从狂热梦境到旷野：Drew Houston 讲 Dropbox 的三个时代

Houston started Dropbox out of personal frustration with file syncing, after forgetting a thumb drive on a trip to New York. Many of his friends had gone through Y Combinator and moved from Boston to California, and he felt left out. To get noticed, he worked backwards from the question of what Paul Graham does all day and guessed the answer was refreshing Hacker News.

Inspired by the book Guerrilla Marketing, he made a simple screencast of Dropbox that stayed at the top of Hacker News for two days. Graham replied that the idea was interesting but he needed a co-founder, with the next YC application deadline about a week or two away. Houston found Arash and got into YC, and says the early days were full of stories like that.

After Demo Day, angel investor Pejman Nozad introduced the founders to Sequoia, and they raised a seed round in 2007. At the end of 2008 they launched at what is now TechCrunch Disrupt, where the live demo failed because the wifi did not work. A second, meme-filled demo video posted to Digg and Reddit lifted the beta waiting list from 5,000 to 85,000 people overnight.

The team then applied engineering rigor to referral programs and shared folders, drawing on the epidemiology-inspired virality playbook of the social media era. For the first several years usage doubled or grew tenfold annually, and the team ran out of wall space taping up printed user counts and moved on to the ceiling. Valuation went from about $6 million in 2007 to $27 million in 2008 and $4 billion in 2011.

Houston dates the end of the first era to around 2013 or 2014, though Apple, Microsoft and Google had all launched competing products years earlier. He compares it to seeing a mushroom cloud in the distance without hearing it: the numbers showed nothing. Steve Jobs named Dropbox on stage in 2011 when announcing iCloud, and the team spent years waiting for the other shoe to drop.

He now sees competition as a boa constrictor rather than a shotgun blast. Being a product for everyone had been a blessing for viral loops, but people used Dropbox for backup, photo sharing and work collaboration, and those use cases pulled in different directions. An IT admin's ideal file server replacement looks nothing like a consumer photo app, and the operating system maker always has a built-in advantage.

To diversify, the team pulled photo features into a separate app called Carousel, built around keeping your whole life in the cloud while feeling local. They also acquired Mailbox, the mobile email client known for its waitlist, hoping it might become their Instagram. In 2014 Houston was on stage describing Dropbox as a new productivity suite, even as he sensed that winter was coming despite strong user and revenue numbers.

In 2015 Houston began hearing louder critics inside and outside the company. Dropbox was fighting on storage, photo sharing against Facebook, Snap, Instagram, Google and Apple, and productivity against Microsoft, Google and newer entrants like Slack. Then, less than a year after he stood on stage promoting Carousel, Google Photos launched with free unlimited storage for photos and video.

It destroyed Carousel's economics, and Houston felt it was an embarrassment because it was so easy to foresee. He realized every incumbent would copy a product, bundle it with its platform and kill the economics, which also applied to Mailbox since swipes and snoozes are not a durable advantage. To think it through he turned to AG Lafley and Roger Martin's Playing to Win, which pushed him to be selective about the markets where he could lead, and to see that Dropbox risked being second best in each of its markets.

Andy Grove's Only the Paranoid Survive supplied the template: Intel faced Japanese memory competitors even while its numbers still looked fine, and Grove and Gordon Moore asked what they would do as consultants to themselves. The answer was to leave memory and bet on microprocessors. Grove warns that CEOs usually want to hedge, while at an inflection point you should put all your eggs in one basket and watch that basket.

After rereading the book over a Fourth of July trip, Houston killed Carousel and Mailbox and went all in on productivity. It was a relatively easy decision because about 80 percent of paying subscribers used Dropbox at work. It did mean giving up photo sharing, consumer use and other things Dropbox was best known for.

Houston wishes he could say things then got better, but the opposite happened. Press and internal sentiment turned sharply negative, with articles every week or two predicting decline. He describes a washing machine of self-perpetuating negative coverage that other companies like Uber and Meta have also experienced.

Reporters, he says, effectively parked a van behind the office and printed what fired employees said anonymously. There was a lot of truth in the coverage, so he does not blame them. Recruiting froze, employees no longer wanted to wear the company T-shirt, and he did not either.

Internally, revenue had scaled much faster than the ability to hire and build infrastructure and operations. Everyone asked what the company was doing now that Carousel and Mailbox were gone, and Houston did not have a quick answer. People looked to him for fixes while also wondering how he had gotten the company into this situation.

Houston says the Apple launches were also initially panic-inducing, though less public, and that competing products took time to bite. He cites Netscape: Internet Explorer 1.0 to 3.0 were not very good, and it was the bundling of versions 4 and 5 that did the damage. Bill Campbell, who was at Netscape, told him "Microsoft did not kill us. We killed ourselves."

Houston says he came to understand this at Dropbox: the company hired smart people yet did seemingly stupid things, products stagnated, and the team was too confident in its ability to respond. Personally he went from feeling stressed but good to mostly feeling bad. He reminded himself that his 18-year-old self would not understand what he was complaining about, and that both feelings were true.

Mindfulness and meditation helped him separate his identity from the company, a fusion he says many successful founders experience. He also took comfort in realizing that most of his entrepreneur heroes had spent time wandering in the desert, which can be a crucible rather than a verdict. Campbell, a former Columbia football coach, would take him out to dinner, and Houston was surprised that he never seemed to be freaking out.

He also realized that chapter one had been a series of boxes to check, from a co-founder to YC to Sequoia to a billion-dollar valuation, and he no longer knew what Dropbox or he wanted to be. Inventing things ten minutes before Google and Apple did was not purpose. He concluded that he had chosen this life, that the learning curve was what he wanted, and that resenting the job or feeling like a victim would be corrosive.

Houston says a couple of unlocks came quickly, but he did not feel better and live happily ever after. He realized he was too busy firing to aim, and that big questions kept dropping below the fold of his to-do list. He began taking think weeks to get off the treadmill, and built a support system of meditation, therapists, coaches, mentors and founder friends.

He also had to face that he had driven the ship toward the rocks, since a founder CEO cannot blame their predecessor. As CEO, he says, strengths and weaknesses are amplified, and blind spots can become cultural dysfunction. The Enneagram helped because, unlike the descriptive Myers-Briggs, it is about what you are running toward and away from.

He describes himself as a type seven, the enthusiast, who is creative and loves new ideas but is bored by routine, conflict-avoidant and chaotic. Those traits meant the company was not telling the truth and employees were confused. The remedy is to work on yourself, hire people without those issues, or otherwise make the company less exposed to your dysfunction.

On the business side, the urgent problem was burning cash in a land-grab against Google on free storage. The company cut unprofitable parts of the business, turned cash flow positive in 2016, reached a billion-dollar run rate in 2017 and went public in 2018. Houston also set out to automate tedious parts of his job and learned machine learning in 2016 and 2017, which proved important later, and he redirected his goals toward the craft of being a great CEO rather than chasing bigger numbers.

Houston says founder mode means different things to different people. He cites Ben Horowitz's product-CEO paradox: companies die first from founders not letting go and second from founders getting too far away. He believes he had drifted too distant from the product, which contributed to the chaos at the end of chapter two.

He thinks the destination matters: after a learning journey, you gain conviction from lived experience and the confusion fades. The flip happens when a founder decides this is not the company they want to run and stops apologizing or negotiating all the time. He suggests that leaders like Elon Musk and Steve Jobs had spans of wandering in the desert before acquiring that conviction.

One of the hardest parts is hiring executives who know far more about their subject than you do, such as someone who ran a double-digit billion-dollar P&L at Google. It is easy to lean out and fail to set direction. He also describes a trap in which feedback collected through execs, HR and coaches turns into a very long to-do list of the CEO's flaws, which can let execs evade accountability.

Houston says his playbook is to get frustrated and solve the problem personally. After the thumb drive, his second frustration was how hard it was to scale a company, and he saw everyone on the same treadmill of meetings all day and email all night. He argues that people are most productive and happy with focus, flow, sleep and purpose, yet work culture rewards who is busiest.

He recalls asking a SpaceX engineering director how his team would get to Mars and being told it would take a lot of emails and files. Tools have gone from force multipliers to suppressors, he says, and the number of them has grown from five to ten to five hundred. He imagines Einstein getting Slack interruptions and says Slack-style tools chop the day up and leave empty carbs in collaboration.

The result was a new mission: Dropbox is designing a more enlightened way of working. Houston says the company was patient zero for the problem, with complacency and entitlement creeping in and attention drifting away from customers. After COVID, he saw an opportunity in decoupling work from location, comparing it to the biggest change in knowledge work since Peter Drucker coined the term in 1959.

Dropbox became a "virtual first" company that is about 90 percent remote, studied other remote-first firms such as GitLab and Automattic, and open-sourced its virtual first toolkit. Houston says retention, satisfaction and engagement have risen since COVID. The next step was to rebuild the product stack for distributed work.

Going distributed loses context, and remote-first companies must document more, which creates a search problem. Houston asked why he has one search box at home and ten at work, and why it is easier to search all of human knowledge than your company's. Around 2018 and 2019 he built a fuzzy, vector-search-based personal search engine for himself, and it worked.

After COVID and ChatGPT it became a company direction. Dropbox acquired a small company called Command E that did universal search and created Dropbox Dash, which connects to a user's apps and answers natural-language questions about their own information. It rests on a connector platform that indexes SaaS apps into a common representation, and it can answer things ChatGPT cannot, such as when your lease expires.

Houston says Dash embodies the problems Dropbox was founded to solve: finding, organizing, sharing and securing your stuff, which used to be files across devices and is now tabs and cloud tools. Dash also includes Stacks, smart collections that give content a container and make sharing easier. At the time of recording, it had launched two months earlier and signed its first customers.

Houston says the second chapter was partly internal, because the business scaled faster than the company could manage. He notes that many SaaS productivity companies suffer a sophomore slump after a hit first product, citing Dropbox, Zoom and Slack. Moving from a functional organization with one product to multiple products creates conflicts over investment and authority, and eventually requires business units, a common technical platform and shared functions, which Geoffrey Moore's Zone to Win discusses.

Culturally, success seeds failure: the outsider mentality that gets you to the top gives way to complacency and entitlement. Houston says he went from earning startup founder merit badges to the stagnation and irrelevance badge, and then the turnaround badge. The turnaround required telling the company that in three years customers had mostly seen a price increase, and resetting around craft, a growth mindset and high agency.

He rebooted the exec team and describes the talent flywheel running in reverse, with people wanting to work for Facebook rather than MySpace. In the late-2010s talent wars, executives had offers from big tech and pre-IPO startups at multiples of the pay, so Dropbox gave out double promotions just to hire. The resulting seniority gap meant too few experienced people to train high-potential staff, and he recommends roughly balanced mixes of the two.

Rachitsky notes that after growth, founders meet the final bosses, such as Apple, Google, Microsoft and Meta. Houston says each time you move up a league the reward is a stronger opponent and a possibly less level playing field. You cannot control that, only how you respond.

Asked whether people should become founders, Houston says he just showed up that way, and that he was insecure about being CEO as a technical person. He advises that it need not be an all-or-nothing decision, since you can keep calibrating, and recalls being told simply to try being CEO and see how it goes. He had once planned to get to a $100 million valuation and retire, but advisors in the Bay Area told him the day they sold their company was the saddest of their lives.

For him burnout is the biggest threat, which is why coping methods matter. He quotes Ben Horowitz that the hardest thing for a CEO is managing their own psychology, and adds that the challenge is not optional but the suffering is. He admits there were periods when he was sad and angry all the time.

Houston borrows from StarCraft: micro is the mechanics, such as design, engineering, distribution and viral loops; macro is the business model, market, competitors and organization; and the meta game is the shifting environment, including business cycles and how marketing changes. He cites the Dash launch, where conventional tech press produced little and direct social posting mattered far more. You need to be good at all three at once.

His most important advice is to keep your personal growth curve ahead of the company's. Reading history and business, from Netscape to Procter & Gamble, shaped Dropbox's decisions, and he shares a book list. He also recommends a community of founders at your stage and two, five and twenty years ahead, since peers give the best advice early and long-tenured founders become almost philosophical.

To stay ahead, he suggests asking what you will wish you had been learning in one, two and five years, and running toward the discomfort of things you are not yet good at. Skills such as management or public speaking are trainable over years. He also warns that smart people struggle to learn because being wrong feels like an assault on identity, recommending the essay "Teaching Smart People How to Learn" and the book The 15 Commitments of Conscious Leadership, and a habit of asking what if you were 100 percent responsible.

In his closing reflection, Houston says starting a company can forge character and builds skills that transfer to other domains, including being a husband and father. There is no easy button, he says, and the founders and CEOs he has met who built iconic companies do it because they love it, even when they describe it as chewing broken glass. Rachitsky adds that product-market fit is not binary and is constantly challenged by others.