I can say fairly confidently that I use YouTube more than any other app. It all started in college when I googled a music video I hadn’t seen since it was on MTV’s TRL. YouTube’s music video catalog was still pretty thin back then, but the idea I could watch any music video anytime I wanted blew my mind. Gradually, YouTube’s algorithm realized I had other interests and the videos it showed me were better than anything I saw on TV. One minute I was watching engineers explain what happened when monsoons crashed into bridges; the next, Jeff Bezos breaking down Amazon’s leadership principles to a class of students thousands of miles away. Anytime I’m confused about a new concept, I search YouTube to find someone to explain it to me like I’m 5. When I want some background noise, I turn on the YouTube TV app. YouTube was — and remains — my best outlet for procrastination. Its endless library of videos entertain me, distract me, and expose me to parts of the world I’d never see without it.

YouTube’s Origins

YouTube was founded in 2005 by former PayPal employees Chad Hurley, Steve Chen, and Jawed Karim. Their original plan was to create a new type of dating site featuring video profiles instead of static descriptions. They reasoned that video profiles would provide far more context than ordinary profiles and transform the way people found love. The plan fell apart, however, when they launched their private beta. They couldn’t find women willing to upload videos, so they resorted to taking out ads offering $20 to women who joined.

As bad as the young founders were at matchmaking, they were exceptional engineers. Nearly all the websites that let people share videos online in 2005 were either terrible or prohibitively expensive. By creating a simple interface for uploading videos and letting anyone use it for free, YouTube had unwittingly discovered a much better way to publish videos online. The first YouTube video on record — Jawed at the zoo — set the stage for creators of all kinds to use YouTube to share any video they wanted. The concept took off almost immediately and within 18 months they sold the site to Google for $1.65b.

Problem 1 — Putting videos online

At the time of YouTube’s founding, it was very difficult to build a website where people could publish and watch videos. Among other challenges, the team had to figure out:

1. Recording. In 2005 there weren’t many people who had videos to upload. Back then, mobile phones didn’t have cameras on them and most people didn’t own video cameras. YouTube’s base of creators was limited to people who had bought video cameras, uploaded their videos onto their computers, and wanted people on the internet to watch what they had recorded.

2. Hosting. Video files were large and cloud hosting platforms like AWS didn’t exist yet, so YouTube had to come up with a way to store videos that users uploaded. The team had no choice but to buy servers and personally maintain them — no small feat during phases of hypergrowth.

3. Playing. Internet connections were relatively slow and bandwidth constraints made it hard to play videos quickly through web browsers. Users essentially had to download videos they wanted to watch using their internet connection, which made for a disjointed viewing experience.

4. Sharing. YouTube could only grow quickly if creators and viewers had an easy way to share videos. That meant every video needed its own webpage and viewers had to be able to share links through channels where their contacts could easily access them.

5. Discovery. Viewers needed a way to find videos that creators uploaded. So on top of all the challenges of getting videos onto webpages, YouTube had to build its own search engine and teach creators how to get their videos featured in search results.

Problem 2 — Moderating video uploads

As soon as people realized they could upload any video and quickly build an audience, they started using YouTube to share videos that threatened its future. Unlicensed clips of film and television scenes invited lawsuits from major studios who were determined to protect their valuable DVD businesses. Graphic clips featuring nudity and violence would tarnish YouTube’s reputation as it courted mainstream viewers around the world. Without a legal department and moderation team to fend off these issues, YouTube would face the same fate as illegal file-sharing sites like Napster.

Problem 3 — Growing too quickly

Most startups don’t have the luxury of growing so quickly that they can’t support their users, but YouTube was one of the lucky few. Just six months after launching its public beta, YouTube had its first video with over a million views. Two months after reaching that milestone, YouTube was getting 8 million views per day. As their viewership scaled up, their costs scaled as well. It was very difficult to create a fast, stable viewing experience when so many people were trying to watch the same videos at the same time. They needed engineers, servers, attorneys, managers, and — above all else — money to pay for their rapidly growing website. While venture capitalists were happy to fund them, the founders could see the writing on the wall. They had stumbled onto what looked like a generational internet business, but growing pains meant that success was far from guaranteed.

Selling to Google

The founders decided they needed the resources of an experienced company to help them navigate YouTube’s growth phase. Instead of going it alone and trying to build out internal infrastructure, they started listening to acquisition offers from companies who had already dealt with the kinds of issues they were facing. In short order, Google and Yahoo emerged as the top bidders. Unlike legacy video companies, Google and Yahoo understood the dynamics of building an internet business. They also had engineers who could quickly get up to speed with YouTube’s technical needs. But Google won out. YouTube already relied heavily on search to help users discover its videos and Google was the world’s leading search engine. Furthermore Eric Schmidt, Google’s then-CEO, promised them autonomy, control, and hundreds of millions in personal wealth. They completed the entire deal in just one week with final closing at a Denny’s in Palo Alto.

Insight 1 — Selling out doesn’t mean giving up

The first insight from YouTube’s success is about the timing of its sale to Google. The founders chose to sell just as they entered the most intense phase of their growth. They had significant content moderation problems, unsettled negotiations with major studios, and a tiny ad business that hadn’t proven itself yet. While viewership was growing exponentially, it was clear the team needed experienced managers to help turn the attention they were generating into a business. The founders could’ve made a lot more money if they had stayed independent, but they also could’ve lost their business in the Great Recession that started just a year later. Selling when they did locked in their generational wealth and gave YouTube the resources it needed to grow into the the largest video-sharing site in the world as well as the second largest search engine.

Insight 2 — Auctions raise sale prices

The second insight from YouTube’s success is about how the founders conducted their sale. They made two brilliant moves that ensured they’d get the best outcome for the company. First they brought multiple bidders to the table. By having several companies submit offers, they enticed them to make big offers that they might not have gotten if they had only pursued a single bidder. Google wanted YouTube to expand its own ad business, but it also wanted to stop Yahoo from stealing it away. If Yahoo had won the bidding war, it could very well be Google’s largest competitor today and costed them many multiples on the $1.65b they ended up paying. The second move the founders made was to put time pressure on the deal. By running a fast process, they didn’t give Google a chance to lowball them and draw out the process.

Insight 3 — Solve hard technical problems

The third insight from YouTube’s success is about the founders’ commitment to solving difficult technical problems. Oftentimes founders pivot when they encounter hard technical problems that aren’t immediately solvable. YouTube’s founders weren’t video engineers, they had worked in payments at PayPal. But despite their inexperience, they committed to solving all the problems with publishing and watching videos online. By solving a technical problem that nearly every amateur video creator had at the time, YouTube created a giant business that changed the entire internet. Their willingness to step outside of their comfort zone gave them generational wealth and a business that remains the leader in its category to this day.

If you want to connect with me more, don’t hesitate to reach out on Twitter or LinkedIn.

As a teenager my music library was one of my most prized possessions. I spent countless hours finding songs online, downloading mp3s, looking up album art, making playlists, and debating who was hot with my friends. I even kept an external hard drive with all my songs just in case something happened to my computer. By the time Spotify came out, I was so locked into my iTunes library that I didn’t want to bother with it. But when my then-girlfriend started sending me links to her Spotify playlists, I had to give it a try. I slowly started copying my playlists into Spotify to share with her, and within a few months I was a regular.

Spotify’s Origins

Spotify was founded in 2006 by Swedish entrepreneurs Daniel Ek and Martin Lorentzon. They were each decorated founders in their own right; in fact Martin’s company TradeDoubler acquired Daniel’s company Advertigo just months before they started working on Spotify. Their plan with Spotify was to create a service that people would rather use than pirating music — no small task in the era of illegal file-sharing services like Napster, Kazaa, and Limewire. Why would people pay for albums when they could just download the songs they wanted for free? Daniel and Martin believed that the answer was convenience: if people could listen to every song they wanted without having to wait for downloads, they’d either listen to ads or pay a small subscription.

Problem 1 — Competing with piracy

It’s hard to understate the impact piracy had on the music industry. For decades, album sales were the best way for musicians and their representatives to make money. Fans had to buy albums to listen to the music they liked when they wanted, so artists who made good music could sell millions of copies relatively easily. Even if you only liked a couple of the songs on an album you had to buy the whole thing, which meant musicians could generate a meaningful income with a tiny catalog. But with peer-to-peer file-sharing services, fans started handpicking their favorite songs and downloading them without paying anything. As long as one fan bought an album and uploaded it to a file-sharing service, millions could listen to it for free anytime they wanted. Even though this was illegal in most countries under DRM laws, it was impossible to prevent fans from doing it. Every time the music industry nailed someone in court, more would spring up and take their place.

Against this backdrop, Daniel and Martin decided to create a listening experience so compelling that the average fan wouldn’t bother to steal music anymore. Their big vision was to eliminate the need to download music altogether. Spotify’s entire catalog would be available to anyone with an internet connection. As long as fans listened to ads or paid a subscription, they could listen to anything they wanted. Then Spotify would share its revenues with musicians and their representatives.

Problem 2 — Getting music rights

With their vision in place, Daniel and Martin started building. Unlike the free file-sharing services that preceded it, Spotify chose to sign licensing deals with major labels before they launched so they wouldn’t have to worry about getting sued. This proved to be their biggest hurdle in getting off the ground. After a series of meetings with music executives in the US, Daniel sensed that piracy had ravaged the industry enough that the labels would work with Spotify. He estimated a 6 month timeline to launch. But it would take nearly two and half years and some creative dealmaking to get the deals over the finish line.

Despite their dire conditions, the labels wanted major concessions to sign on. Among other things, Spotify had to:

1. Pay 1 year of licensing fees up front even if they went belly up

2. Pay 70% of total revenues in royalties to artists and labels

3. Prove the model in piracy-heavy EU markets before expanding elsewhere

4. Allow the labels to buy 20% of Spotify to expand out of the EU

Daniel and Martin decided to accept the costs and move forward. The music rights were too important to their model to quibble over fees and constraints.

Go-to-Market

Once it settled its deals, Spotify was ready to launch. Tactically, they needed to make money in every market to earn back the licensing fees they were paying, so they tried to get as many premium customers as they could as quickly as possible. Their playbook was to launch the premium version first, make the free version invite-only, and use influential tastemakers to drum up interest. Their invite program was especially effective: each free users got 5 invites and tastemakers got even more depending on how large their followings were.

Aside from the labels, Spotify also had to contend with the new king of music: Apple. Not only did Apple have a wildly successful music app and its own deals with the music industry, it also created devices that Spotify would have to use to reach American fans. Spotify’s weapon was its business model. Where Apple sold each song for 99¢ (later raised to $1.29), Spotify offered its entire catalog ad-free for a monthly subscription. Cost-conscious listeners who didn’t want to pay per song could just listen to ads using Spotify’s free offering. Those who didn’t want ads could pay $9.99/mo for unlimited access to Spotify Premium, which offered better value than Apple as long as they found 11 new songs per month.

Insight 1 — Don’t let big deals scare you

The first insight from Spotify’s success is about dealmaking. Startups are often scared to pursue deals with large companies because of the time it takes to negotiate them. Spotify went the other way and bet its entire existence on deals with major record labels. The deals took a year and half longer than expected and saddled them with major costs at the height of the Great Recession. Instead of abandoning their strategy and pivoting, they stayed at the negotiating table and trusted that the wait would be worth it. Their resilience in dealing with large companies earned them legal standing and validated their business all over the world.

Insight 2 — Solve hard technical problems

The second insight from Spotify is about technical innovation. Daniel and Martin set a goal that many thought to be technically impossible: instant access to millions of songs without download times. Even Apple didn’t bother to attempt what Spotify pulled off. They came up with an innovative method where they only downloaded the first fifteen seconds of a given song from the internet, then filled in the rest using peer-to-peer sharing with nearby Spotify users. By setting a high bar for their technology, Spotify differentiated itself from other music players of its time.

Insight 3 — Every competitor is beatable

The third insight from Spotify is about competition. Spotify launched in the US in 2011, a decade after Apple revolutionized music through the iPod. Few companies in the world have ever enjoyed the kind of brand equity in an industry that Apple had in music. Compared to Steve Jobs, Daniel and Martin were nobodies. Apple had devices, apps, an army of fawning press, and ownership of the App Store. Every iPhone had Apple’s native music player and music store pre-installed. But in the face of all of that, Spotify persisted and built the largest subscriber-base in the world. Startups often get discouraged when they see a larger competitor with a successful product, but Spotify showed that startups can compete even in the most unlikely circumstances. In the end, it was Apple that had to change its business model to accommodate the demand for Spotify — not the other way around.

If you want to connect with me more, don’t hesitate to reach out on Twitter or LinkedIn.

I first downloaded Instagram in early 2011 when a grad school friend of mine told me to check out a classmate’s profile. Facebook was the dominant social network at the time, but to me it felt like a high school reunion that lasted an hour too long. Instagram reminded me of what it felt like when I first got Facebook after a half-decade of using Myspace. It was more exclusive, the photos were more interesting, and the people were more relevant. In hindsight, I was exactly the kind of user Instagram was designed to attract.

Instagram’s Origins

Instagram was borne out of a classic startup pivot. The founders — twenty-something Stanford alums Kevin Systrom and Mike Krieger — had been working on a check-in app called Burbn that helped people see where their friends were hanging out. Burbn had a feature that let people attach photos to check-ins, but its main focus was checking-in. Once Kevin and Mike realized their community wasn’t growing fast enough to support a business, they decided to pivot: instead of focusing on check-ins, they’d focus on photos.

When they told the Burbn community about the pivot, many refused to go along with it. They didn’t want another photo app, they wanted check-ins. Despite the blowback, Kevin and Mike had conviction and decided to move forward with the pivot.

The Mobile Photo-Sharing Market

The conditions they saw in the mobile photo-sharing market in 2010 weren’t much different than what someone building a new photo-sharing app might see today. The market was saturated with new apps and dominant incumbents — even at the time Apple and Facebook had a vice-grip that seemed impenetrable. Friends and family questioned their decision; after all, who really wanted another photo-sharing app?

But Kevin and Mike had tried every competing product and knew that people had problems they could solve. They chose to focus on 3 specific problems in the market and build their community around people who cared most about those problems.

Problem 1 — Slow upload times

Mobile networks were much slower in 2010 than they are today, so photo files took a long time to upload. People also liked uploading many photos at once to their albums online, so the slow pace of upload caused people to just wait until they were in front of a computer.

Instagram’s solution was two-fold:

1. Target people sharing single photos in-the-moment instead of albums after-the-fact; and

2. Start the upload as soon as users selected photos instead of waiting until after editing.

This strategy opened up a base of travelers and socialites who didn’t have great options for sharing photos widely while on-the-go.

Problem 2 — Poor picture quality

Again due to the limitations of mobile networks, photos often got degraded during upload and left them looking worse for wear. Kevin and Mike developed a few beautiful filters that obscured imperfections and applied their iconic square formatting so people would know where the photos came from. Unlike unedited mobile photos, Instagram photos looked distinctly like the Polaroids that had transformed a previous era of photography.

Inside the app there was no backdoor around filters: users had to consider filtering their photos to share them. By branding Instagram photos beautifully and making them accessible, Kevin and Mike opened filtered photos to a much wider audience on the Internet than the passionate early-adopters who used them at the time.

Problem 3 — Sharing cross-platform

As remains true today, iPhone users in 2010 had many different platforms where they could share photos. The platforms were different (Tumblr, anyone?), but the problem remained that it was tedious to switch between apps when sharing photos. Kevin and Mike reasoned that they could get users who wanted to share faster by helping them post their photos to other platforms all at once. In a 2016 interview, Kevin compared this part of their strategy to the Lord of the Rings (“One ring to rule them all”); in his eyes, Instagram would be one photo app that shared to all.

Go-to-Market

Armed with solutions to 3 pain points, Kevin and Mike still needed to solve the cold start problem. No one would care about sharing on Instagram unless it had people and photos worth looking at. They knew some Burbn users would make the leap, but many more would not. This was a bet-the-company decision. Who would be the anchors of their new community?

They started their beta by emailing 100 creative professionals — primarily web designers, photographers, and their friends. The choice was crucial because they needed people who already had tasteful audiences. If they could convince a handful of creatives that Instagram was worth adding to their toolkit, they could also convince hobbyists and casual observers to join as well. The strategy worked like a charm. Hoping for 10,000 users within the first month, Instagram attracted 25,000 users on its first day.

Insight 1 — Scale doesn’t matter early on

The first insight from Instagram’s origins is a counterintuitive point about scale. Founders of social apps usually spend too much time worrying about how to get massive numbers of users. Some of this is downward pressure from investors who want evidence of network effects, some of it is their own intuition about what product-market fit looks like for social apps. Instagram showed that people don't get their sense of community from scale, they get it from regular interactions with a relatively small number of people. They didn’t start with billions of users, they started with 100 creative professionals. From there, they segmented their audience further and came up with solutions for segments of their user base. They didn’t focus on being massive until they were already there. If it were true that social apps need scale to get off the ground, none of the giants would exist.

Insight 2 — Antisocial actions can work

The next insight from Instagram’s success is that successful social actions are often quite antisocial. Imagine a teenager sitting alone in a car editing a photo they just took. That’s probably not a scene any of us would think of when describing the most “social” people we know, but it was one of Instagram’s primary drivers of social growth. There's not much fundamentally social (in the sense of forming deep, lasting connections) about posting filtered photos of yourself, recording videos of yourself lip-syncing and dancing to songs, or writing pithy faux-intellectual messages. And yet these behaviors each form the basis of giant communities online. Many of the app-based communities that we're familiar with promote similarly antisocial behaviors. We’re willing to stomach antisocial activities because we want convenient connections, not because we want the underlying social action.

Insight 3 — Young users are more social

The third insight from Instagram’s success is about targeting. Instagram took off with a relatively young base of users. The same was true for Snapchat, TikTok, Facebook, Twitter, Tinder, and host of other social apps. Younger audiences tend to have a lot more time on their hands to worry about their social lives and experiment with behaviors that aren't naturally appealing to adults with full-time jobs and children to care for. Starting with a younger audience actually helps social apps attract busy late-adopters because it gives them a different vector for evaluating the social behavior. Instead of wondering why they should bother with another app doing another thing they’re unfamiliar with, they wonder what the crowd of kids is looking at. The late-adopters are like drivers who slow down when they see an accident surrounded by first responders on the other side of the highway.

Insight 4 — Don’t be afraid to pivot

The fourth insight is about the importance of bold iteration. Kevin and Mike could’ve done what many other overly optimistic founders do and held onto the belief that Burbn would unseat Foursquare. Given Burbn’s base of users and VC funding, no one would’ve held it against them if they had continued to iterate on check-ins. But instead, they decided to scrap most of their product to run a new experiment. Their experimentation wasn’t random though, they pivoted towards their most compelling use case and they were willing to fire users to run a good test. Milquetoast pivots rarely work in early stage social apps. Had Kevin and Mike been unwilling to take bold action, they’d likely have become an acquihire or ended up in the graveyard along with most of the other check-in apps of the era.

Insight 5 — Hardware updates open new doors

The fifth insight has to do with timing. Instagram launched months after the iPhone 4 came out with a high-resolution screen, a powerful rear-facing camera, and its first first front-facing camera. These features greatly expanded the utility of mobile photos by making them prettier and empowering users to take selfies. Just as people began to realize they could take near professional quality photos with their iPhones, Instagram came along and gave them a killer use case. As the number of iPhones grew, so grew Instagram as the default broadcasting service for iPhone photos. There’s a lot of luck in choosing the right time to launch a new app, but new development platforms generally provide better growth opportunities for small developers than legacy platforms — especially in social media.

If you want to connect with me more, don’t hesitate to reach out on Twitter or LinkedIn.

As a kid, I found it really hard to find new musicians I liked. My only options real were the radio, my friends, and my parents — which meant I mostly listened to Top 40 hits, Reggae, and Pop music from the 70s. I didn’t get my own CD player until I was in middle school and I didn’t use it very often because I didn’t want to spend my allowance buying CDs. It wasn’t until I got an iPod that I started trying seriously to find music online. Suddenly I went from having no way to find music to having too many ways to find good music. I used every file-sharing service I could get my hands on, then switched over to iTunes, then Spotify, then YouTube, and finally to SoundCloud. SoundCloud was like having a cool friend who always knew which underground artists were about to blow up. It had its ups and downs along the way, but it knew what I liked and consistently delivered new artists I never would’ve found without it.

SoundCloud’s Origins

SoundCloud was founded in Sweden in 2007 by sound designer Alex Ljung and electronic musician Eric Wahlforss. As music creators themselves, they were sick of using email and Myspace to connect with other musicians. Their plan was to replace those outdated platforms with SoundCloud. If they could get enough musicians to join, they believed SoundCloud would do for music what Flickr did to photos and Wordpress did to websites: empower anyone become to become a creator. Musicians would use it to collaborate and share recordings; then eventually fans would use it discover artists and listen to their music with other fans.

Problem 1 — Making music social

As noble as their ambitions were, Alex and Eric had a monumental challenge ahead of them. They were attempting to start a social music app right as the biggest social network (Facebook) and the biggest music app (Spotify) started to ascend. They’d never be able to match Facebook’s breadth of social features or Spotify’s catalog of mainstream artists. So they focused on a segment both of those platforms overlooked: independent musicians. They built their initial community of budding musicians by giving them tools to share music with each other online. Once they had enough musicians, they started focusing on fans and the social experience. They nested commenting in the songs themselves, allowing fans to see how other fans reacted to every part of a song while listening. They let fans make and share playlists with each other. The result was a unique blend of music and social features that no one else had.

Problem 2 — Making money

SoundCloud didn’t seriously start trying to make money until it had over 175m users. By then it was already mired in controversy, as users kept uploading songs they weren’t licensed to distribute. Hosting music illegally put SoundCloud at odds with the major record labels, who had seen this story before with file-sharing sites like Napster. The labels had also already signed deals with Apple, YouTube, and Spotify, so they didn’t need SoundCloud to make money. They demanded even better terms than they got from the other platforms and the initial negotiations stalled.

Without any major labels on board, SoundCloud’s money problems spilled into public view. The company was bleeding cash and it was still unclear how they’d make money. Each failed fundraising attempt seemed to generate new negative headlines and questions about SoundCloud’s viability. It always seemed to be up for sale without any takers, which damaged its credibility with artists and fans.

SoundCloud’s big break came when it replicated YouTube’s business model using pre-roll audio ads, channel sponsorships, mobile display ads, and native content. This model had the added benefit of bringing the major labels back to the table. Warner Music was the first major label to sign on and gave SoundCloud momentum in its new business model. Universal Music and Sony Music weren’t so enthusiastic though. They wanted SoundCloud to reconsider its business model again before signing. From their experience with Spotify, the labels knew that subscriptions paid more revenue per stream than ads, so they wanted SoundCloud to release its own subscription and restrict non-subscribers from listening to parts of their catalogs. SoundCloud accepted and started competing in earnest with Spotify and Apple Music.

Insight 1 — Fundraising is never that easy

The first insight from SoundCloud is about fundraising. Most founders assume that social apps can always raise money as long as they’re growing quickly, so they don’t have to think about making money until later on. SoundCloud is an important counterexample to that narrative. Even with 175m users and rapid growth, SoundCloud couldn’t get investors at a $1b valuation. It went on life support for half-a-dozen years while it figured out its business model and only just turned a profit 15 years after launching. If the team had tried to make money sooner, they could’ve saved themselves a lot of heartache and potentially become a much larger threat to Spotify than they are today.

Insight 2 — Think different

The second insight from SoundCloud is about its positioning. SoundCloud has the largest music catalog of any subscription music service with over 150m songs. While many of those songs have little if any streaming activity, they say something about SoundCloud’s strategy. Instead of trying to beat Apple Music and Spotify head on, SoundCloud competes by helping relatively unknown musicians build followings. Some of those unknowns go on to find mainstream success, but fans associate the service with smaller musicians who they can’t discover anywhere else. By positioning itself in the long tail of music distribution, SoundCloud has a moat that protects it from both incumbents like Apple and new entrants like TikTok.

Insight 3 — Build a product you want yourself

The third insight from SoundCloud’s success is about the strength of its vision. Lots of startups fail because the founders don’t understand their customers well enough to build a product they want. The easiest way to remedy that is to build products that solve your own problems, as Alex and Eric did. They were each music creators themselves and they deeply understood the challenges musicians faced when sharing music online. On the surface, SoundCloud seemed to be in no-man’s-land competing with companies who had the edge. But Alex and Eric’s professional experience gave them confidence that musicians would work with them anyway. By solving their own problems, they carved out a place in a cutthroat market among a sea of more profitable competitors.

If you want to connect with me more, don’t hesitate to reach out on Twitter or LinkedIn.

Several years ago I had a group text chain with a few tech founders. Most of the time we used it to support each other, but we’d also share memes and make fun of the hard parts of starting companies. One day, someone shared a link to a short video of an elderly man playing an acoustic guitar and singing Bobby McFerrin’s classic feel-good song “Don’t Worry Be Happy.” The video was hosted on TikTok, which I had read about but never tried. I downloaded the app, listened to the song, and spent the next hour scrolling through short videos. One moment I was watching dancers bend their backs in ways that seemed humanly impossible with trap music playing in the background; the next, a mini-lecture on the origins of metallurgy. TikTok was irresistibly chaotic, and I was hooked.

TikTok’s Origins

TikTok was created by a Chinese internet company called Bytedance. They created several popular Chinese apps before TikTok, but their first big hit was a news app called Toutiao (meaning “Headlines”) that launched in 2012. Toutiao was like Facebook’s News Feed; it used sophisticated algorithms to analyze what users were looking at, measure how much they liked it, and find similar content. People didn’t have to set their news preferences, the app would determine their preferences based on their actions. In 2015, Toutiao launched a video feature for creators to share short videos and appear in news feeds. Chinese users loved it. One year later, the feature had turned Toutiao into China’s largest short video network.

Having cornered the Chinese market for short videos, Bytedance moved to win it outright. They created a new app that focused entirely on watching and sharing short videos. The app was called Douyin, but you probably know it by its international brand name: TikTok. It wasn’t easy for TikTok to transition from a Chinese corporate experiment to an international powerhouse, but we can learn a lot from how they navigated the problems that came up.

Problem 1 — Launching in China

Bytedance faced stiff competition getting Douyin off the ground in China. WeChat — generally thought of in the West as the Facebook of China — was a dominant force in social networking with over 800m users. Live-streaming app Inke topped 250m downloads. Short video app Meipai had over 100m users. They were entering a crowded space and needed a unique angle.

They decided to make Douyin a more fun and interactive video app, similar to the lip-syncing app Musical.ly. But they also planned to include their best innovations from Toutiao, like its feed and sorting algorithms.

To solve the cold start problem, they leaned on Toutiao’s distribution and scale. By getting Toutiao’s most popular creators to share their content on Douyin, Bytedance legitimized the new app. Within a year of its 2016 launch, Douyin grew to 100m users and 1b videos viewed per day.

Problem 2 — Launching outside of China

Bytedance thought the idea behind Douyin could work globally, so they took three major steps to expand out of China:

1. Created a new brand under the name TikTok that used the same technology as Douyin, but kept Chinese users / content separate.

2. Localized TikTok in every market they entered so users would primarily see content in their own language.

3. Acquired similar apps with large international user bases and merged their user bases into TikTok.

The two apps Bytedance acquired and merged into TikTok were Musical.ly and Flipagram. Both allowed users to add music to short videos, but Musical.ly was far more successful at the time of its acquisition. Bytedance reportedly paid $1b for Musical.ly’s 200m users, many of whom were young Americans. All user accounts and data ported to TikTok, and existing Musical.ly users had no choice but to switch.

Go-to-Market

Bytedance understood that teenage and twenty-something Americans were critical to TikTok’s growth prospects, especially after they bought Musical.ly. So they focused on their primary competition in that segment: YouTube, Instagram, and Snapchat. If they could get people who used competing apps to join TikTok quickly enough, it would both expand TikTok’s user base and make it more difficult for competitors to launch competing features.

They used a combination of paid marketing and organic marketing to attract new users. On the paid side, they launched one of the most aggressive social media marketing campaigns of all time. By some estimates they spent over $3m per day in 2018 with much of it focused on Instagram. In contrast to Instagram’s polished aesthetic, TikTok videos were more realistic and fun. People danced and played music and sang, which stood out in a feed full of filtered photos.

On the organic side, TikTok made it easy for users to post their videos to other platforms straight from the app. TikTok’s watermark distinguished videos its users posted and served as de facto ads everywhere they appeared.

Insight 1 — Chinese tech companies can win in America

The first insight from TikTok’s success has to do with the quality of Chinese technology. Every one of TikTok’s large competitors copied elements of its core functionality. Instagram created Reels, YouTube created Shorts, Snapchat created Spotlight, and so on. These efforts were defensive, intended to hold onto users spending increasingly more time on TikTok. It’s not often that a Chinese company comes to America and beats trillion-dollar companies at their strengths. TikTok’s success is a sign that Chinese internet companies are capable of leading industries on the global stage.

Insight 2 — Take care of the little guy

The second insight from TikTok’s success is about its commitment to regular users. TikTok recognized that most social media platforms promote influencers at the expense of normal users. It’s discouraging to see people get millions of likes only to have your own posts get 0. So TikTok decided to go the other way and give new users reach to encourage them to post more. The decision established its reputation for being more approachable than other social networks. It also attracted users from other networks who felt they were stuck and couldn’t grow their followings.

Insight 3 — Talk to your users

The third insight from TikTok’s success relates to its method of product development. TikTok’s focus on music videos came from close observation of how teenagers used their smartphones. Its leaders constantly asked users what features they wanted and tested new features with their users before rolling them out more broadly. Product development followed user feedback instead of mandates from senior management. As a result, TikTok cultivated a stronger sense of community than most of its competitors.

Insight 4 — New segments make new opportunities

The fourth insight from TikTok is about its market segmentation. Like cable television before it, TikTok introduced a range of new video segments based on length. But its segments were notably shorter. It started with 15 second and 1 minute segments before later branching into 3 minute segments, live video, and others. Part of the reason their segmentation was so effective is that their segments were underserved in other apps. They pioneered the market for 15 second music videos and that gave them a wedge to introduce more categories over time.

Insight 5 — Don’t be afraid to break conventions

The fifth insight from TikTok is about their willingness to ignore conventional wisdom so they could grow quickly. Many venture capitalists would recoil at the idea of spending billions of dollars to buy your way into international markets, especially when those markets are occupied by Facebook and Google. The conventional approach would be to develop your product until you find product-market fit, then grow through word-of-mouth from your users. But TikTok prioritized speed over all else. They trusted that they’d be able to figure out product issues with a household name brand, so they put everything into growth. It’s hard to argue with their results — they won.

If you want to connect with me more, don’t hesitate to reach out on Twitter or LinkedIn.

Several weeks ago I met with a group of students who were studying entrepreneurship at UCLA. Each student had spent the last 6 weeks developing business plans for startup ideas they had. The ideas spanned a wide range of industries, from healthcare to social media to crypto. My job was to offer perspective on what would happen once they took their business plans out of the classroom and into the real world. I could remember vividly what it was like a decade ago when I was one of them — bright-eyed and armed with my own ambitious plan to take the startup world by storm.

I started by asking the group what they thought makes a startup successful. A young man raised his hand.

“You need a great idea that solves a problem,” he said.

It was a plausible answer; after all the first page of their business plans focused on summarizing the purpose of their new ventures.

“Anyone else?” I asked.

This time a young woman answered.

“The best startups change something fundamental about people’s lives,” she said.

Another plausible answer. I polled the class until everyone who had raised their hand got a chance to answer. Each student who responded gave some variant of the first two responses. The consensus they reached was that great ideas make startups succeed.

I considered leading with the same speech about execution that investors had given me over the years, but I stopped myself. It never really worked on me, so why would it work on them? Instead I decided to ask more questions to tease out a more important point that I hoped they’d carry with them through their journeys.

“Do you guys use YouTube?” I asked.

Everyone nodded.

“What was their original idea?” I continued.

They gave each other puzzled looks, unsure where I was taking them. I then recounted how YouTube started out as a dating website before pivoting to video-sharing.

“Ok what about Instagram, how did that start?” I asked.

Again I was met with confused silence. I explained how Instagram started as a check-in app before pivoting to photo-sharing. We went through the exercise several more times with startups like Pinterest, Slack, and Android to drive the point home.

The students weren’t wrong that great ideas make startups successful, they were just missing a piece of the puzzle. Like many people, their view of startup success was shaped through the lens of scaled products rather than early prototypes. Since coverage of successful startups tends to emphasize superficial milestones like fundraising events and acquisitions, it’s easy to miss deeper insights about how they achieved newsworthy outcomes. Furthermore, early-stage founders usually don’t have the time or notoriety to attract in-depth analysis like their later-stage counterparts.

Superficial analysis of startup success is merely misleading for casual observers, but it’s actually dangerous for early-stage founders. Here’s why:

1. Feature creep. It leads founders to build more features than they need to test their ideas. If you’re building a photo-sharing app and you think the standard is present-day Instagram, you may end up building a bloated prototype that isn’t good enough at any single feature to stand out above the competition.

2. De-motivation. It causes some founders to give up when their initial idea fails. If you believe that success is built on having a singularly brilliant idea and your first idea turns out not to be singularly brilliant, you can easily conclude (falsely) that you’re not cut out for the job.

3. Misdirection. It misses the point of how repeated failure shapes great founders. The best founders almost never start out with the skills they need to get from stage to stage, they learn what their startups need to succeed by trying things, failing, and learning from their mistakes.

The third point is especially important because it actually applies to every function in a startup. In the beginning phase of any new venture it’s all hands on deck. Founders have to talk to customers, build the product, handle customer support, fundraise from investors, hire people to join them, and so on. Over time, specialists come in and refine the duct-taped processes that the founders stitched together so parts that work well become repeatable and predictable. But the first versions of business operations are often as shaky as early prototypes of the product. If the founders fixate too much on trying to emulate the sophisticated practices of large companies, they’ll waste time they don’t have.

A better mental model to use for untested startup ideas is more along the lines of a newly planted seed. Like a seed, an untested idea exists only under the surface. To the naked eye, there’s nothing there yet. Just as the seed needs the right amount of water, sunlight, and space to break out of the soil and grow into something meaningful, the untested idea needs feedback, customers, and a motivated team to grow into a meaningful company.

But even if everything goes right for the seed — or the untested idea — external forces can easily kill it before it grows up. A storm could drown it; an animal could trample it; a drought could starve it. So too with the untested idea. A downturn could push it into bankruptcy; a competitor could steal its customers; a chronic illness could cripple its founders. These risks aren’t always fatal, but they’re far more dangerous for untested ideas (and newly planted seeds) than they are for their grownup counterparts.

The seed model for untested ideas isn’t all that novel; it’s the theory behind the modern venture capital industry, which colloquially refers to initial rounds of funding as “seed” rounds. In fact the most storied venture capital firm — Sequoia Capital — intentionally named itself after the Sequoia tree with the seed model in mind. Sequoia invested in companies like Apple and Google when they were just seeds themselves, hoping the capital they provided would help the founders grow their fledgling ideas into the giant companies they ultimately became.

At the end of my session with the group of students, one of them asked me why the origin stories of startups don’t get more attention. If they’re so important for founders to understand, surely people who write about entrepreneurship should emphasize them more. While there some like Paul Graham who actually do, her point was well-taken. Many of the articles written about startups are little more than press releases that downplay just how long and circuitous the journey really is. They focus on trees instead of seeds. I told her I thought it was because of survivorship bias, but I’d write a few essays myself to help fill the gap. There’s clearly a big audience interested in forests, but it’s far more useful to focus on the seeds when you’re just starting up.

EDIT: Since publishing this essay, I wrote a series of posts on the origins of well known startups including Instagram, YouTube, Spotify, TikTok, and SoundCloud.

If you want to connect with me more, don’t hesitate to reach out on Twitter or LinkedIn.

There’s a scene in the first season of HBO’s brilliant police drama “The Wire” where detectives investigate the killing of a witness who was poised to testify in a murder trial. They had evidence that a drug dealer’s crew killed the witness, but not enough to prosecute them. Eventually, a rival drug dealer named Omar (played by the late great Michael K. Williams) comes forward to testify against the crew. The detectives are shocked. Snitching is the ultimate sin for a drug dealer in West Baltimore, so why would Omar risk it? During his interview, Omar explains that the crew had crossed a line by killing innocent people. “A man must have a code,” the detective responds.

I watched that scene several times over the years and each time I smiled at the irony of Omar’s moral code. But the joke was on me — I had never really defined my own code, so who was I to judge Omar’s? In a sense, a moral code is to a human what an operating system is to a computer. Just as an operating system provides parameters for applications to function, a moral code provides parameters for humans to make decisions. When Omar faced a moral dilemma, his code prioritized justice over his reputation and personal safety. Given a similar fact pattern, I didn’t think my code would lead me to testify as Omar had. But the only way to know for sure was to define my code and test it.

To start, I had to take stock of what my principles actually were. What sort of person did I want to be? How would my friends and colleagues describe me? I made a Notion document where I wrote out my principles and attached short explanations of what they meant to me. For example, I’m organized, so I wrote “Organization: I build effective structures in my life and the lives of others.” I repeated this exercise until I had 10 different principles written down. Then I asked a handful of my friends and colleagues what they thought my principles were. That gave me several more I hadn’t considered, like loyalty, clarity, and impact. I repeated the process for all of my various roles across my personal and professional life. As with my principles, my roles were broadly defined and exhaustive.

My code was almost done, but it was missing something. I wanted to try to answer the question in the middle of the exercise: what’s the point of this? Why did I need a code? These were deceptively simple questions. I started the exercise because I flashed back to a TV show - not exactly the purpose I had in mind for my life’s most important guiding document. Eventually I landed on “To live an exceptionally positive and successful life”.

I suspect this is normal, but before drafting my code I believed myself to be far more principled than I actually was. My demonstrated value system was a patchwork of behaviors I picked up passively to avoid punishment and reap financial rewards. For example, I learned to be respectful by disrespecting my brothers and getting grounded for it; I learned to be hardworking by getting bonuses for working overtime. My values weren’t innate or self-directed, they were externally-driven. In decisions big and small, I mostly just mirrored the behavior of people I looked up to. So when push came to shove, my decision-making was reactive and inconsistent.

Something interesting happened once I wrapped my mind around all of this: I felt relieved, like a weight had been lifted. Part of my code is self-belief, so I forgave myself for being inconsistent and focused more on how I could live up to my code going forward. The answer was obvious: just use it to make decisions. So I started measuring big looming decisions against my code and murky outcomes became clear as day. Asking the simple question, “does this fit with my code or not” is a powerful tool when you’ve taken the time to articulate what exactly your code is.

To put it mildly, the results have exceeded my expectations. I stopped spending time with people who made it harder for me to follow my code and invested more in relationships that make it easier. I stopped worrying about founding a billion dollar company because it’s not in my code. I resolved a long-standing feud with kindness and integrity. In short, I’ve become much more of the man I’d like to be. A code isn’t just aspirational, it’s a framework for introspection and self-improvement. It creates the kind of conviction that motivated Omar to risk his life and reputation to flip on his rivals. Put to Omar’s standard, my code actually dictated that I testify, just as he did.

View My Code (Notion Doc)

If you want to connect with me more, don’t hesitate to reach out on Twitter or LinkedIn.