People didn’t just like no late fees, a flat monthly rate, plus serialized delivery with a queue.
They loved it.
The first day of the test, 90 percent of the people who clicked on the banner ad gave us their credit card information. That’s insane. I’d expected something closer to 20 percent—that’s usually what happens when you ask someone for sixteen digits, even if they’re getting a month free and can cancel before any charges. And it wasn’t a fluke—day after day, the sign-up rates were that high. Visitors to the website were signing up for subscriptions at four or five times the rate they had chosen our à la carte rentals.
If people saw this new offer, they were taking the bait. Hook, line, and sinker.
We scrambled to build a service that could actually do what we were promising. There were a lot of things to figure out—how to run a rolling serialized delivery system alongside our normal operations, how to do automatic subscription billing, how to build a serviceable queue system. But within a week, the results were so positive that we knew we had a winner.
Several times a day, I’d pop over to Suresh’s desk. Suresh extracted all the important data from the river of information we created each day and put it into a form that we could all digest and play with. I’m sure he grew to dread my caffeinated approach to his desk, my jittery requests for numbers. But I wanted to know—Were there more than yesterday? Fewer? How many were signing up for the plan? How many saw it and ignored it? Where in the process were they dropping off?
We’d know for sure after a month, when people who had signed up for their free trial (and given us their credit card information) could cancel their subscriptions. But things were looking up. It had taken hundreds of failed experiments, thousands of hours of work, and many millions of dollars, but it appeared that we had finally come up with a workable model for DVD rental by mail.
No one was more surprised than me. Not only had I fought against taking the risk of testing all three of our ideas simultaneously, but this was perhaps the least likely solution I could ever have imagined. If you had asked me on launch day to describe what Netflix would eventually look like, I never would have come up with a monthly subscription service. Even if you had tried to make it easy for me and put it in the form of a three-part multiple-choice question, I still would have had a one-in-three shot at picking the right answer.
A few days after we launched the test, Lorraine brought the kids over to Los Gatos for lunch. No more jogging for now. Instead, we ordered a pizza and had a picnic in the park. Afterward, Logan, Morgan, and I climbed up into the steam-powered train that ran along the park’s periphery. Lorraine slid into the row behind us, holding a grunting Hunter. As we circled the lake in the middle of the park and I talked through our exciting new idea with Lorraine, I thought of my father, setting up his steam-powered train in the basement and calling me down to watch the wheels turn.
“Guess I was wrong, huh,” Lorraine said when I told her about the initial numbers. “This thing is gonna take off, isn’t it?”
“I really think so,” I said. “But don’t feel bad. It wasn’t that good of an idea a few years ago. And besides, nobody knows anything.”
Lorraine laughed. She knew I was quoting William Goldman, whose book Adventures in the Screen Trade we’d both just finished. You might not have heard of Goldman. He writes screenplays, so he largely labors behind the scenes and out of the headlines. People of my generation can thank him for writing Butch Cassidy and the Sundance Kid. Those a bit younger may have enjoyed the script he wrote for The Princess Bride. He also did Misery, Heat, Magic, Marathon Man, The General’s Daughter…and over twenty-five more. He’s won two Academy Awards for screenwriting.
But William Goldman is most famous for writing three words:
Nobody. Knows. Anything.
According to Goldman, those three words are the key to understanding everything about Hollywood. Nobody really knows how well a movie is going to do…until after it’s already done it.
For instance, how is it possible that you can have a film directed by an Academy Award–winning director (Michael Cimino), starring a best-actor Academy Award winner (Christopher Walken), with a can’t-miss script and a $50 million budget…and end up with Heaven’s Gate, one of the biggest Hollywood flops of all time?
On the other hand, how can you have a film with a first-time director, a handful of amateur actors, no script at all, a budget under $50,000…and end up with The Blair Witch Project, which, after grossing more than $250 million, is one of the most successful independent films of all time?
There’s a simple explanation.
It’s because Nobody Knows Anything. And it’s not just in Hollywood. It’s true in Silicon Valley, too.
“Nobody Knows Anything” isn’t an indictment. It’s a reminder. An encouragement.
Because if Nobody Knows Anything—if it’s truly impossible to know in advance which ideas are the good ones and which aren’t, if it’s impossible to know who is going to succeed and who isn’t—then any idea could be the one to succeed. If Nobody Knows Anything, then you have to trust yourself. You have to test yourself. And you have to be willing to fail.
Silicon Valley brainstorming sessions often begin with someone saying, “There are no bad ideas.” I’ve always disagreed. There are bad ideas. But you don’t know an idea is bad until you’ve tried it.
And, as Netflix shows, sometimes bad ideas have a way of becoming good ones.
Not only had all the people who told me that Netflix would never work (including my wife) gotten it wrong, but so had I. We all had. We’d all known that the idea could work, but in the end nobody knew anything about how—until it did.
We had conceived of Netflix as an online version of Mitch’s Video Droid: a video store. Informally, we even called it that—we never called Netflix.com a website, or a rental service. It was still always The Store.
But now we had a new model, one we never could have brainstormed into existence. The most revolutionary structure in e-commerce was the result of years of work, thousands of hours of brainstorms, dire finances, and an impatient CEO. The subscription model saved Netflix and quickly came to define it. But it wasn’t something that we thought our way toward—it wasn’t something anyone could have predicted ahead of time. It took a lot of hard work, a lot of hard thought.
It also took a lot of cards falling just right.
Other people call that luck. I call it nobody knowing anything.
A subscription model had the potential to solve a lot of our problems. But it also provided a number of new ones.
The first was our existing promotions. I had finally convinced the DVD manufacturers to include our coupons in their boxes—after they’d said no countless times. I had promised them that we could be counted on to fulfill our promises. And now, as a result, there were hundreds of thousands of coupons promising “3 Free DVD Rentals!” in circulation. And due to the lags in the DVD manufacturers’ supply chains, they would be coming out of the woodwork for years. We knew that the best way to jump-start our subscription program would be to substitute every request for free DVD rentals with a free month of unlimited rentals fulfilled through Marquee. But would customers accept that? Or would they consider it a bait-and-switch? We were also worried about the manufacturers themselves, who would have every right to insist that we fulfill the terms of the coupons to the letter.
The second issue was our “first month free” promotion, in which we gave each customer a free month to evaluate the program for themselves before having to decide to continue paying for it. We liked the free month plan—it had brought thousands of new users to our service—but we couldn’t quite agree on what was going to happen after that first month. How could we convert users taking advantage of a promotion into paying customers? You could always just ask them if they wanted to continue. But I felt strongly that we should utilize a “negative option”—that is, not even ask. Instead, we would automatically roll customers into their next month of membership—and bill their credit card—unless they proactively canceled. You see this all the time now—Amazon Prime and virtually every subscription plan does it. But at the time it seemed like an overly aggressive money grab—verging on sleazy. Reed hated it.
The third issue was our à la carte DVD rental service. While it had never reached the point of being able to support the company on its own, quite a few people liked being able to rent DVDs one at a time, with no long-term commitment. But just as almost exactly twelve months earlier we had been confronted with the complexity of doing rentals and sales at the same time—and realized that our best chances of success were to focus on one—we now had to make a similar decision. Should we focus all our effort and resources on the program that might save us, or try to offer both models simultaneously?
The first problem was easier to solve than I expected. Turns out it’s easier to negotiate with an enormous consumer electronics company when you’ve been working with them for a year and your new initiative is proving to be a massive success with users. What was obvious to us was obvious to the Sonys and Toshibas of the world, too—a subscription model was a game changer. It’s hard to imagine that, now that tech startups offer subscriptions for everything from socks to sex toys. But in 1999, we were doing something that no one had done before: we were convincing people to pay for potential. By saying that they would be paying the same amount, no matter how many movies they watched, we were, in effect, daring them to use our service as much as possible. And by eliminating the penalties for keeping discs for days or weeks at a time, we were providing a viable alternative to the video stores for the heavy renters—traditionally the stores’ most valuable customers.
We were operating from a position of confidence, in other words. So when I approached Mike Fidler and Steve Nickerson at Sony and Toshiba, I didn’t ask if we could change the terms of our promotion. I explained the shift in our business model and gave them some numbers about the popularity of the program. Customers would still get their DVDs—and for free. But they’d be signing up for a subscription to do it. It took all my persuasive powers to craft a perfectly pitched argument, but in the end, it worked—nobody jumped ship.
The issue of the negative option was a little thornier.
“You can’t just charge people’s cards without asking them,” Reed said. “It’s totally unethical.”
“It’s totally the norm, Reed,” I told him. “Haven’t you ever subscribed to a magazine?”
“I don’t like it.”
“They have a chance to get something for free,” I said. “We have a chance to get them hooked. That’s the trade-off. They know it when they start.”
“Maybe they forgot.”
“Listen, if they liked the offer enough at the beginning to hand over credit card information, chances are they’ll like us enough to let us keep it.”
Reed frowned. He didn’t agree. But in the end, I won: After all, we were sending them a hundred dollars’ worth of DVDs. Customers had to input their credit cards to take advantage of the trial. That didn’t feel funny to them.
“Let’s just start from the assumption that everyone is going to like it,” I argued. “If that’s the case, they’ll be happy to have their subscription automatically continued and their credit cards automatically charged.”
Despite my optimism, I’m not completely crazy. Four weeks after we launched the free trial, I was half-prepared for a rash of cancelations. All day, I shuffled back and forth between my desk and Suresh’s, checking on the numbers. By five o’clock he was starting to shout them at me before I’d gotten to him. All day, the message was essentially the same.
“They’re doing it!” he said. “They’re letting us charge them!”
The thorniest problem by far was the issue of à la carte rental. Some renters loved it, especially low-volume renters who didn’t watch a ton of movies but liked the convenience of online ordering.
But a lot of renters loved the subscription service. In the first three months of its existence, Marquee drove up our site traffic by 300 percent.
The question we had to ask ourselves was: Was it worth trying to offer both models? Or did it make more sense to focus on subscription, jettisoning some of our earliest users?
To answer that question, I’d like to tell you about something I called the Canada Principle.
Netflix, for its first twelve years, limited its services to the United States. When we were just starting out, we didn’t have the infrastructure or the money to serve the international market. We had a couple of guys in a bank safe hand-stuffing envelopes, and our entire business model was based on U.S. postage rates. Nonetheless, we thought frequently about expanding into Canada. It was close, the regulations were easy, and the postage and transport costs were low. When we ran the numbers, we saw that we could probably get an instant revenue bump of about 10 percent.
But we didn’t do it.
Why? Two reasons.
First, we knew that it was inevitably going to be more complicated than it looked. Because French is the main language spoken in some parts of Canada, we would have translation headaches. Canadians use a different currency, which would have complicated our pricing—and the fact that Canada also calls that currency a “dollar” threatened to be a communications nightmare. Postage was different, too, so we would have had to use different envelopes. In other words, even something seemingly simple was bound to be a pain in the ass.
But the bigger reason for staying out was even simpler.
If we took the amount of effort, manpower, and mind-power Canadian expansion would require and applied it to other aspects of the business, we’d eventually get a far greater return than 10 percent. Expanding to Canada would have been a short-term move, with short-term benefits. It would have diluted our focus.
When Reed started advocating for dropping à la carte rental, I was initially against it. Even though the numbers were good, I was nervous about the financial hit of jettisoning that part of our customer base. Why couldn’t we do both for a little while longer, easing the transition for our users and our bottom line?
But once I realized that the decision was similar to the one we had faced six months earlier, when we’d decided to drop DVD sales—once I realized, in effect, that we were facing an opportunity to apply the Canada Principle—I was on board. Reed was right—if we knew that the subscription model was the future, there was no point in continuing to work on the à la carte past. À la carte users only made up a small percentage of our users. We were only diverting energy, money, and talent to a model we had outgrown. Plus, as with DVD sales, we were confusing customers, giving them too many options.
By February of 2000, we’d dropped à la carte rentals and switched entirely to the subscription service, which now cost $19.99 a month. Now Netflix was Marquee and Marquee was Netflix.
Focus. It’s an entrepreneur’s secret weapon. Again and again in the Netflix story—dropping DVD sales, dropping à la carte rentals, and eventually dropping many members of the original Netflix team—we had to be willing to abandon parts of the past in service of the future. Sometimes, focus this intense looks like ruthlessness—and it is, a little bit. But it’s more than that. It’s something akin to courage.
Moving the business entirely to Marquee almost immediately took one of our biggest liabilities—delivery time—and in one fell swoop turned it into one of our biggest advantages. Now we weren’t several days slower than going to a Blockbuster—we were many times faster! If you wanted to watch a movie, no more driving to the store. There was already a stack of movies waiting for you, right on top of your TV. It was as close to “movies on demand” as we could get.
We imagined a user with a rotating, constantly refreshed library of DVDs. Watch a movie at night, drop it into the mailbox on the way to work the next morning, and by afternoon get an email that the next DVD was on the way.
Not quite instant gratification, but close.
We didn’t know what this meant for our shipping methods. Tom Dillon had already restructured all the picking, packing, and shipping systems we’d been using for the previous year, to make them more efficient and user-friendly. He’d also figured out that it was much cheaper—and more efficient—to ship all DVDs separately and as they became available, even if a user had ordered more than one. (I’m reminded of this every time I go to a hip new small-plates place with Lorraine and they inform me that they’ll be bringing each dish out as soon as it’s been cooked. Sounds like a chill dining philosophy, but really it’s just easier for the kitchen that way.)
But although Marquee didn’t necessarily require faster shipping—since our customers already had movies they could watch sitting on top of their TV—we thought it would be pretty cool if subscribers could get a new movie the day after returning one. It would be like magic. And after all, who wanted to wait a week for a DVD?
Some of our local customers were already enjoying next-day delivery. By virtue of proximity, Netflix users in San Jose, where our warehouse was, tended to get their DVDs within a day of ordering them, while users in Florida often waited six or seven. But when we looked at the numbers, we didn’t see any correlation between delivery time and customer retention. After a few months, the customer retention levels in the Bay Area and in Florida were roughly equal.
“What gives?” I asked Reed one afternoon, bouncing a tennis ball off the side of a cubicle. “You’d think those people in Florida would say, ‘Screw it, this ain’t worth the fifteen dollars.’”
“They’re probably just used to it,” he said. “They know we’re across the country and probably just assume things will take longer for them. We might be dodging a bullet here. If we don’t have to build warehouses all over the country to facilitate overnight shipping, we’ll save a lot of money.”
“It just doesn’t make sense,” I said. “Next-day delivery should move the needle. There has to be something we’re not seeing.”
I tossed the ball slightly too hard, and it bounced past me to Reed’s desk.
“I have an idea,” I said. “We’ve never turned on next-day delivery for a city, offering it from the beginning. If we do that, we can measure the impact on all our variables, see if it matters.”
I’ll never forget the look in Tom Dillon’s eye when I told him we needed to test next-day delivery in another market, just to make sure. I wasn’t quite certain how to do it—obviously, we couldn’t just build an entirely new distribution center to test the service in one city. Right?
“Just do it in Sacramento,” he said, chuckling. “Don’t build a warehouse. Just drive everything up from here every night for a month, and drop it off at the Sacramento post office.”
“Are you volunteering?”
“Hell, no,” he said. “It was your idea.”
Which is how Dan Jepson found himself driving a panel van two hours up I-80, all the windows down, the edges of thousands of Netflix mailers fluttering gently behind him in the breeze.
For the next few months, Dan drove to Sacramento every morning to pick up the mail and bring it back to Los Gatos, and then a few hours later, he did the entire trip a second time to drop it back off in Sacramento again. For months, we measured the results. What we found was incredibly surprising. Next-day delivery didn’t really change our cancelation rates. Where it mattered was in new customer sign-ups.
“This makes no sense,” I said, standing next to Christina’s desk with a printout of new sign-ups in my hand. “We’re not telling them ahead of time that they’ll be getting their movies the next day—we’re just doing it! Do they just…intuit that they’ll be getting things quickly?”
Christina rolled her eyes. “Marc, no. You’re missing the forest for the trees.”
“They’re telling their friends. It’s word-of-mouth advertising.”
Christina was right. The longer we ran the test, the more apparent it was that next-day delivery was a real game changer—just not in the ways we thought. It didn’t affect retention—it affected sign-ups. Next-day delivery inspired real dedication, the kind that makes you tell all your friends about this new service you’re using. Over time, we noticed that our penetration into the Sacramento market was approaching Silicon Valley levels. Silicon Valley! Where all the early adopters of DVD technology lived!
The whole saga had provided a valuable lesson: trust your gut, but also test it. Before you do anything concrete, the data has to agree. We’d suspected that next-day delivery was important, but we’d been myopic in our analysis of our tests, so we hadn’t understood why. It took an additional test, with a truly outside-the-box execution, to understand what we’d already intuited to be true. And once we understood it, we could refine the idea and maximize its potential—which was huge. Next-day delivery was like magic. We knew it had to be part of our plans going forward. Now we just needed to figure out a way to make it work without driving the DVDs ourselves or building enormous warehouses all over the country.
“I’m on it,” Tom Dillon said.
Whenever anyone asks me what my favorite movie is, I never tell the truth.
The public answer—the convenient lie—is Pulp Fiction. All the cinephiles and tough guys in the audience nod their head with approval when I mention it. And it’s true—I love that movie. I love the writing, I love the cinematography, I love the performances by Samuel L. Jackson and John Travolta and Uma Thurman. I’ve probably seen Pulp Fiction more than any movie but The Wizard of Oz.
But it’s not my favorite movie. My real favorite movie is Doc Hollywood, a 1991 comedy you’ve probably forgotten about, if you saw it at all.
In Doc Hollywood, a young Michael J. Fox plays an arrogant plastic surgeon in Washington, D.C. Driving across the country in his Porsche, he gets into a wreck in small-town South Carolina. He’s mowed down some fences, and as community service, he’s ordered to work shifts at the local hospital.
Complications ensue. It’s essentially a fish-out-of-water story—he’s a big-city surgeon in a small town with small-town values, and he eventually realizes that being a small-town doctor is what he really loves.
Doc Hollywood is no one’s idea of a masterpiece. But it speaks to me—I’m not sure why. Maybe it’s simply because it taps into some deep desire to live an uncomplicated life with a real connection to people, family, and place. In many ways, Doc Hollywood is my fantasy. It makes me yearn for the simple life, for a place where everyone knows and cares for each other. Where you go to work, you come home, sit on the porch, and then get asked to judge a barbecue cook-off.
Doc Hollywood is not the first movie I’d name if you asked me for the greatest films of the twentieth century, or even the 1990s, or even 1991. But if I see my copy lying around the house, I’ll slide it into the DVD player more times than not. It’s not the best movie, or a classic, or a hot new release—it’s just my favorite.
Helping people find their favorite movies, movies they’d love, was our real goal at Netflix. From the beginning, we’d known that our company couldn’t be tied to a shipping service or a mere product—because if it was, we’d be obsolete the second the technology changed. If we wanted any chance of surviving long-term, we had to convince customers that we were giving them something better than an online library and quick shipping. Neither the technology nor the delivery method mattered. What counted was seamlessly connecting our users with movies we knew they’d love. That would be relevant regardless of what direction future technologies took us.
Easier said than done, of course.
One disadvantage of being an online store was that it made browsing difficult. If you knew what you were looking for, you could just search for it. But if not, finding movies was surprisingly difficult. You could only view one page at a time, and there was a limited number of movies you could fit on a page. You had to make a snap judgment based on the cover art or a synopsis. This was a problem in brick-and-mortars, too, of course. According to Mitch, most people walked into video stores completely unsure of what they were looking for, and simply drifted from section to section. But in a brick-and-mortar, you could ask a clerk for help. Or, at the very least, you could wander the aisles, and hope that you’d serendipitously stumble across something that looked promising.
We wanted to make browsing easier, and we also wanted to connect users with recommendations and reviews. So Christina, the editorial content team, and I designed content-rich landing pages for a variety of genres. If you were looking for a thriller, we had an entire page dedicated to them, replete with top-ten lists, reviews of recent and classic thrillers, and highlighted selections from our inventory. If you liked Tom Cruise movies—same deal. The idea was to provide gentle suggestions and guidance, something akin to what a sympathetic (and knowledgeable) video store clerk could offer.
We wanted to offer a personalized touch. The problem was, it was enormously expensive—not to mention time-consuming—to do it all manually. When we had 900 titles, it was somewhat feasible to create content to match. But by late 1999, we had almost 5,000 movies to work with. It was hard to keep up, and even harder to browse.
Reed, in typical Reed fashion, pushed for automation.
“Forget the landing pages,” he said. “We’re redesigning the site anyway. Instead of hard-coding pages, how about we just do it like this: Create a frame on the home page that has slots to display four movies at a time. Each slot can show the cover of the movie, run time, date of release, a little capsule synopsis—the data we already have. Then just make a list of fifty movies you might want to have appear there, and have the site randomly pick which four to display. Or better yet, just define how to build the list—maybe call the list ‘thrillers’ and let the system randomly pick from any movie we have that is tagged as a thriller.”
If I recall correctly, I reacted with horror to this suggestion. I hated it. It seemed cold, computerized, random—all the things we weren’t trying to be.
But have you used Netflix lately? Reed’s slot structure survives—with alterations. The most crucial of which is that the films in the slots aren’t randomly chosen. They’re the product of a complex algorithmic matching service, one that’s calibrated to both your taste and Netflix’s needs.
That algorithmic matching service can be traced directly back to 2000 and Reed’s slots. Because he was right, of course—users needed a more efficient, easier way to find movies they would like, something even more intuitive than an editorially curated landing page. Putting DVDs into slots was a start. Now we just needed to figure out some way to arrange them that wasn’t random.
In talks all that fall, we discussed ways to build a service that would give users movies they’d love while also making our life as a distributor easier (and more profitable). When users sat down to decide which movies to order next, we wanted them to see a list of films that had been customized to their taste—and optimized for our inventory. If we could show customers what they wanted to watch, they’d be happier with the service. And if we could also show them what we wanted them to watch? Win-win.
Put simply: Even if we were ordering twenty times more new releases than any Blockbuster (an enormously expensive gambit), we wouldn’t be able to satisfy all demand, all the time. And new releases were expensive. To keep our customers happy and our costs reasonable, we needed to direct users to less in-demand movies that we knew they’d like—and probably like even better than new releases.
For example: Say I rented (and loved) Pleasantville, one of the best movies of 1998 and a clever dark comedy about what happens when two teenagers from the nineties (Tobey Maguire and Reese Witherspoon) are sucked into a black-and-white television show set in 1950s small-town America. The ideal recommendation engine would be able to steer me away from more current new releases and toward other movies, like Pleasantville—movies like Doc Hollywood.
That was a tall order. The thing about taste is that it’s subjective. And the number of factors in play, when trying to establish similarities between films, is almost endless. Do you group films by actor, by director, by genre? Release year, award nominations, screenwriter? How does one quantify a thing like mood?
I worked with Reed and the engineers for months on a solution. The problem was coming up with an algorithm that actually spat out movies that made sense together. Since it could only use the data available to it—things like genre, actors, location, release year, language, and so forth—the algorithm often made suggestions that made sense to a computer but didn’t really take into account any kind of real-world similarity. Or, it would give unhelpful suggestions: “You like Top Gun? Here’s another movie that came out in 1986!”
In the end, we realized that the best way to give users what they wanted was to crowdsource data from them. At first, we did what Amazon did. Using a process called “collaborative filtering,” Amazon would suggest products to you based on common buying patterns. They still do this. Essentially, if you buy a wrench from Amazon, it groups you with other users who have bought a wrench, and then suggests that you buy other things that they’ve bought.
Here’s how it worked with rentals: Let’s say Reed and I each rented three movies from Netflix. I rented Armageddon, The Bridges of Madison County, and Casablanca. And Reed rented Armageddon, The Bridges of Madison County, and The Mighty Ducks. Collaborative filtering would say that since we’d both rented two of the same movies, we would probably each enjoy the third movie that the other person rented. Therefore, the site would recommend that I rent The Mighty Ducks and that Reed rent Casablanca.
The problem with this method, of course, is that filtering for rental history doesn’t really tell you whether I liked Casablanca, or if Reed liked The Mighty Ducks. It just tells us that we both rented those movies. We could have hated them. We could have rented them for our kids (or our wives).
If we were going to use collaborative filtering to group customers and recommend films, we needed to know what customers enjoyed rather than just what they rented. We needed a reviews system: a movie rating system. Grouping customers by ratings—by “clustering” users according to overlapping positive or negative reviews—meant that we could efficiently recommend films to users based not on what they’d rented but what they liked. Ultimately, the algorithm would become much more complex than that. But for it to work at all, we needed users to review movies—lots of them.
Ultimately, we decided that we would ask our customers to rate movies by assigning each movie from one to five stars. Five stars for a movie they loved. One star for a complete time waster.
It sounds simple enough, but that stupid star rating system was the source of hundreds of hours of argument. More battles about fewer pixels have never been waged. Could you give something zero stars? Should we offer a half-star option? When you gave a rating it was whole stars, but when we predicted a rating, should it be in whole stars or in tenths? When should a user be prompted to review a film? Where should the widget go?
In the end, we asked Netflix users to review films early and often. We would ask them to rate films whenever they visited the site, whenever they returned a movie, and whenever they rearranged their queue. The great thing about movie rentals is that you don’t have to rent a movie to have already seen it—unlike buying a wrench, a review didn’t have to be tied to a sale. Theoretically, a user could review every movie he or she had ever seen—even if he or she had never rented a single movie from us. And it turns out that people love to be asked for their opinion. Everyone’s a critic.
It was remarkably easy to amass enough reviews to build a collaborative filtering function that could actually predict—with reasonable accuracy—what someone might like. After that, Reed’s team went to work integrating these taste predictions into a broader algorithm that made movie recommendations after weighing a number of factors—keyword, number of copies, number of copies in stock, cost per disc.
The result—which launched in February of 2000 as Cinematch—was a seemingly more intuitive recommendation engine, one that outsourced qualitative assessment to users while also optimizing things on the back end. In many ways, it was the best of both worlds: an automated system that nonetheless felt human, like a video store clerk asking you what you’d seen lately and then recommending something he knew you’d like—and that he had in stock.
Actually, it felt better than human. It felt invisible.
If it sounds like two of the most innovative and influential developments in the history of Netflix happened quickly, hot on the heels of Reed and I deciding to run the company together—well, if it sounds that way, that’s because it’s true.
Reed and I came to our CEO/president agreement in September of 1998. Within a year, the subscription plan was live. Within a year and a half, it was the only way to rent from Netflix—and a redesigned site was connecting with customers using an innovative algorithm that gave them exactly what we knew they’d want…and what we wanted them to have.
Those two key innovations would be enough to prove to almost anyone that we’d made the right choice when it came to running the company. We were really singing together. The team I’d built was bursting with creative ideas to connect with our users, and Reed’s had a singular focus in streamlining our vision. Reed’s laser focus helped us concentrate on the future. My goal was to make sure that however quickly we moved, however efficient we got, we were always fundamentally seeking to connect with our users.
Past and future, heart and brain, Lennon and McCartney—Reed and I were a perfect pair.