Song Maestro - McMaster University

advertisement
Song Maestro
Jace Silva, a budding entrepreneur from Mississauga, Ontario, was preparing to appear
on the American reality television show Shark Tank. Like a similar Canadian television show –
Dragon’s Den – entrepreneurs pitched their business ideas to successful businesspeople and
venture capitalists hoping to win an investment. Jace didn’t like to boast but his tone was
confident and assured. “The music industry has always used two criteria to determine if a song
will be a success: 1) it sounds like a hit; and 2) the song and artist can successfully be brought to
the market. While the industry pays professionals to determine if a song sounds like a hit, they
only have a 10% success rate – only one in ten songs that get promoted ever hits the charts. The
new technology that I invented adds a third criterion – that it has optimal mathematical patterns.
With it, I can significantly increase success rates. Song Maestro generated unusually high scores
for Adele, a British singer who most industry insiders expected to have limited commercial
impact but whose album later rose to the top of the charts. It also correctly predicted each of the
hits of the band Maroon 5.”
To date, Silva had invested nearly US$400,000 of his own funds developing Song
Maestro (much of it to legally acquire a database of songs). He felt he needed US$100,000 to
bring the service to market though he still was not certain of the ideal target customer for the
service. He hoped to find an investor willing to give him the money for 20% of his company.
Even if his appearance failed to gain an investor, he hoped the exposure would either generate
sales leads or other investors. He was not worried about attracting copycats. Though he could
not patent Song Maestro, he was certain no one could reproduce his technology.
The Song Maestro Technology
To create Song Maestro, Silva – with Ph.D.’s in both mathematics and engineering –
analyzed thousands of songs released by music labels since the 1950’s. His database was
updated weekly with new releases. He devised a way to “listen” to a piece of music and isolate
particular patterns. The process, known as “spectral deconvolution,” considered over 25
characteristics including melody, harmony, tempo, pitch, octave, beat, rhythm, duration, fullness
of sound, noise, sonic brilliance, and chord progression. Based on its mathematical
characteristics, each song was then mapped onto a multidimensional grid called the “music
universe.” Songs with mathematical similarities were positioned very close to one another in
this universe. This organization could create groupings which appeared counter-intuitive. As
Silva explained, “a Beethoven composition could fall very close to a song by rock band U2 or
pop singer Mariah Carey.”
Using this feature of Song Maestro alone, it would be possible to develop a music
recommendation system. When a shopper entered a music store (or its online equivalent), s/he
could share the names of some favourite songs. Song Maestro could see where these songs fell
---------------------------------------------------------------This case was written by Marvin Ryder. Case material is prepared as a basis for classroom
discussion only. Copyright 2012 by Marvin Ryder, DeGroote School of Business, McMaster
University, Hamilton, Ontario. This case is not to be reproduced in whole or in part by any means
without the express written consent of the author.
2
in the “music universe” and then recommend other music with similar mathematical patterns. “I
believe that people don’t just like a certain genre of music but that they like specific
mathematical patterns that transverse music genres. Genres are just old-fashioned, marketing
terms,” stated Silva. Song Maestro could be licensed to physical retailers (like Best Buy or
Virgin Megastores) or e-tailers (like iTunes) for up to US$250,000 a year. Silva felt that the cost
would be more than offset by increased sales.
Because so many physical record stores had closed in the 2000’s, Silva took his analysis
to a new level. He looked at songs that had made it to the Billboard Hot 100 list (compiled by
Nielsen Broadcast Data Systems and Nielsen Soundscan and published in the American music
trade publication Billboard Magazine). He found that hit songs had common mathematical
properties – there were only about 60 hit “clusters” in the music universe. The extent to which
new releases “fit” those clusters should indicate their hit potential. Using Song Maestro, the
closeness to a cluster was indicated with a “hit” score on a scale of 1 to 100 (he borrowed this
approach from the old American Bandstand television show) with higher scores indicating a
greater hit potential. The clustering technique also allowed Song Maestro to provide insights
into the coherence of an album, that is, the extent to which songs fell into the same or nearby
clusters. It could produce a list of the album’s songs with similar mathematical properties.
According to Silva, “I can take an unreleased album and examine how the songs on that
album map onto the clusters. If a song falls within one of these clusters, I can’t necessarily say
that it will be a hit. I just know that it has the potential. The song must still sound like a hit, be
promoted like a hit, and the artist must be marketed correctly. But if a song falls outside of the
clusters, I know it will probably not become a hit. Aggregating the values from all the songs, I
can also make predictions about the popularity of the entire album.”
By giving higher weights to songs with higher sales levels, Silva was able to determine
the popularity of various clusters. He could also segment the database by year so he could adjust
clusters for changing music tastes. Generating a report on an album of ten to twelve songs was
relatively cheap and quick – it cost him US$300 for the two hours of analysis.
The Music Industry
Worldwide revenues for CD’s, vinyl, cassettes and digital downloads fell from US$36.9
billion in 2000 to US$15.9 billion in 2010. Geographically, the largest market was the United
States. (See Figure 1) About 30% of all recorded music was produced there and the retail
market was roughly 40% of worldwide sales. European countries accounted for 34% of
worldwide sales with the United Kingdom, Germany, France, Italy and Spain being the largest
component markets. While music was as popular as ever, industry insiders attributed the decline
in revenues to online and offline piracy.
The music recording process typically started with the artists who wrote song lyrics,
composed music, and performed it. Recognized “talent” and some new artists had contracts with
a “label” within a record company that stipulated the terms under which they were to deliver one
or more albums over a certain period and work exclusively for that label. They were supported
by legal advisors, managers, and agents who helped negotiate contracts, book concerts, and
3
Figure 1 Regions' Share of Music Sales
Latin America
3%
Other
4%
Asia
18%
North America
41%
Europe
34%
schedule recording sessions. Experts estimated that there were 10,000 artists with a record
contract in the U.S. and Europe but only several hundred with some name recognition and
commercial success. Hundreds of thousands of individuals or groups were hoping to secure such
a contract. “Every high school has a band who thinks it will be the next big thing,” said Silva.
Unsigned artists often used “demo” (demonstration) recordings to attract a music publisher’s
interest. It was estimated that record companies received 500 demos each week but less than 1%
led to record contracts.
There were five big record companies, “the majors,” that controlled 75% of the global
recorded music business: BMG Entertainment, EMI, Sony Music, Universal Music Group, and
Warner Music Group. (See Figure 2) Each of those companies had various labels and music
publishing companies under their umbrellas. For example, Universal Music Group incorporated
at least a dozen labels aimed at the U.S. market including Motown, Interscope, Geffen, MCA,
Universal Classics, Universal Records, Universal South, and Universal/Island. It had three
dozen labels aimed at international markets. Some labels covered all music types while others
specialized in music genres. In the U.S. alone, there were a thousand small and mid-sized
recording companies. When labels signed an artist, they typically had an in-house A&R (artistand-repertoire) producer guide the project. These people were typically young, enjoyed music,
and were believed to have “good ears” (i.e., that they could pick hits). A&R producers looked
for talent, found songs that they could record, selected the best producer for each project,
participated in the recording process, and checked the master recordings. Each A&R producer
mentored 20 artists through their early career development and had a vested stake in their
success. Most A&R producers only lasted three to four years because of lack of success. Those
with greater success had a longer career span and became well known in the music industry.
Established artists generated 90% of a record company’s sales but without successful new talent
revenues would decline over time.
4
Figure 2 Major Record Companies' Share
of Music Sales
Independents
24%
BMG
11%
EMI
12%
Sony
14%
Warner
14%
Universal
25%
Artists could also work with independent producers. Like A&R producers, they would
help the artist select music and develop a musical style, oversee recording schedules, recruit
sound engineers, and watch over recording budgets. Perhaps an independent producer said it
best, “A producer can be responsible for a lot of things – writing the song, recording the song,
mixing and editing it, making sure it has the right vibe. The producer is the pivot between the
label, the publisher, the manager, and the artist. A good producer can see what an artist is
supposed to sound like and come up with the right sound, or take an artist who has a vision and
help him or her realize it. I once saw a performance by a young woman and had a strong feeling
that she could be a star. I found a sound that worked for her and helped her score several huge
hits.” There were 20 to 30 producers who were responsible for the majority of Top 40 success, a
larger group of a few hundred producers who had a hit once in a while, and thousands of people
trying to establish themselves as producers.
The Role of Marketing
The recording process ended with the delivery of a completed master copy of the album.
At this point, the record company planned a promotional campaign for the album (or individual
singles) and established a timeline for the physical or virtual production and distribution of the
records. Generally, the record label president and marketing director made final decisions about
promotional strategies but the A&R person often acted as an influencer/advocate on behalf of the
artist. One of the most important decisions for a label was which single of an album (usually
containing ten to twelve songs) to release first. Radio airplay was the primary advertising
vehicle for popular albums. Music television was also an important advertising channel. A
strong first single – particularly for new artists – was often a make-it-or-break-it decision. If the
first single underperformed, the album could be left to “die” – the company would not release
any more singles from the album and gave up on the advertising campaign. It could be the end
of the artist’s career – a no-hit wonder!
5
Promoting music was expensive – the release of a single from an unknown artist involved
at least US$300,000 in promotional expenditures. Campaigns involved anything from the
production of music videos to concert tours, cooperative advertising with retailers, in-store
merchandising materials, radio and television commercials, websites, social media, and press
kits. To promote the song and get it played on the radio, free records and digital recordings were
sent to hundreds of radio stations. Popular music stations only added four to six new songs per
week to their play lists so competition for airplay was intense. For established stars, the
promotion budget could swell to US$1 million. Labels did not expect to make that money back
on single sales. The album delivered the largest share of revenues. Typically, the album and
first single were released on the same date.
Recorded music was distributed either in a physical form (compact disk) or digital form
(say through iTunes). Each year approximately 30,000 new albums were brought to the market.
Not all of those albums were aimed at a broad, mainstream audience – about 2,500 albums
released each year were accompanied by one or more singles. The physical distribution arms of
the five majors accounted for the majority of shipments of discs.
Revenues and Costs
The suggested retail price to consumers for a physical album was just under $17 in the
U.S.; a physical music retailer paid about $10.50 for a CD. Physical CD singles were priced
around $5.00. When distributed online, a single sold for between $1.00 and $2.00. Record
labels sought a profit margin of 30% of the price to retailers after paying royalties to artists, fees
to the publisher (about 5% of the price to retailers), manufacturing and distribution costs (about
10% of the price to retailers), A&R expenses (about 15% of the price to retailers), administrative
expenses (about 10% of the price to retailers), and marketing and promotion costs.
Music generated three different royalty streams: 1) from the sale of music recordings; 2)
from performance of the music (by radio stations, orchestras, nightclub singers, etc.); and 3)
from accompanying visual images (say in a movie). A record company typically paid an artist
an up-front fee and, once sales had passed a level that allowed the company to recover its costs, a
royalty – between 5% and 15% of a record’s retail price depending on the artist’s track record
and stature. Industry contracts generally dictated that most of the costs of making a record were
to be repaid out of the artist’s up-front fee and royalties. Managers and talent agents were paid
25% to 40% of a performer’s income. Outside producers received a production fee and often a
royalty of 1% to 5% of retail selling price. Production fees varied dramatically – from
US$100,000 for a first record for a new artist to US$1,000,000 for an established artist.
In a typical year, in the U.S., the top 25 albums sold 3,000,000 units on average. The next
40 albums sold an average of 1,000,000 units. Industry experts estimated that less than 15% of
music titles released were profitable. In the U.S., 3,000 singles were released annually and only
10% made it to the Billboard Singles Top 40 list. 2,500 albums were released annually and only
250 had a single that made it onto the Top 40 list (of course, a few had more than one single
make it to the list). While 10% success was an industry average, the best A&R producers could
sometimes average 13% to 16% success. Figure 3 shows estimated U.S. revenues for singles and
albums under different scenarios.
6
Figure 3 Estimated Revenues for Singles/Albums With & Without a Top 40 List Position
Expected Revenues for …
Single not reaching the Top 40 list
Single reaching Top 40 list
Album with single not reaching the Top 40 list
Album with single reading the Top 40 list
Low
Estimate
$0
$100,000
$0
$300,000
Medium
Estimate
$10,000
$200,000
$90,000
$2,000,000
High
Estimate
$100,000
$2,000,000
$300,000
$40,000,000
Most record companies engaged in research activities to help them forecast sales levels.
“Everyone has their own technique,” remarked Silva. “One top record-label executive likes to
round up kids on the streets of New York and let them listen to new artists and new songs.
Another executive prides himself on his ‘ears’ – unless it sounds like a hit to him, it is not, and
he does not want to know what anybody else thinks. While a few companies rely completely on
‘gut instinct,’ most do some testing. Focus groups can cost record companies US$10,000 per
song. A more popular technique is ‘call-out’ research. Enlisting the help of a market research
company, potential respondents are contacted by phone at home. After screening for
demographics, music listening behaviour, or other characteristics, respondents participate in a
music ‘test.’ They are played 15 to 30 second fragments (sometimes called hooks) of songs and
are then asked to rate them either by pushing a button on their phone or verbally giving a score to
a staff member. Depending on the number of people surveyed and the method used, these
studies cost between US$5,000 and US$7,000 per song. The Internet has reduced the cost of this
research to US$3,000 per song.”
The vast majority of music released never became a hit. As one executive noted,
“Releases in the music business traditionally are a big gamble. We can spend millions of dollars
to put a product on the market and not be sure whether there will be any demand for it or whether
it will get any airplay. Las Vegas gives you better odds than the music industry. We might as
well just put a few million dollars on red and spin the wheel.” For each label, a few successes
covered the losses made on many failed titles.
Silva argued, based on the results of initial beta tests, that he could give the industry
much better success rates. “Song Maestro had an 80% success rate – it correctly predicted
whether a single would reach the Singles Top 40 List eight out of ten times,” Silva boasted. “If a
song scores a rating of 70 out of 100, I believe it has the potential to become a hit.”
Despite an extensive search, Silva was not aware of any comparable product or service
on the market.
The Target Market Question
Silva felt that there were four target markets who could use this technology:
1) Record label – helps the company decide whether to launch an album, which singles to
release first, and which new artists to sign to a record deal;
7
2) Producers – can test songs or albums at some stage of the production process allowing for
tweaks, additions or deletions before launch;
3) Unsigned artists – can test the hit potential of their songs, improve the quality of their
demos, and improve their chances of being signed to a contract; and
4) Music retailers and e-tailers – allows customers to find new music that they will instantly
like leading them to be more satisfied and, possibly, to buying more music.
Silva had done limited marketing research. “When we tell music executives about the
concept, they typically look at us with glazed eyes, check their watch, and then think of an
excuse why they need to leave as soon as possible. Many people simply cannot imagine that
science could play a role in picking hit songs. Song Maestro is to the music industry what the xray machine was to medicine. The first time someone told a doctor that he could look inside a
patient’s body without cutting it open, it probably sounded like science fiction. But in the end,
the x-ray machine is a tool that helps the doctor see something that he could not see before, and
he can use that information to make better decisions. That is exactly what Song Maestro does
and that is what matters. I am just a millimeter away from this thing taking off.”
The manager of one artist disagreed, “What creates a hit is that people have an emotional
reaction to a song, particularly the lyrics. I can’t believe that a machine could gauge that.” A
musician noted, “I doubt pop music could get any worse. This is a meaningless tool.” A third
expert argued, “There are always musicians who will do something different. They could be
missed if the industry retires A&R people and relies too much on a data-crunching machine.”
However, not all executives were negative. One beta tester noted, “This business has
always been run by instinct and gut, and even my own colleagues might have a hard time
believing this, but my tests with Song Maestro have been fantastic. It was extremely accurate
picking the tracks that we have taken to commercial radio.” Another beta tester noted, “When
this service came along, I was skeptical. I put together a ‘crazy’ album with some really bad
songs and some really good ones, just to see what would happen. The report generated by Song
Maestro clearly showed the duds and the hits. Now that I have tested it, I realize that this is an
amazing tool for producers.”
While Silva felt all four targets could benefit from Song Maestro, he needed to generate
some quick sales. If he could convince a venture capitalist to invest US$100,000, he would be
able to hire enough salespeople to target only one of these markets. It was important to choose
the market that both had the greatest likelihood to buy and could generate significant revenues.
For the first three markets, Silva had decided that he would only sell reports – he would not
license the technology. He wondered what price point he could charge for the service? What
benefit would most motivate customers to purchase? He expected some potential customers to
ask for a “free” trial of the software. He worried that such an approach would “cheapen” the
value the software could bring to a potential consumer. For the fourth market, only part of Song
Maestro would be made available and that portion he was prepared to license. While he
announced a free of US $250,000 per year, he had no idea how to price this value-added feature
for record stores.
Download