2. Year One of Shazam – Startups in Action: The Critical Year One Choices That Built Etsy, HotelTonight, Fiverr, and More

© JP Silva 2020
J. SilvaStartups in Actionhttps://doi.org/10.1007/978-1-4842-5787-6_2

2. Year One of Shazam

JP Silva1 
(1)
Toronto, ON, Canada
 
Shazam (www.shazam.com) is a music recognition service founded by Chris Barton, Dhiraj Mukherjee, Philip Inghelbrecht, and Avery Wang (Figure 2-1). In September 2018, Apple Inc. completed the acquisition of Shazam. The value was not disclosed, but TechCrunch reported it to be around $400 million.
Figure 2-1

Screenshot of the landing page of shazam.com as of January 2020

The first year account that follows is based on a phone call and several e-mails exchanged between January and November 2019 with two of Shazam’s founders, Chris Barton and Dhiraj Mukherjee.

Month 0: September 1999

Barton, Inghelbrecht, and Mukherjee had been talking about starting a company for several months. Although the conversations between Barton and Inghelbrecht had been separate from those between Barton and Mukherjee, the three of them knew that they were eventually going to work together. In fact, they had verbally agreed halfway through the summer of 1999 that they were going to start a business together, once they came across an idea that interested them.

Barton and Inghelbrecht met in August 1998 at the start of their MBA at Berkeley, while Barton and Mukherjee had met in the early 1990s, when they were both living in San Francisco. Mukherjee meanwhile had moved to London to set up the UK operations of a US technology company, and Barton was also in London for the summer in between his first and second year of business school. Barton was doing an internship facilitated by Barton’s MBA program. As part of the program, students were encouraged to spend the summer break with companies interested in having them as interns, and Barton decided to spend some time at Microsoft in their London office.

During the summer of 1999, Barton and Mukherjee spent many hours in London cafes and pubs talking about their interest in technology, startups, and constantly brainstorming ideas for businesses that they could build as well as discussing their feasibility. Barton and Inghelbrecht continued to have the conversations that they had at Berkeley, although with Inghelbrecht having stayed back in California, they were less frequent, but still not less intense and lively. Moreover with the summer of 1999 somewhat being the height of the dot-com euphoria, Barton, Inghelbrecht, and Mukherjee were especially motivated. They felt that there was so much opportunity for startups to be established and to thrive.

The decision to go ahead with developing what would become Shazam came toward the end of September 1999. Barton, Inghelbrecht, and Mukherjee had discussed tens of ideas over the previous months, but they were not sure which one to pursue. As both Barton and Inghelbrecht wanted to use their second year at Berkeley for startup development, they all decided that they had to make a decision before the end of the month. As they couldn’t pursue several distinct ideas at the same time, they eventually agreed to pick one and focus. At this point, the idea that gathered most consensus was one that Barton had been brainstorming for some time on his own: a music recognition service.

For the longest time, Barton had been fascinated with sound and music recognition and the possibility of identifying songs when they were played on the radio. In his first conception of a service, a piece of software would be built, which would be provided to all radio stations to help them track their programming in real time, so that they could improve internal record keeping and royalty payment purposes. Barton was excited by the fact that this real-time data could then be used to provide a music recognition service to anyone with a mobile phone who wanted to know what was playing on a radio station. One evening in September 1999, Barton was thinking about the “defensibility” of this novel approach and tried to think of ways a competing service could leapfrog access to the real-time playlists of radio stations. He then suddenly realized that someone could simply identify the music via a mobile phone using the ambient sound in the air. He realized that this would leapfrog the original idea of software inside radio stations. He also realized that it would enable mobile phone users to identify music in places where the source was not radio such as TV, movie theaters, clubs, bars, stores, and others. He became obsessed with this as the breakthrough idea that he wanted to tackle.

Barton enthusiastically convinced Inghelbrecht and Mukherjee that they should develop a technology that could enable ambient music recognition via a mobile phone. Barton’s vision was greater than the feature of music recognition. Once a song was recognized, people could eventually have access to a whole host of related features such as buying songs, watching music videos, reading lyrics, creating playlists, and so on. Moreover, it was a mobile-first idea, and the entire team believed that mobile phones would soon become the default access device to the Internet. It was an idea for the future.

Months 1 to 3: October to December 1999

Barton, Inghelbrecht, and Mukherjee realized that they first had to understand the domain that was relevant to music recognition. To this end, they spent several weeks looking into academic research related to the analysis of sound and music. They eventually realized that the most relevant domain was called audio signal processing, which was a subcategory of digital signal processing. The next step was to find an expert that was not only able to develop a robust audio signal processing algorithm but also keen to become their technical co-founder.

Products and services:   None

End-users:   None

Staff:   None

Founders:   Barton, Inghelbrecht, and Mukherjee

Month 4: January 2000

Upon completing the exchange program, Barton went back to the United States to join Inghelbrecht in their final semester at Berkeley, and Mukherjee stayed in London to continue working for his employer. While Mukherjee typically could only afford to work evenings and weekends, Barton and Inghelbrecht used the idea in as much MBA coursework as they could, which allowed them to make significant progress researching the problem, understanding what the addressable market could be, who were the potential competitors and substitutes, and so on.

Mukherjee worked on a potential business plan in his spare time, while Barton and Inghelbrecht plowed ahead on the search for a technical expert and eventual co-founder. They were able to reach out to more experts who were all PhDs in electrical engineering from institutions such as MIT and Stanford that had focused on audio signal processing. However, no one seemed to have any idea of how to tackle the problem that involved inventing music pattern recognition technology that would work while facing the challenges of both noise and scale. The most encouraging conversations were typically that it was a very interesting and intriguing idea, but also extremely challenging.

Products and services:   None

End-users:   None

Staff:   None

Founders:   Barton, Inghelbrecht, and Mukherjee

Month 5: February 2000

Invigorated by the commitment of his co-founders, Mukherjee decided to quit his job to work on the project full time. After several months of working together, their relationship had become very close and solid. Mukherjee also had a good financial safety net, as he had saved a fair bit of money over the years and he could afford to spend a year or more without earning an income. If eventually he had to look for a job again, his MBA from Stanford, a consulting background of several years, and his involvement on several leading edge projects and initiatives would most likely make the search brief.

At this point, Barton and Inghelbrecht had already approached most people in their list of audio signal processing researchers in the United States who had specialized in music. One of the people that they had approached, but had not responded, was a professor at Stanford’s Center for Computer Research in Music and Acoustics (CCRMA) . After asking a Berkeley professor for an introduction, they finally managed to get their meeting with Julius Smith. The meeting with Smith was very positive, as he was immediately excited with the idea. Even though he didn’t know of any technology that could be used for music recognition in a noisy environment with sound recording over a mobile phone microphone, he nonetheless felt that it could be invented.

During a subsequent meeting at the CCRMA, Barton and Inghelbrecht were able to get Smith to review their list of thirty or so most relevant digital signal processing researchers in the United States and rank the top 5 most talented candidates. In this, the intention was not just to assess the list based on acoustics signal processing expertise but also on theoretical, strong mathematical and statistical background, as well as software development skills. The person that ranked number one was a previous student of Smith, Avery Wang. Smith connected Barton and Inghelbrecht with Wang, and the three met. Upon hearing about the problem and the possible solution, Wang suggested that a technology could eventually be invented, but the chances of coming up with something effective in a reasonable amount of time were actually pretty dismal.

With Wang possibly being the only chance to make the music recognition service a reality, even if still very improbable, Barton, Inghelbrecht, and Mukherjee decided to invite him to join the founding team. The amount of due diligence research that was presented to Wang, not to mention the detailed analysis of the competitive landscape, business modeling, and operational planning, strongly impressed him. He was an entrepreneur at heart, someone that wanted to build a company from the ground up and who happened to be struggling with his own audio hardware startup. More importantly he was utterly curious to see if the problem was solvable. Wang accepted and jumped on board as the fourth co-founder.

Products and services:   None

End-users:   None

Staff:   None

Founders:   Barton, Inghelbrecht, Mukherjee, and Wang

Months 6 to 8: March to May 2000

Wang started working on a solution on his own while brainstorming with Smith as frequently as possible. Progress toward developing an algorithm was almost nonexistent however. The challenge was immense. The solution that had to be produced had to be able to quickly search a database of songs that would eventually be in the millions, have a high recognition rate, have a low false positive rate, be robust despite noise and distortion, be very tolerant to room reverberation, and be able to work with the limited audio data caused by voice compression.

Mukherjee, on the other hand, had made quite a bit of progress on the operational front, as he managed to eventually incorporate the company in the United Kingdom, open a bank account, find an accountant to do the bookkeeping, investigate interactive voice response (IVR) systems , and start looking for a small office space.

Products and services:   None

End-users:   None

Staff:   None

Founders:   Barton, Inghelbrecht, Mukherjee, and Wang

Month 9: June 2000

As soon as Barton and Inghelbrecht completed their MBA in early June, Barton moved to the United Kingdom and joined Mukherjee in London while Inghelbrecht remained in Berkeley not too far from Wang’s home in Palo Alto. The team had decided that the company was going to be based in London, as the mobile communications market and industry was far more developed in Europe than in the United States at the time. If there was any chance to convince mobile network operators to deploy a mobile music recognition service, it would be those in the United Kingdom—the most innovative in Europe. Unlike US mobile operators, European mobile operators were already offering premium SMS, which allowed third parties, such as ringtone companies, to use it as a billing mechanism. Also, the music market was very strong in the United Kingdom, with greater music purchases per consumer than anywhere in the world. Finally, investors in Europe were attuned to business models targeting the consumer mobile market, whereas the US investors were not yet there.

With Mukherjee, Barton, Inghelbrecht, and Wang all working full time on Shazam, the company needed to quickly fundraise in order to keep going and limit personal burn rates. In preparation, Mukherjee worked on writing the business plan, Inghelbrecht on the financial model, and Barton on the slides and the pitch. In parallel, some meetings with potential partners were organized, but they were unwilling to make any commitments as Shazam did not have a proven technology, a commercial product, or even significant funding to build these.

In mid-June, Wang was staring at some graphs when he realized that there might be a very elegant and simple solution ahead. He recognized that there was a certain trace and a scatterplot of matching audio fingerprints. He was also able to realize that his finding could be a very strong statistical indicator toward recognizing a song. He quickly ran his hypothesis through several songs and found that the algorithm was actually able to properly recognize them, even if in some cases there was a lot of noise in the background. Although it could be premature, Wang became confident that the new solution could scale to possibly millions of songs and thousands of recognitions per second, without the need for massive amounts of computing power.

Once Barton, Inghelbrecht, and Mukherjee had a chance to celebrate, they immediately started requesting meetings with mobile network operators and investors. They knew that even if there could be eventual scalability issues in the future, the demo could be sufficient to convince the mobile network operators. With one or several operators on board and signed letters of intent from them, the chances of securing funding from venture capital investors had to improve dramatically.

Products and services:   None

End-users:   None

Staff:   None

Founders:   Barton, Inghelbrecht, Mukherjee, and Wang

Months 10 and 11: July and August 2000

With Wang confident about scalability, everyone agreed that he should immediately fly to London to quickly build a prototype. Meanwhile, several mobile network operators and investors had replied positively to meeting requests as well as asked for demos.

Barton, Inghelbrecht, Mukherjee, and Wang eventually secured several letters of intent from UK mobile network operators. In their perspective, the technology was very promising and there was very little risk for them, so they enthusiastically agreed to sign them. Investors also reacted very positively, and soon several business angels had agreed to invest about $1 million. The demo was of course central to it all. Going through the demo and witnessing the technology recognize music in a real environment recorded over a mobile phone microphone and search against a database of songs in just a few seconds was nothing short of wizardry. Everyone was astounded, as they had never experienced anything similar, and they had never thought it would be possible.

Barton wanted to raise the initial seed funding from high-profile angel investors, people that were known in the mobile and music industries, as they could help validate the business in the eyes of the venture capitalists who would be the next round of funding. In fact, when the angel round was finally closed, the investors that committed, all had relevant professional backgrounds, as they had been executives at the major record labels such as EMI and BMG or at technology companies such as Amazon Europe and British Telecom.

Products and services:   None

End-users:   None

Staff:   None

Founders:   Barton, Inghelbrecht, Mukherjee, and Wang

Month 12: September 2000

September ended up being consumed with paperwork, talking to lawyers, discussing with investors, checking and signing up documents, and getting everything lined up so that the money could be transferred into Shazam’s empty bank account. Still, there was no time to lose, as Barton, Inghelbrecht, Mukherjee, and Wang had to start discussing how they could take the technology from the prototype to the commercial release. They had to figure out how to build a database of millions of songs, not to mention how to make the service available across all the mobile operators in the United Kingdom.

Products and services:   None

End-users:   None

Staff:   None

Founders:   Barton, Inghelbrecht, Mukherjee, and Wang

Beyond the First Year

In November 2000, Shazam raised $350,000 in a second angel round.

In February 2001, it raised $1 million in a third angel round.

In July 2001, it raised $7.5 million in a series A round from three venture capital investors: IDG Ventures Europe, Lynx Capital Partners, and FLV Fund.

In August 2002, it launched commercially across all four major UK mobile network operators, thus available to be used by virtually every person with a mobile phone in the United Kingdom on the day of launch. See Figure 2-2.
Figure 2-2

Landing page of shazam.com as of August 20021

In October 2002, it raised $2 million in series B round from several undisclosed investors.

In August 2003, it raised $4 million in series C round from several undisclosed investors.

In August 2009, it raised $5 million in a series D round from several undisclosed investors.

In April 2011, it raised $32 million in a second series D round co-led by Kleiner Perkins and IVP.

In April 2013, it raised $40 million in a third series D round led by Carlos Slim.

In December 2014, it raised $50 million in a series E round from several undisclosed investors.

In December 2017, it accepted an acquisition from Apple for a reported $400 million.

Table 2-1 shows the evolution in the number of offerings, number of monthly active users, and staff up to Shazam’s sale to Apple in 2017. Unfortunately, the year-on-year evolution cannot be effectively gauged, as additional data was not disclosed in press releases or conversations. Still, it’s possible to note that when the company was acquired, it had already achieved significant milestones, that is, grown its number of monthly active users to over 300 million and its staff to over 250.
Table 2-1

Evolution of key indicators until acquisition

Today

Today, Shazam is one of the most recognized services for mobile music recognition. There are a number of competing services, but Shazam is still the most popular and recognized option, to the point of being one of the most downloaded apps of the decade in the music category. In fact, it has become so ingrained in popular culture that new startups often refer to themselves as the “Shazam for clothes” or the “Shazam for plants and animals.”

Reflections

In addition to exposing the in the making of Shazam during its first year, the month by month account also revealed that:
  • Successful technology companies can be started as empathy in action: Shazam started off from a strong desire by one of the co-founders to develop a sound and music recognition service and allowed anyone to identify songs when they were played on the radio. In this, it was not just the yearning for a service that flawlessly worked, but more importantly one that was seamless and easy to use (what if someone could simply identify the music via a mobile phone using the ambient sound in the air?).

  • Successful technology companies can be started as side projects: The founders of Shazam spent a large part of their first year only working evenings and weekends in their new company. They eventually concluded their graduate degrees (Barton and Inghelbrecht) and left their full-time jobs (Mukherjee), but not before they saw momentum being gathered.

  • Successful technology companies can be started with many co-founders: One of the initial strengths of Shazam was its close-knit group of co-founders, particularly when they were simultaneously looking for a technical co-founder and talking to potential partners and investors in the United Kingdom and the United States. If it wasn’t for the good relationship as well as the very diverse set of skills, the initial momentum might not have taken place, which most likely would have discouraged Mukherjee to quit his job or Wang to join later, once he realized the complexity of the algorithm that had to be developed.

  • Successful technology companies can be started as slowness in action: Shazam took several years to be commercially unveiled. Although there was an expectation that the technology was very complex and was going to take time to develop, the founders never expected that it would take almost three years from start to commercial launch, from October 1999 until August 2002.