blog,” www.broadcastlawblog.com, accessed April 2009.

26 David D. Oxenford, “Satellite and Digital Cable Sound Recording Performance Royalties,” e-mail message to Alex Rosenfeld, February 26, 2009.

27 Passman, The Music Business, pp. 295-297.

28 U.S. House Subcommittee on Courts, the Internet and Intellectual Property, Committee on the Judiciary; Internet Streaming of Radio Broadcasts, testimony of Jonathan Potter.

29 17 USC § 114(f)(2)(B).

30 Library of Congress Copyright Office, Digital Performance Right in Sound Recordings and Ephemeral Recordings, Federal Register, Vol. 68, No. 84, May 1, 2003

31 Joe Kennedy, interview by author.

32 Davis Wright Tremaine LLP, “Internet Radio.”

33 U.S. Copyright Office, “Copyright Arbitration Royalty Panels (CARP),” www.copyright.gov/carp/index.html, accessed April 2009.

34 Determination of Rates and Terms in the Matter of Digital Performance Right in Sound Recordings and Ephemeral Recordings, United States Copyright Royalty Judges, Docket No. 2005-1 CRB DTRA (March 2, 2007); Final Determination of Rates and Terms in the Matter of Digital Performance Right in Sound Recordings and Ephemeral Recordings, United States Copyright Royalty Judges, Docket No. 2005-1 CRB DTRA (April 23, 2007).

35 Determination of Rates and Terms, p. 24; Final Determination of Rates and Terms, p. 25.

36 Determination of Rates and Terms, p. 40; Final Determination of Rates and Terms, p. 41.

37 Determination of Rates and Terms, p. 19, n.7; Final Determination of Rates and Terms, p. 20, n.8.

38 Daniel McSwain, “Webcast Royalty Rate Decision Announced,” Radio and Internet Newsletter, March 2, 2007, www.kurthanson.com/archive/news/030207/index.shtml, accessed April 2009.

39 U.S. Senate Committee on Commerce, Science and Transportation; The Future of Radio, testimony of Tim Westergren, Founder and Chief Strategy Officer, Pandora Media, on behalf of the Digital Media Association, October 24, 2007, http://commerce.senate.gov/public/_files/TimWestergrenFINALFINAL1022505pm.pdf, accessed April 2009.

40 Ibid.

41 Ibid.

42 Joe Kennedy, “Pandora Case,” e-mail message to Alex Rosenfeld.

43 U.S. Senate Committee on Commerce, Science and Transportation; The Future of Radio, testimony of Tim Westergren.

44 Ibid.

45 “Copyright Royalty Board rejects appeals from Web broadcasters,” April 16, 2007, MarketWatch, www.marketwatch.com/news/story/copyright-royalty-board-rejects-appeals/story.aspx?guid=%7BD848DFF4 %2DD20C%2D4EF1%2D8E83%2DA55833C13922%7D&dist=msr_2, accessed April 2009.