March 7, 2005
Come together | The developers of Goombah want to tap into the growing digital music market, one person at a time
By Taylor Smith
During the mid-1980s, Gary Robinson toted an acoustic guitar around New York's Greenwich Village, going to songwriting workshops and playing gigs at coffeehouses. He saw loads of talent in his fellow songwriters, but he also saw how many of them struggled to eke out a living in the highly competitive music industry. "A lot of them disappeared into a black hole because they weren't considered to be marketable," he says. "You just don't have that kind of reach unless you're picked up by the mass-market machine."
Twenty years later, Robinson finally feels like he's helping alleviate the plight of under-appreciated musicians, albeit with a laptop rather than a guitar. A mathematician and computer programmer by trade, Robinson has refined a program he's spent years developing with Bob Swerdlow, a programmer and veteran of the Boston technology industry. The program, called Goombah, was designed as a music recommendation application that matches music lovers based on each user's personal tastes. The point of the application is simple enough: Create groups of like-minded music lovers who can pick through each others' online music libraries, learn about new bands or artists and, ultimately, buy a song or album.
Robinson says the complex technology behind Goombah is unique. The program uses a rigorous mathematical analysis of each user's musical library to generate a list of other Goombah users with similar musical tastes. And, the Goombah theory goes, one user is likely to have songs in his or her own library that another user doesn't have, but probably would like. With the increasing fragmentation of the music industry, Robinson and Swerdlow say the market is ready for a product like Goombah. "It's a very interesting time," says Swerdlow. "There's more music and more outlets to find that music, but we believe there are many people who still can't find the music they're into."
After years of quietly developing Goombah, Robinson and Swerdlow plan to launch the program publicly over the next few months. Although they have high hopes, the launch will present their company, Brunswick-based Emergent Music LLC, with a range of challenges. For starters, Goombah isn't the only music recommendation application available: Companies such as Savage Beast Technologies, Soundflavor and Audioscrobbler all lay claim to having the best method for recommending new tunes to music lovers. ("If you go to their services and try them, the music it thinks is similar ˆ some people would think they're nuts," says Robinson.) Meanwhile, Emergent Music CEO Dianne Sammer and the rest of the management team need to figure out the crucial question of how to generate a steady revenue stream from the program.
Phil Leigh, a stock analyst and founder of Inside Digital Media, a research firm in Tampa, Fla., says Goombah's premise is interesting because it gives people with similar musical tastes a venue to discuss new bands or different artists. Those affinity groups, says Leigh, are increasingly important to the music industry because they inevitably result in record sales. The traditional music affinity groups ˆ listeners of a particular radio station, for example, ˆ in recent years have taken a hit, Leigh says, from the spreading influence of digital music. "If people have an iPod, they're not getting exposed to new music," he says. "One of the big problems the record labels have is how to stimulate the demand for new music if people aren't listening to the radio."
And these days, there's no end to the amount of new music flooding the market. Chris Brown, vice president of operations and marketing at Portland-based Bull Moose Music, says each of the company's 10 stores in Maine and New Hampshire carries 40,000-60,000 music titles and that every week well over 1,000 more titles are dumped into the marketplace. Only a handful of those titles will ever make a dent in the weekly SoundScan ratings, while the others are left to languish in record bins and warehouses.
But Robinson says part of a hit record's success is the network of fans who build buzz for an artist. "Having a community is a very important part of a phenomenon like Britney Spears," he says. "We want to have a community for people whose tastes are very unusual."
Making a match
Robinson began developing the seeds of Goombah during the 1980s, when he worked as a consultant for New York Telephone, building complex databases with data culled from millions of customer records. His interest in mathematics led Robinson to found Microvox Systems, a company that featured just one product, a voicemail-based dating service called 212-ROMANCE. Unlike many telephone dating services, 212-ROMANCE employed a math-based formula called collaborative filtering to play matchmaker (which Robinson says was the first commercial application for such filtering technology). Robinson figured that users would find more success if they didn't have to slog through every voicemail in the system in search of the perfect date. Instead, Robinson set the system up to do much of the vetting work by trying to match people with similar interests.
Fast forward 20 years, and Robinson's grasp on collaborative filtering has progressed significantly. (In fact, Robinson in 1999 received a patent for a mathematical basis of collaborative filtering. Goombah is the latest evolution in a string of applications Robinson and Swerdlow have developed using collaborative filtering technology.
Most recently, Robinson and Swerdlow were trying to market a predecessor of Goombah called Trumpet to corporate clients as a tool for knowledge management (corporate jargon for the ability to access and manage mounds of company data, from corporate policies to industry news). Trumpet also included a way for some companies to predict market cycles or online activity. Pharmaceutical giant Pfizer, for example, was interested in Trumpet after Robinson and Swerdlow explained that it offered the company a way to evaluate how drug sales in certain markets could be used as a predictor for other markets. Their company, Transpose LLC, received a $100,000 grant from the National Science Foundation in late 2002 to continue developing the Trumpet technology, but Robinson says Transpose had difficulty convincing companies to pay for an expensive knowledge management tool.
("Database heroes," Nov. 25, 2002.) A deal with online auction site eBay fizzled when neither side would part with proprietary information, and Pfizer balked when Transpose couldn't offer around-the-clock product support. "It was the tiger-by-the-tail problem," says Swerdlow.
But while Trumpet was trying to gain respectability in the corporate world, Robinson, Swerdlow and Matt Goyer, a Canadian programmer and entrepreneur who helped build Trumpet, were developing an early version of Goombah called Emergent Music that aimed to give non-mass-market artists a fair shake in the major-label dominated music industry. The scope of Emergent Music eventually widened to include music recommendation, and, in late 2003, Robinson and Swerdlow released an early beta version of Goombah. (It was replaced by a new beta version last October.)
Meanwhile, the pair began recruiting a management team to market Goombah. Diane Sammer, a former telecom executive and founder of Maine Angels, a venture capital group, joined the company in late 2004 after running into Swerdlow at a handful of entrepreneurial events. Sammer brought in Randy Labbe, a record producer and founder of the Accord Music Group, a Waterville-based record label that Sammer invested in, for his expertise in the music industry. And to round out the group, Madeline Mooney, a marketing veteran, and Alec Wysoker, an engineer from Cambridge, Mass., were brought into the fold early this year.
With Goombah, the premise works like this: Instead of classifying a user as a thrash-metal lover or bluegrass junkie, the software filters through her online music library, cataloging each artist, song and album. It then compares that information to the rest of the online Goombah community, generating a list of that user's Goombah Neighbors ˆ people on the Goombah network who the software determines to have similar musical tastes.
Once a list of neighbors is generated, users can then scroll through their libraries. Each song title, album and artist contains links to retail websites like Amazon.com and Apple's iTunes music store, as well as reference sites like AllMusic, which contains detailed information on more than 778,000 albums.
The revenue question
CEO Sammer says that Goombah benefits music lovers by offering recommendations for new music; it also benefits online retailers like Amazon and iTunes, she says, by driving more business to their sites. "There's value for these sites, and they're willing to pay us for that," she says.
Goombah has struck affiliate deals with both Amazon and iTunes, where any purchases made through the result of the Goombah matchmaking process will net the company a cut of roughly five percent. Those affiliate programs, however, are the same ones used by bloggers and individual website owners; the relationships don't give any preferential treatment to Goombah.
Analyst Phil Leigh says that while the Goombah concept is interesting, the company could face heavy challenges on the business side. The problem, he says, is that many users who purchase music through sites like Amazon, which has its own recommendation capability, and iTunes eventually will develop direct relationships with those retailers, bypassing Goombah entirely. "The idea is good," says Leigh. "But it's going to be difficult to make a lot of money the way they've got it set up."
Mike Goodman, an analyst with the Yankee Group, a Boston-based research firm, agrees that Goombah faces challenges on the road to profitability. "They can build a great recommendation model and get bought, or license it out to music services like Napster," he says. "Those are the two business models that have a chance of success. I don't think there's much opportunity for them, period." ("I guess we'll see, won't we?" responds Sammer. "If we have zillions of users, I can assure you that we're going to make some money.")
But with five experienced executives ˆ many of whom have had the luxury of fat paychecks in recent years ˆ running Goombah, it's not a stretch to believe that they're expecting a big payoff down the road. (In the near term, says Sammer, all five have agreed to forego any compensation.) Sammer says the company is developing other relationships that she says could create "more lucrative situations." She declines to discuss those relationships in detail, but says that if they demonstrate to the industry that Goombah can play matchmaker with a great deal of precision, affiliate programs won't be the only source of revenues. "We see a big opportunity here," says Sammer. "Our mission is to attract as many of those music lovers as we can and build our community. If we do it well, there will be lots of interesting revenue opportunities."
But before the company can begin exploring new revenue models ˆ or demonstrate its proficiency as a matchmaker ˆ Goombah has to build a community. So far, efforts at publicizing Goombah have been relatively muted. In late 2003, Robinson and Swerdlow posted the early beta version of Goombah to VersionTracker, a software website frequented largely by Macintosh aficionados. They hoped to entice about 500 people to test the free software; in just two days, more than 300 people downloaded it. A new beta version of Goombah was released last October for the Macintosh, and this month the company was scheduled to release a beta version of the software for PC users.
Roughly 1,200 registered users have signed onto the Goombah network, which still is a very small sample to work with, says Robinson. He'd like to see millions of people registered; such a sizeable network, he says, would allow for much greater accuracy when matching Goombah neighbors because of the vast amounts of data the software is able to draw from.
Madeline Mooney, Goombah's vice president of marketing, plans to increase the software's visibility during the next few months and admits that the company isn't likely to go through the traditional marketing channels like advertising or PR campaigns. Instead, Mooney says Goombah will be launched through a low-key, grassroots campaign that will be dependent on good word of mouth from users.
Mooney, who worked as the vice president of marketing for Internet search firm Lycos Inc. until moving to Ogunquit early last year, says many promotional avenues will cost little or nothing. Best of all, she says, the product features cool technology and is so user friendly that it won't alienate non-techies. "I love a challenge, but I wouldn't take it on if I didn't think the company had the right idea at the right time," says Mooney.
From a programmer's perspective, Robinson agrees that Goombah has the potential to be a blockbuster application. But he also recognizes that even if it finds success with users, further development of the company will depend on strong financial performance. "I never would have pursued it if we didn't think it would be a huge success," he says. "The reason I have this passion about it is because it's solving a problem. But in the real world, you also have to be able to make money."
Twenty years later, Robinson finally feels like he's helping alleviate the plight of under-appreciated musicians, albeit with a laptop rather than a guitar. A mathematician and computer programmer by trade, Robinson has refined a program he's spent years developing with Bob Swerdlow, a programmer and veteran of the Boston technology industry. The program, called Goombah, was designed as a music recommendation application that matches music lovers based on each user's personal tastes. The point of the application is simple enough: Create groups of like-minded music lovers who can pick through each others' online music libraries, learn about new bands or artists and, ultimately, buy a song or album.
Robinson says the complex technology behind Goombah is unique. The program uses a rigorous mathematical analysis of each user's musical library to generate a list of other Goombah users with similar musical tastes. And, the Goombah theory goes, one user is likely to have songs in his or her own library that another user doesn't have, but probably would like. With the increasing fragmentation of the music industry, Robinson and Swerdlow say the market is ready for a product like Goombah. "It's a very interesting time," says Swerdlow. "There's more music and more outlets to find that music, but we believe there are many people who still can't find the music they're into."
After years of quietly developing Goombah, Robinson and Swerdlow plan to launch the program publicly over the next few months. Although they have high hopes, the launch will present their company, Brunswick-based Emergent Music LLC, with a range of challenges. For starters, Goombah isn't the only music recommendation application available: Companies such as Savage Beast Technologies, Soundflavor and Audioscrobbler all lay claim to having the best method for recommending new tunes to music lovers. ("If you go to their services and try them, the music it thinks is similar ˆ some people would think they're nuts," says Robinson.) Meanwhile, Emergent Music CEO Dianne Sammer and the rest of the management team need to figure out the crucial question of how to generate a steady revenue stream from the program.
Phil Leigh, a stock analyst and founder of Inside Digital Media, a research firm in Tampa, Fla., says Goombah's premise is interesting because it gives people with similar musical tastes a venue to discuss new bands or different artists. Those affinity groups, says Leigh, are increasingly important to the music industry because they inevitably result in record sales. The traditional music affinity groups ˆ listeners of a particular radio station, for example, ˆ in recent years have taken a hit, Leigh says, from the spreading influence of digital music. "If people have an iPod, they're not getting exposed to new music," he says. "One of the big problems the record labels have is how to stimulate the demand for new music if people aren't listening to the radio."
And these days, there's no end to the amount of new music flooding the market. Chris Brown, vice president of operations and marketing at Portland-based Bull Moose Music, says each of the company's 10 stores in Maine and New Hampshire carries 40,000-60,000 music titles and that every week well over 1,000 more titles are dumped into the marketplace. Only a handful of those titles will ever make a dent in the weekly SoundScan ratings, while the others are left to languish in record bins and warehouses.
But Robinson says part of a hit record's success is the network of fans who build buzz for an artist. "Having a community is a very important part of a phenomenon like Britney Spears," he says. "We want to have a community for people whose tastes are very unusual."
Making a match
Robinson began developing the seeds of Goombah during the 1980s, when he worked as a consultant for New York Telephone, building complex databases with data culled from millions of customer records. His interest in mathematics led Robinson to found Microvox Systems, a company that featured just one product, a voicemail-based dating service called 212-ROMANCE. Unlike many telephone dating services, 212-ROMANCE employed a math-based formula called collaborative filtering to play matchmaker (which Robinson says was the first commercial application for such filtering technology). Robinson figured that users would find more success if they didn't have to slog through every voicemail in the system in search of the perfect date. Instead, Robinson set the system up to do much of the vetting work by trying to match people with similar interests.
Fast forward 20 years, and Robinson's grasp on collaborative filtering has progressed significantly. (In fact, Robinson in 1999 received a patent for a mathematical basis of collaborative filtering. Goombah is the latest evolution in a string of applications Robinson and Swerdlow have developed using collaborative filtering technology.
Most recently, Robinson and Swerdlow were trying to market a predecessor of Goombah called Trumpet to corporate clients as a tool for knowledge management (corporate jargon for the ability to access and manage mounds of company data, from corporate policies to industry news). Trumpet also included a way for some companies to predict market cycles or online activity. Pharmaceutical giant Pfizer, for example, was interested in Trumpet after Robinson and Swerdlow explained that it offered the company a way to evaluate how drug sales in certain markets could be used as a predictor for other markets. Their company, Transpose LLC, received a $100,000 grant from the National Science Foundation in late 2002 to continue developing the Trumpet technology, but Robinson says Transpose had difficulty convincing companies to pay for an expensive knowledge management tool.
("Database heroes," Nov. 25, 2002.) A deal with online auction site eBay fizzled when neither side would part with proprietary information, and Pfizer balked when Transpose couldn't offer around-the-clock product support. "It was the tiger-by-the-tail problem," says Swerdlow.
But while Trumpet was trying to gain respectability in the corporate world, Robinson, Swerdlow and Matt Goyer, a Canadian programmer and entrepreneur who helped build Trumpet, were developing an early version of Goombah called Emergent Music that aimed to give non-mass-market artists a fair shake in the major-label dominated music industry. The scope of Emergent Music eventually widened to include music recommendation, and, in late 2003, Robinson and Swerdlow released an early beta version of Goombah. (It was replaced by a new beta version last October.)
Meanwhile, the pair began recruiting a management team to market Goombah. Diane Sammer, a former telecom executive and founder of Maine Angels, a venture capital group, joined the company in late 2004 after running into Swerdlow at a handful of entrepreneurial events. Sammer brought in Randy Labbe, a record producer and founder of the Accord Music Group, a Waterville-based record label that Sammer invested in, for his expertise in the music industry. And to round out the group, Madeline Mooney, a marketing veteran, and Alec Wysoker, an engineer from Cambridge, Mass., were brought into the fold early this year.
With Goombah, the premise works like this: Instead of classifying a user as a thrash-metal lover or bluegrass junkie, the software filters through her online music library, cataloging each artist, song and album. It then compares that information to the rest of the online Goombah community, generating a list of that user's Goombah Neighbors ˆ people on the Goombah network who the software determines to have similar musical tastes.
Once a list of neighbors is generated, users can then scroll through their libraries. Each song title, album and artist contains links to retail websites like Amazon.com and Apple's iTunes music store, as well as reference sites like AllMusic, which contains detailed information on more than 778,000 albums.
The revenue question
CEO Sammer says that Goombah benefits music lovers by offering recommendations for new music; it also benefits online retailers like Amazon and iTunes, she says, by driving more business to their sites. "There's value for these sites, and they're willing to pay us for that," she says.
Goombah has struck affiliate deals with both Amazon and iTunes, where any purchases made through the result of the Goombah matchmaking process will net the company a cut of roughly five percent. Those affiliate programs, however, are the same ones used by bloggers and individual website owners; the relationships don't give any preferential treatment to Goombah.
Analyst Phil Leigh says that while the Goombah concept is interesting, the company could face heavy challenges on the business side. The problem, he says, is that many users who purchase music through sites like Amazon, which has its own recommendation capability, and iTunes eventually will develop direct relationships with those retailers, bypassing Goombah entirely. "The idea is good," says Leigh. "But it's going to be difficult to make a lot of money the way they've got it set up."
Mike Goodman, an analyst with the Yankee Group, a Boston-based research firm, agrees that Goombah faces challenges on the road to profitability. "They can build a great recommendation model and get bought, or license it out to music services like Napster," he says. "Those are the two business models that have a chance of success. I don't think there's much opportunity for them, period." ("I guess we'll see, won't we?" responds Sammer. "If we have zillions of users, I can assure you that we're going to make some money.")
But with five experienced executives ˆ many of whom have had the luxury of fat paychecks in recent years ˆ running Goombah, it's not a stretch to believe that they're expecting a big payoff down the road. (In the near term, says Sammer, all five have agreed to forego any compensation.) Sammer says the company is developing other relationships that she says could create "more lucrative situations." She declines to discuss those relationships in detail, but says that if they demonstrate to the industry that Goombah can play matchmaker with a great deal of precision, affiliate programs won't be the only source of revenues. "We see a big opportunity here," says Sammer. "Our mission is to attract as many of those music lovers as we can and build our community. If we do it well, there will be lots of interesting revenue opportunities."
But before the company can begin exploring new revenue models ˆ or demonstrate its proficiency as a matchmaker ˆ Goombah has to build a community. So far, efforts at publicizing Goombah have been relatively muted. In late 2003, Robinson and Swerdlow posted the early beta version of Goombah to VersionTracker, a software website frequented largely by Macintosh aficionados. They hoped to entice about 500 people to test the free software; in just two days, more than 300 people downloaded it. A new beta version of Goombah was released last October for the Macintosh, and this month the company was scheduled to release a beta version of the software for PC users.
Roughly 1,200 registered users have signed onto the Goombah network, which still is a very small sample to work with, says Robinson. He'd like to see millions of people registered; such a sizeable network, he says, would allow for much greater accuracy when matching Goombah neighbors because of the vast amounts of data the software is able to draw from.
Madeline Mooney, Goombah's vice president of marketing, plans to increase the software's visibility during the next few months and admits that the company isn't likely to go through the traditional marketing channels like advertising or PR campaigns. Instead, Mooney says Goombah will be launched through a low-key, grassroots campaign that will be dependent on good word of mouth from users.
Mooney, who worked as the vice president of marketing for Internet search firm Lycos Inc. until moving to Ogunquit early last year, says many promotional avenues will cost little or nothing. Best of all, she says, the product features cool technology and is so user friendly that it won't alienate non-techies. "I love a challenge, but I wouldn't take it on if I didn't think the company had the right idea at the right time," says Mooney.
From a programmer's perspective, Robinson agrees that Goombah has the potential to be a blockbuster application. But he also recognizes that even if it finds success with users, further development of the company will depend on strong financial performance. "I never would have pursued it if we didn't think it would be a huge success," he says. "The reason I have this passion about it is because it's solving a problem. But in the real world, you also have to be able to make money."
Most Recent
Most Recent
Comments