WE MOVED THE COMPANY to Los Gatos in March of 1999. The new office was just over the hill on 17, as close to Santa Cruz as you could get and still be in the Valley. It was fourteen minutes from my house. A far cry from the five-minute drive I’d grown accustomed to, that first year. But long enough to get three, or sometimes four, run-throughs of Sweet Adeline, or Down Our Way, or whatever barbershop quartet song I was working on at the time.

广告：个人专属 VPN，独立 IP，无限流量，多机房切换，还可以屏蔽广告和恶意软件，每月最低仅 5 美元

Let me explain. A few years before Netflix, in the mid-nineties, Lorraine was worried that I was burning myself out. She suggested I get a hobby that had nothing to do with work. “You always sing in the car,” she said. “Why don’t you join a choir?”

I not only joined a choir, I joined the Society for the Preservation and Encouragement of Barber Shop Quartet Singing in America. SPEBSQSA, for short. No women allowed. (They had their own group: the Sweet Adelines.)

The society had chapters all over the world, and the closest one was in Santa Cruz. Every Tuesday night, there was a group sing in the community room of the Felton Bible Church. Anyone who was a member of SPEBSQSA could drop in, and because of the standard repertoire, there wasn’t any confusion, once you knew the songs and your part in them. Every sing started with “The Old Songs,” the official theme of SPEBSQSA. After that, the director would call out different songs, sometimes relying on member requests. And after two hours or so, sometimes we’d go out for a beer.

There are four voice ranges in barbershop singing. Tenor is the highest, followed by lead, baritone, and bass. Since there are only men singing, there are no alto or soprano parts, and the ranges are much closer together. That leads to really tight harmonies—singers used to the wider ranges of mixed choirs often have a difficult time adjusting to the tricky parts that barbershop demands. Singing in a big mixed choir feels like being in an orchestra, with the timpani and double basses in the back and the flutes and violins in the front. But barbershop feels more like a guitar, one person playing chords made up of strings tuned pretty close together, in timbre and pitch.

I loved singing barbershop. I loved feeling like part of an instrument, feeling a chord come into being. I rarely sang the lead melody—instead, I was often tasked with intricate close harmony, just off the main line. I was a supporting voice, completely necessary but not the first thing you heard. Barbershop’s like that—it’s a truly collaborative form. Take out any of the parts, and the song doesn’t sound right.

I never performed in public with SPEBSQSA. That wasn’t the point. The point was those Tuesday nights. I attended them religiously. They were like AA meetings for me, except instead of sad stories and burnt coffee, there was joyful, old-timey music. Those nights kept me sane.

They almost drove my family nuts, though. To practice, I sang along to barbershop tapes in the car—special tapes that isolated my part. On the A-side would be just your part, and the B-side would be all the others, with your part dropped out. The idea was, you could play the A-side ten times in a row to really nail down your harmony part, and then flip the tape to practice singing it with the rest of the ensemble. This is a useful technology, but it’s exceedingly annoying for any passenger not enamored of barbershop quartets. Like my son.

“Stop singing!” Logan used to yell, strapped into his car seat, with his hands over his ears. “No more singing!”

I’d stop. But when I was alone, driving to and from work, I really let it rip.

It strikes me now that those morning singing sessions were useful preparation for the work I was doing in the office in late 1998 and early 1999. Every day, I was redefining my role. I wasn’t always the lead voice anymore. I wasn’t always in front of the ensemble. But I was part of a group, and we were making a big, beautiful noise together. I was learning how to sing a tight, close harmony with Reed.

Officially that spring, my title was “president.” Day-to-day, little about my job had changed. I was still in charge of the aspects of Netflix that I loved (and was good at): customer relations, marketing, PR, web design, all the movie content, and our ongoing relationships with DVD player manufacturers. Reed took over the back end: finances, operations, and engineering. As far as I was concerned, the job titles were irrelevant. But titles mattered to VCs, and I wasn’t stupid: I knew that when it came to raising money for a rapidly growing (and still not profitable) startup, having Reed as CEO was one of our best assets. Reed’s presence calmed the board and reassured potential investors. That spring, I was more than happy to take a backseat pitching our company. I did what I did best: helping Reed soften his edges with both investors and employees.

Another person tasked with smoothing things over where Reed was involved was Patty McCord. Reed had brought her into the fold to run HR soon after we announced that we’d be running the company together. She’d been director of HR at Pure Atria, and she’d long been Reed’s right-hand point person. She was a sort of Reed-whisperer. She understood him like few people did and, more importantly, knew how to nudge him into social niceties. Reed can be…blunt. Patty can be, too. But she’s blunt in a charming, Texan way—she understands social graces. Patty knew that Reed didn’t always notice when he’d ruffled someone’s feathers, and that he was often oblivious to other people’s hurt feelings—especially people who didn’t know him well, like I did. If there was a contentious meeting, Patty knew how and when to take Reed aside and gently suggest that maybe he should apologize for calling somebody’s idea “totally unsupported by reason.”

Once, I overheard her telling Reed that our executive meeting had been very productive, before asking him who’d done most of the talking.

“Marc and I did,” he said.

“Do you think other people in a meeting should talk, too?”

Reed stared at her for a second, and I wondered if he’d answer.

He nodded. “Point taken.”

Patty’s role went far beyond merely Reed, however. It’s hard to overstate how enormous an impact she made on Netflix, as director of HR. Frankly, it’s hard to overstate the impact she made on the entire field of human resources. She completely redefined it.

I’ve written in this book about how Netflix’s culture, at least originally, wasn’t the result of careful planning—of aspirational principles or cultural manifestos. How it was a reflection of the shared values and behaviors of the founders. How we trusted each other, worked hard, and had zero patience for traditional corporate bullshit.

All of that is true. But what happens when the team grows?

When a company is small, trust and efficiency go hand in hand. If you’ve got the right people on your team, you don’t need to tell them exactly how you want them to do things—in fact, you often don’t even need to tell them what you want them to do. You simply need to be clear about what you want to accomplish and why it’s important. If you hired the right people—smart, capable, trustworthy—they’ll figure out what needs to be done, and they’ll go ahead and do it. They’ll solve problems on their own before you even know the problems exist.

And if you didn’t hire the right person? It’ll be apparent, really quickly.

Our early culture at Netflix was born completely out of how Reed and I treated each other. We didn’t give each other a list of tasks we expected the other to be doing and then “check in” frequently to make sure everything got done. We just made sure that each of us understood the company’s objectives, and which aspects each of us were responsible for. It was up to us to figure out what needed to be done to accomplish those objectives. And it was up to us to be honest with each other—radically honest.

I’ve written about what that looked like—or, more accurately, what it sounded like. Raised voices, argumentative meetings, blunt statements about how an idea was stupid or wouldn’t work. Sometimes it was hard for people to understand that Reed and I really liked each other—that we’d found that we were most productive when we dropped all the bullshit and just said what we meant. Reed and I had been doing this since those early days in the car on 17, and we’d never stopped. Whether it was just the two of us or a twenty-person departmental meeting, we felt that we owed it to the company (and each other) to make sure that we’d teased out the proper solution—or, more accurately, beat it out of each other with 4-irons and billy clubs. Sometimes the discussions would become so boisterous that Reed and I would lose ourselves totally in academic gymnastics, finally stopping only when we realized that one or the other of our ideas—or, more usually, some combination of the two—was obviously the solution, and that it was time to move on. It was not unusual for Reed and I to look up after a particularly loud session to see the quiet, stunned faces of our colleagues around the table, with expressions that seemed to ask, “Why are Mommy and Daddy fighting?”

But they got used to it.

Radical honesty. Freedom and Responsibility. These are phenomenal ideals, but for our first couple of years, they weren’t really written down. We approached things on an ad hoc basis.

Here’s an example.

At some point in 1999, one of our engineering managers came to me with a peculiar request. His girlfriend had moved to San Diego and he was trying to keep their relationship on stable footing. “How would you feel,” he asked, “if I left work early on Fridays to fly down to San Diego?”

He explained that he would work from there on Monday, fly back home Monday night, and be in the office Tuesday morning.

My answer probably surprised him. “I don’t care where you work, or what hours you work. Work from Mars, for all I care. If all you’re asking me is about when you work and where you do it, that’s an easy answer: it makes no difference to me.

“But,” I continued, “if what you’re really asking me is whether I’m willing to lower my expectations for you and your group so that you can spend time with your girlfriend? Well, that’s an easy answer, too. No.”

He looked at me uncertainly. I could see his dreams of San Diego weekends fading away.

“Look, where and when you work is entirely up to you. If you can run your group effectively on three and a half days a week in the office, all power to you. Go ahead—I’m envious. Wish I was smart enough to do that. Just remember: You’re a manager. Part of your job is making sure that your team knows what you want them to accomplish and why it’s important. Do you think you can do that without being around?”

Needless to say, his girlfriend was a free agent shortly thereafter.

I gave that engineer freedom to make a choice, but also reminded him of his responsibility to the team. I was radically honest with him—I doubted that he’d be able to keep up his end of the deal if he took off early for San Diego every week—but ultimately I left it up to him.

The manager felt empowered, free to make a choice about his own lifestyle, and the company ended up benefiting from his renewed focus. Everyone won.

Well, almost everyone. The girlfriend in San Diego probably didn’t see it the way I did.

Freedom and Responsibility weren’t just for managers, either. Take our receptionist, for example. When he started the job, there wasn’t a seven-page list of all the things he could do or not do all day—Keep the desk clean. Don’t eat at the desk. Instead, his job description was a single sentence: Put the best face forward for the company.

We gave our receptionist a clear responsibility and near-total freedom to figure out how to accomplish it. It was entirely up to him what hours during the day someone needed to be there; up to him to figure out how to cover when he was away, or sick, or needed a day off. It was up to him to figure out which behaviors didn’t put the best face forward for the company (like eating lunch at his desk), and which helped. (I have a strong suspicion that he bought the popcorn machine.)

And you know what? We had a damn good receptionist as a result.

A culture of freedom and responsibility, coupled with radical honesty, worked like a charm. Not only did we get great results, but employees loved it. People who have the judgment to make decisions responsibly love having the freedom to do so.

They love being trusted.

But that just makes sense, right? If you fill your company with people who lack good judgment, then you have to build all kinds of guardrails to keep them in line. You have to define everything for them: how much they can spend on office supplies, how many vacation days they take, when they are expected to be at their desk.

Most companies end up building a system to protect themselves from people who lack judgment. And that only ends up frustrating the people who have it. Remember the engineers in the hot tub? If you treat people like children, it doesn’t matter how many beanbag chairs and beer parties you throw at them. They will resent you.

In 2000, we were growing fast. And we were still hiring people with good judgment. But even people with good judgment had questions about culture and rules—and they shouldn’t always have to find me or Reed to ask them.

We started to ask ourselves: What if you could build a process that was meant for people who had great judgment? What if you could free them from all the petty restrictions that drive the top performers crazy? How can we scale up this set of ideals that came so naturally to us, so that a growing company can benefit from them?

How do you codify culture?

That’s where Patty McCord came in. She was brilliant at pushing the boundary of rules and freedom. She identified that what was special at Netflix was our particular combination of freedom and responsibility. And then she endeavored to put structure in place not to limit freedom—but to encourage and preserve it.

How far can you take freedom? How do you ensure shared responsibility?

Patty came down on the side of common sense. For example: If you were traveling for work, common sense said there had to be some mechanism for expense reimbursement. But none of us wanted lengthy, time-consuming, and ultimately pointless approval processes for it. If we were trusting them to make decisions on the company’s behalf that could make or lose millions of dollars, we could certainly trust them to make decisions about what type of plane tickets they should book for themselves.

Same with vacation days. We hadn’t kept track of them before because we didn’t need to. The attitude was: If you need to take a day off, just take it. I don’t need to know about your root canal, or your kid’s school schedule. Just get your work done, and cover for yourself when you’re gone.

But when a company has fifty employees, things get more complicated. People want to know what they can and can’t do. Patty could have just replicated the standards of the time: fourteen days per year paid time off. Instead, she was curious. If we wanted our employees to take time off when they needed it, why couldn’t we also let them decide how much time to take and when to take it? What if we didn’t have a set number of vacation days? What if we just trusted our employees to get things done?

Unlimited vacation days and hassle-free expense reimbursement are almost clichés now. But they were groundbreaking at the time. In Netflix, Patty saw an opportunity to redefine the role of HR departments. No longer was HR just a lonely cubicle filled with employee handbooks, sexual harassment claims, and benefits summaries. Instead, she envisioned the department as a proactive agent for culture.

She saw an opening, and she drove a truck through it. She dismantled all the systems we had in place that limited the amount of freedom we granted our employees, and designed systems that were almost totally on the side of employee freedom. She fought hard to ensure that we didn’t inadvertently create new paradigms that collared our workers, while also implementing structures that made it clear what we expected from them. Part of why she was so successful is because she held everyone, including senior leadership, accountable. It didn’t matter who you were—Patty would call you on your bullshit. She was never afraid to speak truth to power.

She knew how to do something rare: scale up culture.

A good example: Remember new employee dress-up days? I’d long assumed that they would fall by the wayside as we got bigger. Asking someone to make a costume and endure a fake interview is easy when you’re only hiring one person a week. But once we had five, six, or a dozen new hires every week, it wasn’t all that practical.

But Patty saw the value in a quirky, film-oriented ritual. So she made it easier and more efficient: she filled an entire room in the new office with dozens and dozens of film-related costumes—Batman outfits, Wonder Woman capes, cowboy hats and fake six-shooters fit for a Western. New hires still had to dress up, but everybody had the same pool of outfits to wear. The pressure was off. It was just fun.

Patty also smoothed over some of our rougher edges. She tried to, anyway. For example, one of the only pieces of decoration in my office was an Austin Powers: International Man of Mystery promotional poster that a movie studio had sent me. It depicted the entire text of Dr. Evil’s soliloquy to his therapist, halfway through the film, in which he recounts in baroque language his bizarre childhood, describes in detail his comically insane father, and rhapsodizes about the “breathtaking” feeling of “a shorn scrotum.”

I knew it wasn’t the most HR-friendly wall hanging. But I couldn’t help it: I loved that movie. And that scene made me snort with laughter. It became an inside joke between Patty and me: Every time she popped her head into my office and saw my poster, she’d stifle a laugh and order me to take it down. And I would—at least until her back was turned. Then I’d put it right back up.

Just because Netflix was growing and had an honest-to-God HR person didn’t mean that we couldn’t have a little fun in the offices. Case in point: a little game we called Coins in the Fountain.

I don’t remember who came up with it. All I remember is that the male employees of Netflix played it constantly. The rules were simple: You put a coin in the bottom of the urinal. The next person to use the facilities would see it and either ignore it, or reach into the bowl and take it. It was a sort of sociological experiment: How much money would it take for someone to do something disgusting and unsanitary, and reach into the bowl?

The game only worked, of course, if not everyone knew they were playing. But whoever seeded the urinal would usually tip me off. We learned a lot of interesting things about human nature, playing that game. For instance: a quarter would disappear much more quickly than three dimes. No one would touch paper money, unless the denomination was over five dollars. The highest cash value ever achieved was when someone threw in a twenty-dollar bill. It languished in the urinal all day, and was still there when I left at six for dinner with my family. But when I came back to the office later, at eight or nine, it was gone.

I still have my suspicions.

Another game we liked to play involved the kitchen. It was a typical mid-90s setup, the sort of thing you might recognize from Dilbert cartoons or The Office: refrigerator stocked with forgotten Tupperware, microwave stained by dozens of exploding popcorn bags. This was years before nitro-cold-brew taps dominated the kitchen spaces of American startups. We were decidedly more old-school. No on-site chef for us. Most of us brought our lunches to work.

This game was also a kind of willpower test, but in reverse—and with snacks. It arose from a common problem with shared kitchens and communal food. Anytime anyone brought a snack to share among the whole office—a dozen donuts from the shop down the street, or a bowl full of leftover Halloween candy—it would vanish within minutes. That’s what long hours and stress will do to you. Crumpled fun-size Milky Way wrappers and powdered sugar littered the tabletops mere seconds after a batch of snacks had been dropped off.

Eventually, we made a game out of it. Was it possible to bring a snack that would last more than a few minutes in the kitchen? Could you bring something that would get eaten…but only over the course of an entire day?

The challenge wasn’t to bring a food item so disgusting that no one would touch it. That would have been too easy. Just bring rocks. The point was to bring something just strange or unpalatable enough that it would eventually disappear, but would take an entire workday to do so. You had to walk a tightrope between delicious and unsavory, familiar and strange.

Here’s an example:

One day, I brought in a huge bag of dried shrimp and seaweed from the Asian market in Sunnyvale. Delicious, if you like that kind of thing. But pungent, strange-looking, and most definitely not for everybody. I opened the bag and filled one of our popcorn bowls with the stuff, then camped out at a table with a good view of the kitchen. Within seconds, Boris had sidled up to the bowl and, his mind elsewhere, working on some coding problem in his head, reached in for a handful. The look on his face when he realized that he wasn’t eating popcorn or M&M’s was priceless.

I cackled inwardly. For the next three hours I watched as Te, Christina, and the rest of the office ambled up, tasted a bit of the sea, and left after one bite. The only person who showed absolutely no reaction to the shrimp was one of the engineers. He took a little bowl of them back to his desk and snacked away happily.

They lasted until five.

Another time, I brought in a dozen Balut. You might have heard of them? They’re a delicacy in Laos and Cambodia: fertilized duck eggs that have been incubated for seventeen days, then boiled. There are embryos inside, tiny little ducklings. For obvious reasons, they’re off-putting to most people. As a result of their preservation and curing, their yolks are a deep, dark green. The whites are dark brown. They look—and smell—like dinosaur eggs.

I neatly sliced a few, arrayed them on a paper plate, set out forks, and even made a sign: Duck Eggs! Try one!

Surprisingly, they only lasted two hours.

The new offices were at the northern edge of Los Gatos, and bordered Vasona Lake Park. They were big. Two stories, open floor-plan. This was no converted bank. It was a Silicon Valley office building, built to house a company. It was big enough to grow in. Whenever you hired somebody new, all you had to do was fit a few cubicle walls together.

I was on the south side, upstairs, with all the front-end web people, content producers, analytics, and marketing. Reed was on the other side of the building, hunkered down with the finance team and back-end developers. If we both stood up at the same time, we could see each other across the cavernous space.

We’d come to a détente on the issue of the shares. In the end, I’d agreed that a third of the shares Reed wanted, if he was to come in as CEO, would come from me. The other two-thirds he was going to have to ask the board for. Which he did—and which he got.

That spring, soon after the move, Reed had brought in two major hires who made a huge impact on the business. Barry McCarthy was the first one. A seasoned executive and former investment banker, he’d been working as the CFO of Music Choice, which piped music into homes through a set-top cable box. He had an MBA from Wharton and decades of experience as a consultant and investment banker. He was unlike anyone in the office—a hard-edged, East Coast preppy with a diploma from Williams College. In the shorts and sandals world of Los Gatos, his Brooks Brothers blazers stuck out like a sore thumb. Which is, I suspect, exactly why Reed liked him.

I liked him because he was smart, no-nonsense, and efficient. Also, he called me “Mr. Founder,” even after I told him to say “Marc.”

Barry’s arrival spelled the end of Jim Cook’s time at Netflix. Jim had always, from the start, wanted to be the CFO, and once Barry came in, it was clear that it would never happen. His departure wasn’t dramatic—these things rarely are. But it underscored what was happening, that spring and summer: the startup team was starting to peel off, and the next phase was replacing them.

This is one of the facts of startup life: change. When you’re building something from nothing, you rely on talented, passionate generalists: people who can do a little bit of everything, who buy into the mission, and whom you trust with your time, money, and ideas. But once you’ve gone from 0 to 1, and the seed you’ve planted is starting to grow, some shuffling happens. Often the person who was right for the job at the beginning is not right for the middle. Sometimes bringing in people with decades of experience and institutional know-how is the necessary thing to do.

That was definitely the case with Tom Dillon, whom we brought in as head of Operations after Jim left, in early 1999. Tom was semi-retired, in his mid-fifties, and had spent his entire life managing global distribution for massive companies, most recently as Chief Information Officer at Seagate and Candescent. These were huge businesses. Seagate in particular was massive and complex. They had twenty-four factories all over the world, staffed by more than 100,000 employees. It’s hard to fathom what it takes to be in charge of tech for a company of that size and scale. Even more amazing, Tom presided over a time when Seagate decided to automate all its factories—which allowed the company to cut the number of factories (and employees) in half.

I don’t know where Patty McCord found him, but I think Tom was one of the most important hires that Netflix ever made. I’m still kind of astounded that we got him. We were doing two thousand movies a week, all shipped within the United States, and this was a guy who had overseen companies doing millions of shipments, all over the world. Frankly, he was way beyond our pay grade—literally. We could only pay him about twenty percent of what he was probably used to.

But Tom Dillon’s a different kind of cat. He’s a total type B personality—surprising, for someone charged with the extreme detail-work his job required. A tall, somewhat shambling fellow, he had a big beard and a shock of receding white hair. He favored loose clothes and laid-back jokes. I never saw him stressed. He was like a real-life embodiment of Jeff Bridges as “The Dude” in The Big Lebowski—everyone’s favorite stoner grandpa. Tom Dillon abided.

He saw our little company as a kind of retirement hobby. I think he liked the challenge. I mean, we had one warehouse and were still sorting everything by hand on card tables. It was like hiring Miles Davis to play your kid’s bar mitzvah.

We had a new office, filled with new faces. But we had the same old problem: No one wanted to rent DVDs from us.

Sounds insane, doesn’t it? Within a year, Netflix would be almost synonymous with rental. But from ’98 to ’99, the only way we could convince people to rent DVDs from us was to do it for free. We were a year and a half in, and we’d tried everything we could think of: rent-one-get-one-free, give-a-ways, bundles, promotions. We’d tried every single possible home page design we could think of. But we were coming up short. We still hadn’t developed a way of obtaining customers—and then a way of getting them to come back—that made more money than it cost to acquire them in the first place.

Not exactly a stellar business plan.

But Amazon, as we’d always known they would, had started selling DVDs the previous November. And after a few months of directing our customers to Amazon when they wanted to buy, Reed had quietly shelved the initiative. We’d invested hundreds of hours adding Amazon links to our site, all designed to send our customers to Amazon to buy their DVDs. We fully expected them to make the same efforts pushing their customers back to us to rent—but the returns had been paltry. Amazon’s links back to us were lackluster and hard to find. We were sending them tens of thousands of customers—they were sending us hundreds.

When the deal collapsed, Reed informed us that it had never been that important. Everyone was crestfallen, especially Christina. From the beginning, she had opposed the deal, but as always, she’d been a complete team player—and had worked hard on the Netflix end. A lot of conversations with Reed (and with Patty) were needed to get him to realize how demoralizing it could be, when you made something a priority, asked people to break their backs doing something they didn’t agree with—and then didn’t honor the work they put in.

Also demoralizing was the fact that without new renters from Amazon, and now without DVD sales to keep us afloat, we were hemorrhaging money. Reed and I put on a good face in front of our teams, spinning our depressing results as a kind of blessing. If we were ever going to figure out a way to make Netflix work, we pointed out, the company had to focus on one thing. And that thing was rental.

By the summer of ’99, things had reached a breaking point. I was spending most of my lunch breaks jogging through the park next to our offices, hoping that at some point during my sweaty peregrinations on the Los Gatos Creek Trail, I’d land on a solution that would keep people renting from us.

There was one idea I couldn’t shake. On one of my last trips to our warehouse in San Jose, I noticed that we had thousands—no, tens of thousands—of discs just sitting unused and unwatched on the warehouse shelves. When I came back to the office and shared my observation with Reed, it sparked another interesting Reed and Marc conversation: Why were we storing all those DVDs in a warehouse? Maybe we could figure out a way to let our customers store the discs. At their houses. On their shelves. Just keep the DVDs as long as they wanted.

What if we did away with late fees?

The more we thought about this idea, the more we liked it. We knew that one of the biggest problems with our current rental program was that it relied on a somewhat organized, prepared renter. Someone who thought several days in advance about what they might want to watch.

In other words, pretty much nobody that any of us knew. Most people (and loath as I was to admit it, I included myself in this category) figured out what movie they wanted to watch right around the time they pulled into the parking spot in front of Blockbuster. And in my world, that would have been considered prepared. Most people made the decision what to rent about ten seconds after they spotted it on the new-release rack.

But if they could keep the disc as long as they wanted? That changed things. Now they could let that disc sit on top of the TV as long as needed. And when the mood struck that it was time to watch a movie, it would be instantaneous. Even faster than driving to a Blockbuster. And if you had a bunch of movies sitting there, you could choose what to watch based on your mood. Was The Thin Blue Line a little bit too intense after a hard day at the office? Fine, let it languish. Luckily Groundhog Day was happy to step into the breach and lift your mood.

In a single stroke, this would completely turn our biggest weaknesses into our greatest strength.

And when they were done with a disc? Here we weren’t sure what should happen. Well—what if users simply mailed the discs to the next renter, a peer-to-peer approach?

In other words, we were basically pulling things out of our asses. But by mid-summer, after weeks of debate and about a hundred miles of running, we’d come up with three ideas that we didn’t think were total trash. They were:

Our plan was to test each of these initiatives separately, to see what worked and what didn’t. We’d done this from the beginning, at Netflix. We’d designed our site so that the impact of even the slightest change could be measured and quantified. We’d learned, before launch, how to test efficiently. It didn’t matter, in the end, how great a test looked—there could be broken links, missing pictures, misspelled words, you name it. What mattered was the idea. If it was a bad idea, even more attention to detail in our test wasn’t going to make it a good one. And if it was a good idea, people would immediately fight to take advantage of it, despite obstacles or sloppiness on our end. Faced with a problem on our website, they would try again and again to make it work. They would reboot the site. Try to find a way to work around the problem. Call us up to place their orders (and we had an unlisted number!).

If people want what you have, they will break down your door, leap over broken links, and beg you for more. If they don’t want what you’ve got, changing the color palette won’t make a damn bit of difference.

So, by mid-1999, we were old pros at testing. We could do it quickly. But even quickly, each test would take about two weeks. When I ran this by Reed, he looked at me like I was crazy.

“That makes no sense,” he said. “We don’t have the time for that.”

“Listen,” I told him. “We’ve got to do something. We can’t retain anyone, and no one’s renting, and—”

“Exactly. That’s why you should just test everything at once,” Reed said, cutting me off.

I started to argue, but then I remembered the previous year of tests. It wasn’t a bad idea. And it fit in with our ethos of faster, more frequent testing. We were always trying to avoid one of the number one pitfalls of startup entrepreneurship: building imaginary castles in your mind, meticulously designed, complete with turrets, drawbridges, moats. Overplanning and overdesigning is often just overthinking—or just plain old procrastination. When it comes to ideas, it’s more efficient to test ten bad ones than spend days trying to come up with something perfect.

What the hell, I thought. I told Christina and Eric to combine all three tests into one offer. At this point, we had a pretty good flow of customers coming to our site to redeem their Free-Rental coupons, so it wouldn’t take long to get some results. We set it up so that every tenth customer clicking the redeem button on the site would be directed to a custom page, offering them the opportunity to try—for free—a monthly subscription to the Netflix Marquee program: no due dates, no late fees. We would send them four DVDS, and when they sent one DVD back, we would send them another one. As many times as they wanted. And at the end of the month—if they didn’t cancel—we would automatically (and here it was me describing it as “automagically”) hit them up for only $15.99 a month, payable by any major credit card.

Home Rental Library × Serialized Delivery × Subscription. Just the last three halfway decent ideas we’d had, thrown into one pot.

“This probably won’t work,” I told Christina. “But, hey, at least we’ll know.”