“We should have built more automated testing earlier” — usually, you know that it was failing and investing time in testing would not have helped
You will be more successful with:
“We should have decreased scope in X way”
“We could have de-risked by doing X”
Other failure modes:
Assuming simplicity in someone’s domain
Other things I’ve learned recently:
Even if you are not responsible for implementing change, you can be responsible for highlighting problems and sharing the state of the world, and shining a light, even if it is embarrassing.
One of the most useful things you can do is have data that is presented well.
There’s a difference between saying “Ok, I agree” and then feeling responsible for the risks vs. saying “Ok, I agree, but here are the risks” and having the peace of mind that you pointed out all the problems (and that your party is also aware of these problems)
As consumers, we tend to think that we have completely free choice — but the truth of the matter is, that many forces beyond our control frequently give us a default — the ultimate goal of any business is to become the default. In the past, the way to become the default choice was simple: land. Land helped propel McDonald’s into becoming the most popular fast food chain in the country. Land is also the reason you see two Starbucks within a block of each other. More storefronts translates into two things: (1) more convenience (easier to reach by proximity) and (2) increased familiarity (trust over unknown choices) — each additional store brings customers closer to making that business their default choice.
Winning with Proximity and Familiarity
Default choices are hard to resist. In 2002, Google reached a deal with AOL to guarantee a payment of $50m if AOL made Google its first-choice search engine shown to users by default. Today, Google reportedly pays Apple billions of dollars a year — $9 billion in 2018 and $12 billion in 2019 — to remain Safari’s default search engine. Google’s annual payment to Apple is an example of why proximity (convenience) remains important despite familiarity (trust) — I may believe Google is superior to Bing but will still use Bing if I have a browser that defaults to that because I am lazy.
Some businesses struggle to become the default, even though their product is already proximate to consumers everywhere. For example, shampoo is known for having poor brand loyalty — 73% of people change their shampoo occasionally. People change their shampoo sometimes because there are other options at the places they go to buy their shampoo. Shampoo brands cannot win with convenience. Contrast this with the automotive industry, where 61.5% of Subaru customers end up repurchasing another Subaru. A car is a large enough purchase that you’re willing to travel a little bit and there does not tend to be options at similar distances. For example, I live in San Francisco — there isn’t a Tesla on every corner while I have to drive all the way to LA to go buy a Ford. Proximity is less relevant for big consumer purchases.
In the world of land, shampoo companies couldn’t simply take over proximity, except now in the new world with Amazon, they can because they can just pop up as the Amazon top choice when someone searches “shampoo.” This is what Amazon is doing with creating their own brands — if the first three things I hit on the search page have high ratings and there are several variations of Amazon shampoo, I’m going to choose the top ones listed. Ordering is something that the shampoo companies cannot do at CVS, because there are dozens of bottles next to each other. While it may help to be on the top shelf, a consumer is going to browse around and generally it will be very easy for them to switch — it requires no more thinking and no more effort to switch. Amazon, on the other hand, has set up a new store, where I actually have to scroll to switch.
The Shift from Land to Digital Real Estate
Previously, land was the most important driver. Zara, for example, strategically positions its retail location in the luxury shopping district in order to target higher end customers. If a consumer’s default choice for jewelry is Tiffany’s, and Zara that is physically next to the default choice store they are already familiar with, they will be more likely to go to Zara than had Zara been located somewhere else. Retail stores play this game all the time by seeking to build themselves on the block where consumers already shop. In a similar manner, a consumer drives themselves over to Amazon to get the book because Amazon is the default. Once they get over there — surprise, they see the movie store next to it, Amazon Prime Video. This play is similar to the physical proximity that retailers were leveraging to win over consumers; however, this time, the proximity is taking place on a landing page.
Businesses no longer need physical real estate – the new battleground is now the internet. With this new battleground comes a narrowing of the number of pipes consumers can choose from. This phenomenon is already playing out in entertainment. Where someone used to spend time channel surfing or looking through the TV guide, now the default is to go to Netflix and watch a show recommended by Netflix. An increasing amount of watched content on Netflix is coming from algorithm-recommended content rather than a user clicking on something they had predetermined they had wanted to watch. Netflix seems like a one-stop shop because it has so much content, even though you end up just clicking whatever it recommended for you. We increasingly have only one portal not just for video — the same is also true for listening to music, as Spotify makes decisions for us about what to listen to.
While consumers have more choices of goods, they also have less pipes from which to consume goods. Contrast this to the past, where monopolies used to be prosecuted because lawmakers argued that they led to less choice and left consumers worse off. Now, monopolies are democratizing content creation and arguably delivering more consumer choice. The number of movie titles released, for example, has been steadily increasing over time, and hit its record high in the US and Canada in 2018. The number of companies funding this content, however, has been shrinking.
The Three-Step Flywheel
So what is the key — how do you build an empire that becomes every consumer’s default choice?
Step One: Become the one-stop shop
The first step is to become the one-stop shop for many choices — instead of going to individual sites or stores, consumers only need to go to one place. The convenience of on-demand – whether it’s being able to instantly watch a movie or being able to instantly cook a delicious homemade meal — further helps spin this flywheel to retrain consumer behavior. Step Two: Retrain consumer behavior
This one-stop convenience leads to retrained consumer behavior, where consumers are conditioned towards particular brand or service. Changing consumer behavior is the key to unlocking billion-dollar businesses. In 2011, I discovered a new service where I could stay at someone’s house, rather than a hotel when I was visiting a new city. When I told my friends about it, they all replied similarly, “I would never use it.” That quickly changed as Airbnb began to override previously socially accepted norms and learned behavior.
Step Three: Leverage your monopoly
This retrained habit leads to the third step: monopoly leveraging. A business can further become an entrenched monopoly by bundling products together to create leverage. Amazon, for example, has a monopoly in online retail and bundles Amazon Prime Video with Amazon Prime to move towards securing a monopoly in video. As a consumer, you already go to Amazon for retail and now you think, “Oh, I can start going to Amazon for my entertainment too.” This is how companies use monopoly leveraging to retrain consumer behavior. Amazon implicitly tells consumers, Why not stay here and you can do it all. Monopoly leveraging is naughty — it’s a way for companies to get an unfair advantage in a completely different product category and essentially get customer acquisition for free. Twenty years ago, Microsoft employed this same tactic when they dominated the operating system (OS) space. They took advantage of their monopoly in operating systems to get into the browser space, replacing Netscape and instead packaging its Windows operating system with their own browser, Internet Explorer, to win the browser market. Essentially, Microsoft used one monopoly to create another monopoly. This is a great deal for Microsoft because get essentially free customer acquisition – whereas the customer acquisition cost (CAC) would have been very costly for a business that did not already have a monopoly.
A Coming Shift
Increasingly, a fewer number of businesses are owning more and more categories. Simultaneously, there will be entire categories where consumers are eliminating their need to choose and instead empowering online portals to do it for them. However, there is a missing element that made brick and mortar businesses a default choice that has not been perfectly replicated in the digital space yet.
It is not sufficient to have a web presence or even a very popular one. A digital brand can become familiar, and can make its varying services proximate to each other (e.g. Google Search and Google Shopping Express) – and yet, if it cannot serve the customer in a timely manner, it will fail. Put differently, internet producers have not yet overcome one brick-and-mortar advantage: physical proximity of good to customer. Amazon recognizes this — the reason for Amazon’s acquisition of Whole Foods was not the Whole Foods grocery business, but the real estate footprint. Amazon now has Whole Foods locations that can double as distribution centers all over the country. Where many businesses have failed and gone out of business because of high shipping costs and mismanaged distribution, Amazon has succeeded.
In the next essay, I will dig deeper into this next battleground and cover the qualities of those who have failed and succeeded.
Thank you Mike Ng, Chris Pagliarella, Savannah Perkins, Alexis Kim, Suthen Siva, Jessy Lin, Ash, and David Perell for reviewing earlier versions of this essay.
Paul Graham once tweeted, “Anything badly broken is a dam holding back a lake of unrealized desires. A startup that can bore a hole through such a dam can liberate all that energy.” I like to think of it more simply: sometimes the best way forward in life is through your own misery.
Oskar Barnack had asthma and couldn’t go outdoors for long periods of time, which is what partially drove him to invent a lightweight camera, the Leica.
Google Maps was started by 2 Danish brothers, Lars and Jens Rasmussen who were laid off from their startups during the dotcom burst. After moving back to live with their mother in Denmark, they realized they had a poor sense of direction and built a personal maps project.
Sophie Wilson, who invented the low-power chips we use today, failed out of the Math Department at Cambridge, so she switched to Computer Science, where she combined her mathematical and computing expertise to invent the chips.
Scandinavian nations pioneered wireless before many others out of sheer necessity — routing telephone wires through vast expanses of rocky, snowy terrain was difficult… The Nordic Mobile Telephony system, which was established in 1981, marked the re-conception of the phone as something that could — and should — transcend borders, and reshaped the way that people thought about mobile communications, from an implement useful in local markets, to a more general, universal tool. (Merchant, The One Device)
I asked my co-worker Blake recently about his feedback on an escalation path I had written for reporting issues with Tesla’s in-car entertainment system. He was very good at thinking in terms of frameworks (“Is this supposed to be an algorithm that customer service follows to a T or rather guiding principles?”), alternative solutions (“diagrams are more consumable than lists”), and edge cases. I told him he thinks surprisingly long-term for someone at Tesla, or in any fast-paced startup environment where the default is to rush urgently to solve a problem, and then sprint to the next fire. He said for him it’s important to take things slowly, because he can think through why it happened and make sure it doesn’t happen again. I realized in this way, he develops a repository of repeatable approaches to solving any kind of problem. My boss does this too. People say she’s going to run the entire org very soon. I notice she thinks carefully about everything — whether she’s negotiating a business deal with Spotify, or phrasing the wording of an email with a random startup who wants to partner with us. This allows her learnings from each situation to compound, so that whenever she runs into a problem, for her it’s always: “just another one of those x issues.”
This reminds me of one my favorite quotes, “Even though your work seems very trivial and contemptible, make sure you regard it as great and precious, not on account of your worthiness, but because it has its place within that.“
I recently had the opportunity to sit down with Li Hongyi and discuss Singapore’s approach to government and technology. I’ve been fascinated by Singapore since reading From Third World to First by Lee Kuan Yew, the nation’s founding father, and how he took a poor and corrupt country and built it into one of the richest countries in the world — where the average per capita income is nearly twice that of America’s.
1. Singapore is nearly twice as efficient with government spending as America is. Their government expenditures make up 17% of total GDP — and not only do they have universal healthcare, but also the government manages the retirement savings of all its citizens (and invests that money). The US’s government expenditures are 38% of total GDP. A Rube Goldberg machine is an example of how an arbitrarily simple task can be arbitrary complex. This is how most governments work.
2.What worries you about the culture in Silicon Valley? He responded that despite all of the wealth and abundance in Silicon Valley, people tend to accept wealth disparity and move to higher levels of esoteric consumption (taking on more interesting hobbies, eating at nice restaurants, traveling to more exotic places). Buying experiences is not better than buying things. People here look at the homeless and accept that that’s the way things are going to be. This is not the case in Singapore — people in Singapore expect a lot out of their government and will complain about even the smallest things — and the government is quick to respond.See the response to a citizen’s complaint about a grocery store not carrying a yogurt drink for his children (1):
3. “One way to see countries is as companies — what are the main pillars of government strategy?”
Li Hongyi runs the Data Science and AI division of the government — he responded that AI itself isn’t the main source of value, rather, it is automating all the things leading up to that.For example, the police was having a hard time tracking down peoples’ addresses when they would call to track down crimes. The police thought AI would solve their problems — in actuality, all they needed was a faster way to query. Hongyi used Amazon ElasticSearch to bring the query time from 2 minutes to a speedier 0.5 seconds, and their issues were resolved.
AI as a national strategy resource does not exist. Their competitive strategy is to attract smart people. What do smart people care about? Having a good quality of life.
Two competitive strategies:
Develop an edge – e.g. China focuses on IoT, manufacturing, electric, and autonomous vehicles; Japan similarly cherry picks industries
Just get the most successful people in our country – e.g. Singapore
The other day I had a difficult meeting where I was under a tremendous amount of pressure from our engineering directors. My boss J was sitting in it for the first time — J didn’t say anything during the meeting, but after the meeting, he gave me a step-by-step rundown of how the meeting went, the things I said, and how if I had framed things a certain way, I would have been more effective. Two things I learned from our debrief:
(1) Use detail as a tool to build trust.
There are little things, that when phrased differently, can have a big impact. In the meeting, I was under a lot of pressure to reach alignment with an Autopilot manager to decide whether his features were shipping with this release, but I was having trouble getting in contact with him all morning. You could explain the situation to the executive team in two ways:
Less useful: “I couldn’t get a hold of them.” (your bias for action is abstract)
More useful: “I tried calling them, but they didn’t pick up.” (your bias for action is concrete)
It’s a small difference in word choice but a big difference in how people perceive you as someone who is capable of driving results.
(2) Being decisive about the world enables you to make progress faster.
Most people who ask someone to make a decision, ask for a decision in an open-ended way. e.g. “what are you going to do about X?” Even the directors fell into this trap when they asked Autopilot manager, “What are we doing about these new features?” It’s much more useful to instead, define the world for them, and ask them to answer the question in the context of the world you have defined.
Less useful: “What are you doing about this new feature?” (the conversation goes on for 30m+ as we navigated through all the assumptions that needed to be made in order to answer this question)
More useful: “We are shipping this firmware in 5 days. Are you going to be on the train or not?” (by indicating when you are going to ship the firmware, you define the world for them)
You can always change the context later, but defining the world makes it easier for them to make a call, because you are scoping down the unknowns.
It was refreshing to get this level of feedback — most feedback at work is about overall performance than at a day to day level. I remember when I first started playing rugby at Brown, which went undefeated in the Ivy league for several years, I was amazed at how quickly I picked it up. At first, I arrogantly attributed it to my athletic dexterity — then I realized it was the nature of the training program — the immediate feedback from our coach on all the micro-drills created the perfect feedback loop. I was being given targeted guidance on exactly what I was doing wrong and why. And wasn’t just rugby — the same dynamic also helped me in coding. And it’s not just me — as people, we are beholden to making progress in environments with tight loops.
When you’re building a company or a product, how do you decide whether to own the entire stack or whether to outsource parts of the product? Owning the entire stack could allow you to create a better user experience (vis-a-vis WeChat and how you can text message, order a taxi, or make a doctor’s appointment all in one app), but sometimes this power is wielded in ways that stifle innovation.
I used to think it was better to outsource things you are not good at, and instead focus on your key strengths — then why is Tesla trying to bring manufacturing in house when they could move much faster following Apple’s footsteps and outsourcing their manufacturing to China? After all, the Chinese manufacturers have decades more experience producing electronics and cars and could much more easy troubleshoot problems than we could.
Bringing things in house is what has enabled Tesla to get so far and be the leading electric vehicle company in the world — having no credible competition in the electric car market for 5+ years. Cutting out the middle man enables three things:
Reduces client implementation complexity -> we can build more things!
Lowers costs – we’re not paying someone who is incentivized to charge more
To innovate – we can quickly iterate in-house
I’ve been reading the Master Switch: The Rise and Fall of Information Empires, and it’s fascinating to learn about companies who started with one core competency, and expanded moved things in house to gain a competitive edge. Radio Corporation of America (RCA) for example, originally started as a radio broadcasting company, but decided to start producing and selling radio sets so that they could control the frequencies that the radios could receive.
Are there industries then, that lend themselves more to vertical integration than those that don’t?
The answer is usually mixed. In the computer industry, for example, Microsoft is not vertically integrated but won most of the market share — Apple, on the other hand, is vertically integrated and won the lion’s share of not the market, but the profits.
Similarly, in the mobile phone industry, Google Android is not vertically integrated and has won the market (87% market share), while the Apple iPhone (12% market share)  has contributed to making Apple the most profitable company in the world .
On my plane ride recently to Portugal, the server asked me if I wanted anything to drink. I replied no. He said in a Portuguese accent, “Just one cup! The air is really dry here,” and thrusted a cup of water in my hand. I’ve never had this experience on an American airline. I was touched by his rare sense of ownership to people around him.
Once I got to Portugal, I was there for an overnight layover, but I ended up having one of the best nights of my life. I met up with a mutual friend, and even though it was my first time meeting her, I felt like I was meeting up with a long lost friend in a new country. She was disappointed that she couldn’t meet me for dinner because she had work, but gave me an itinerary of all the things I should do that evening — which restaurant I should go to, what I should order, and the order of the sights I should see around the city. She told me she’d meet at 11pm, so after I enjoyed the steak at Carvoaria, the restaurant she recommended, I ordered an espresso. Drinking coffee sometimes feels like pushing a button where you suddenly have twice as much life to live.
We ended up going to a discotheque until 5am.
I don’t like clubbing in America. It feels transactional; a game where people try to hook up. Two men in the corner of the club were trying to do the same thing as they approached two Portuguese women. The women looked very uncomfortable and kept moving away as the men tried to moved closer and twirled them around. They were not having it. Later that night, two of the Portuguese men from our group of friends were talking to the same two women. They talked for several hours on end. It wasn’t until several hours later into the night at 4am that they started dancing. I like how people take their time with each other. The question of “Are you coming home with me tonight?” felt absent.
It was as if I had been thrust into a more evolved version of America, a world where machines have automated all jobs and people are instead on Earth to enjoy life, value their friendships, and cultivate meaningful, non-transactional relationships. My best friend told me that you can think of life as traveling on a bus. Most people, as they travel through life, simply collect other people onto their bus. The passengers are either heading in their direction or they’re no longer on the bus. He doesn’t like to live life like that — he’d prefer to bring someone onto his bus and together, decide where to go. I like this kind of friendship much more — and I realize I don’t have to live in Portugal to adopt these values. When I travel, I learn about how things we consider normal are completely weird in a different context. I can choose what I like from all these different ways of living, and piece it back into my own life.
I meet many people who bash Snapchat or Facebook as products. They think of the products as a waste of time. That’s one way to approach making products. Another way is to do the opposite. Put Snapchat on a pedestal. Ask “What makes Snapchat addictive? How did they get to over a 100 million users? Can we take their strategies and leverage them to achieve our product x’s goals?”
You can adopt this strategy with your career choices too. Gary Gensler worked for Goldman Sachs and became a top manager (this is a career decision some would ordinarily pooh-pooh as selling out), but then he left and worked for the CFTC (Commodity Futures Trading Commission) to help fix the broken and corrupt derivatives market — and he was the only person fit for the job, because no one else in the government understood how derivatives worked. He eventually helped bring transparency to the $400 trillion previously opaque swaps market, transforming the sleepy CFTC into a powerful regulatory force.*