2005 Entrepreneurship Conference - Taking on the Challenge: Jeffrey Bezos, Amazon

I want to talk a little bit about how we think about innovation at Amazon.com and, uh, give you a couple of examples from the world.

This is the whiffle ball and the guy, his name is David Nelson Malany, and in 1953 he took a Cody perfume package and, out of frustration because his son had broken a window with a regular baseball, created the whiffle ball. His son named it. Uh, this woman is named Betty Nesmith Graham, and you will, uh, know of her accomplishment in just a moment if you don't recognize her name.

But she was an executive assistant at the time when, uh, typewriters transitioned to film cartridges from ink ribbons. And, uh, she was very annoyed; she wasn't an espe, she was a great executive assistant, but she was a rather poor typist. She was annoyed by her inability to erase her mistakes with the film cartridges, so she invented Liquid Paper, um, which after, uh—uh, let's see—by then she did this in 1956, and in 1975, she sold Liquid Paper to the Gillette Corporation for $47.5 million. So she did pretty well, uh, with her innovation.

And by the way, it was just white paint, so, uh, there was no very, very clever person because there was no, uh, you know, special chemistry involved here. This was not a high-tech solution, um, but an extraordinary observation of a problem and a solution.

Now, so sometimes people see the problem, and the problem is really annoying them, and then they invent a solution. Sometimes you can work this from the backwards direction, and, in fact, in high-tech, I think a lot of the innovation sometimes comes from this direction. You see a new technology or there's something out there, some new understanding in the world, and you work backwards from a solution to find the appropriate problem.

Um, uh, carbon dating is a little bit like this. So you know they'd always wanted this solution to be able to date things, but until they really understood radioactive decay, uh, that wasn't possible to come up with that solution. So some new understanding, some new concept enters the world, and there are a lot of things like that in high-tech.

Another critical ingredient for anybody who would be innovative is persistence, and there are many examples of people being persistent in the world in order to, uh, make something work. Um, this is one of my favorites.

Um, WD-40 was, uh, this a small team, a small company of three people who got a government contract to, uh, develop a, u, some kind of coating that they could put on the skin of Atlas missiles while they sat in their silos to keep them from rusting. And they, uh, they worked on this for a long time to find this right compound that would do this job and finally did it.

Um, now it turned out that the Atlas missile market was a fairly small one, so the company wasn't really able to make much progress. Now, what does WD-40 stand for? It stands for water displacement, 40th attempt. Um, and that—that's the name right out of the lab book.

Um, when they did all this work to invent WD-40, uh, so when they, uh, properly recognized the size of the Atlas missile market, um, they started to think about a different business plan, and the company was called the Rocket Chemical Corporation, and it was their only chemical and it was their only product. And so about 10 years later, they renamed the company the WD-40 company, and you know the rest is history.

One of the most, uh, pernicious obstacles to invention is learned helplessness. So people get, uh, the problems that you encounter. Uh, if you encounter them for a long enough duration, uh, humans—and actually biology in general, but humans in particular—are so remarkably adaptive that we pretty soon see right through the problems.

Um, we don't even notice them. And so great inventors and people who—they're very good at, you know, ordinary things bother them. Um, you know, they wake up every morning and take a shower, and they think this shower is terrible. You know, and then—but if you—but it's very difficult to do that, to kind of push through this learned helplessness. You get used to something.

Here's an example. This woman is named Mary Anderson. Um, and Mary invented the windshield wiper. Let me get Mary's paper out here. Oh, well, I know the story well enough. Um, but Mary, uh, was—this was around 1913—and cars didn't have windshield wipers. In fact, when it rained, people would pull off to the side of the road, and they would take a rag and clean the windshield off, and they'd drive another mile and repeat this process, and they thought nothing of this.

And Mary, you know, looked at this and she said that it's ridiculous and that there must be a better solution than stopping every so often and cleaning off your windshield with a rag. And so she invented the windshield wiper, and she was roundly criticized that this was a ridiculous thing.

And people had all sorts of reasons why it wouldn't work, uh, primarily that it would be a huge distraction for drivers to have this thing going back and forth in front of them. And, uh, but she persisted. Uh, and this happened very quickly. Um, within 10 years of her invention of the windshield wiper, they were standard equipment on all cars.

So it was one of these things that, you know, it actually—even though people poo-pooed it—once they tried it, they were like, you know, stopping, not having to stop every mile with a rag is actually a pretty good idea. Um, and so Mary definitely got the last laugh.

And this is one of my favorite inventions and a fantastic example of learned helplessness. Do you know toilet paper was invented? Um, and people see the year—this is in the mid-1800s. Um, a guy named Joseph G.T. invented toilet paper in 1857. This is the first toilet paper. It's hard to read, but it says, "Many people have wooed their own destruction, physical and mental, by neglecting to pay attention to ordinary matters." Rarely, in an advertisement, does one find such a profound statement.

Um, and, uh, so, but this is one of those things—people didn't know they needed toilet paper until Joseph G.T. invented it, and then very shortly thereafter, they began to realize they didn’t know how they lived without it.

Uh, and it's certainly difficult to consider going—[laughter]—back now.

Um, let me talk about—let me make one very important disclaimer right up front that Amazon.com does not claim any inventions nearly as important as toilet paper. Um, this is—this is clear.

Uh, but we have been working hard on innovation and we've done a bunch of different stuff. And I want to—I want to, uh, and we have been for the whole nine and a half years of our history. And I want to talk a little bit about that.

So first of all, it's sometimes useful to take a little time travel trip and move backwards. This is what Amazon.com looked like in July of 1995. I think this is actually a screenshot from August, and, uh, it—you know, it's changed a lot since then.

Um, this is, by the way, I wrote all of this HTML myself. No, really, please hold your applause. There's no need; you're embarrassing me.

Um, and th—this—uh, notice some of the very obvious, uh, problems that this has. For example, there's no search box on this page. There—the search box is a click away. Um, you have to click on, um, I think you—you know, "one million titles" will take you there, or search Amazon.com's million-tile catalog.

So, uh, this is what the website looks like today. And, uh, we invest about $250 million in the last year, about $2 million a year for several years now—$250 million the last year on technology and content development—and so the website should be better.

Um, these are significant sums of money. The company has generated over the last three years $950 million in free cash flow in the context of, uh, investing $650 million in technology in that same period of time. So, we've gotten smarter about a few things—the search box is now—you don't have to click; it's not one click away; it's right there.

Um, and there are a bunch of other things going on here, but, uh, but the point is that you lose track in the moment of how much innovation there is over a nine-year period.

One of the things that is very important at Amazon that we're constantly trying to do is reject this, uh, either-or thinking that really is at the heart of one of the big impediments to innovation.

Uh, let me give you a simple example which is in customer service. Uh, what we want to do is make the service better so that we can have the ACSI rating that you're talking about of 88, which we are very, very proud of.

Um, and because that's a very—they have a panel of like 60,000 customers that they pull, so it's a very statistically significant kind of result and something that we do, uh, you know, it’s a metric that people have worked very hard—not specifically on that metric but on the things that lead to it. So it's a great, uh, for us—it’s something that we, you know, are very proud of this. We get a lot of customer contacts, and one of the things that we would like to do is minimize customer contacts in order to, uh, save money so we can offer people lower prices but still offer great customer service.

And so how do you solve that problem? Well, you do it a couple of ways. One is you eliminate defects and eliminate root causes of defects because if customers are contacting you, they probably didn't want to.

Um, I think actually the number four reason we get contacted is to say thank you, uh, and so that's kind of nice; we don't need to eliminate those contacts. But most of the contacts that we get are not to say thank you—they, you know, sometimes come in all capital letters. They, um, uh, you know, and, uh, so if you can eliminate the defects, you're going to save money because you're not going to have to handle the customer contact at all, and you're going to improve the customer experience. So there's no tradeoff in reducing defects.

Um, and the more—the closer you reduce defects to the root, uh, to the root cause, the more that's going to be true—you're going to both improve the experience and save money—no tradeoff.

But the second thing that we did is build, uh, customer self-service. And this is our—you know, this is the page where you go to do, uh, customer self-service, and we let people cancel their own orders. We let people, uh, u, consolidate orders and do all sorts of things and change their shipping address and so on and so on and check on the status and, uh, and when is my thing going to arrive, etc., etc.

And that has really empowered customers. And what you find is when—in other industries like the hotel industry, Embassy Suites often wins the best customer service, uh, award in hotels when they poll heavy hotel users.

And of course, the—the secret to Embassy Suites winning that is that they don't actually have any service; it's all self-service. And so people, when they're serving themselves, think the service is terrific. Um, and, uh, so you have the buffet out, and if the buffet is out and the food is free, so you just take the food and you leave, there's just not much that can go wrong.

Um, and that—that really helps. So that's something that, um, that we've, uh, worked hard on, and it's an example of rejecting either-or thinking. And we have been able to reduce contacts over the last seven or eight years by 85%, which is a huge reduction, and that, uh, that level of that driving down contacts per unit continues.

The other thing you have to do if you're going to, uh, innovate and that we try to do at Amazon.com is maximize the rate of experimentation. So—and if you're going to do that, you have to make sure that your cost of doing experiments is low.

Uh, if your cost of doing experiments is high, then you're only going to get to do that—the company is only going to do a few experiments per year. And if you're doing a few experiments per year, the—you’re going to have to do some kind of global prioritization to see which kind of experiments you can do, and that's going to have a couple of downsides.

One is you’re not going to do much experimentation, but the second is that the best, most inventive people in your organization are going to become frustrated because they're going to have to—get—ultimately, if we—if Amazon could only do three experiments a year, then if you wanted to do an experiment, you'd have to get my permission to do it, and I'm not scalable.

So that is going to put a huge impediment in the way of invention. And if you want, on the other hand, if you want to make experimentation that everybody can do it and it doesn't take, you know, sort of—you know, the institutional apparatus doesn't have to approve every experiment, then you need to make the cost of experimentation low.

If you're still going to make it a reasonable thing to do and, uh, and not irresponsible, you can't just let—if the cost of experimentation is high, you can't just say, "Yeah, this is going to cost $100 million to do this experiment, but go for it."

Um, so getting that cost low is key, and one of the things that we've done over the years is put a lot of effort into making sure that our infrastructure and the framework that operates our website and other parts of our business really make it easy for people to do experiments in a self-service way without coordination with the institutional apparatus.

Uh, one of the great things about operating on the web is that you have, uh, you have so much of the ability to collect data.

Um, and all many things that in the physical world—because the cost of experiment would be high—you have to use judgment to decide those things. You have to use intuition to judge those things because you can't do the experiment at realistic cost. So you end up having to argue over it.

Um, and on the web, you can often just know the answer. So, you know, a simple example: this was a control, this page, and you'll see the difference here. A new line is going to pop up: "Want it delivered tomorrow, May 5th? Order in the next two hours and 53 minutes and choose one-day shipping at checkout."

So this is an experiment we did about a year ago, and 50% of the customers saw it this way, and 50% of the customers saw it this way. And you can just literally measure the impact that this has on all sorts of important metrics, you know, first and foremost sales, so you can see just telling people that they should order in the next two hours and 53 minutes increase sales or decrease sales, um, and—and—and by how much and is this worth doing and is it worth pushing harder on, and so on and so on.

So being able to do that kind of experiment and not having it be a big deal is really wonderful. Here's another example, an even simpler example: this is—you know, customers who bought this also bought something that we're well known for.

And, uh, this has been improved over the years dramatically. Here's a small example: this was tested too. Al, it's not the one I'm showing here; is we added series to it so that you're—here's an actual individual DVD, but here are series from different things that was a successful experiment, so it stayed.

Um, one of the experiments that we ran with this "customers who bought this also bought" was it—it seemed very intuitive that if this works well then adding images might make it work even better, so we added images.

Turned out that it didn't make it work better; it made it work worse. Um, and even though that's a little counterintuitive, I mean, to me, I can see it both ways, but that's the—that's kind of the whole point here.

Um, you know, you—you have your a prior guesses, but then you actually validate them with real data. Um, here's another try at images—since the—this—the one that didn't work, people say maybe it's too clunky, maybe it's pushing things down below the fold, so this one—oops, this one was tried, but it also didn't work as well, so we stuck with this one.

Um, the other thing we try to do at Amazon.com is have a customer-centric obsession instead of a competitor-obsessed, uh, point of view. And there are, by the way, it's not clear in all companies at all times that this is the right strategy.

Uh, competitor focus—and, in fact, close-following strategies—can be very effective, uh, in certain business environments and there's certainly nothing wrong with that. But in our industry and, uh, given the DNA that we have inside the company, the kind of people that we've attracted over time are people who like to invent, people who like to pioneer.

So, you know, we’d have—we have the wrong culture to be close followers, so there's part of this that, you know, we are what we are, um, and we kind of have to stick with that and make the best of that.

I actually think in our space this is also very handy because the rate of change in the online world is so rapid that close following doesn't work as well as it might in a more stable industry that—that changes more slowly over time.

But one of the things about these fast change in industries is I often get asked, um, by audiences if I—Q&U, you know, what's going to change in five to ten years? One of the questions that I rarely get asked, and I think is at least as important of a question and maybe in some ways an even more important question, is what is not going to change over the next five to ten years.

And that's another reason why customer-centric strategies can be so powerful, because competitor-focused strategies have to change a lot; the competitive set changes so rapidly. Over the last ten years, our competitive set has changed, and the technologies have changed, so many things change but the basic drivers of our business, the core things that customers want, do not change.

Um, you know, we know that what customers want are selection, low prices, and convenience, and I guarantee you that ten years from now they’re still going to want selection, low prices, and convenience.

Uh, we are not going to wake up ten years from now and have customers say, "This is all well and good. I love this thing, Amazon.com, but could you make it a little less convenient?" Um, it's just not going to happen.

And so, uh, so when you work on low prices, you have to figure out inventive ways to do that. And there are inventions at different granularities; there are a lot of the most important inventions that happen with the finest granularities, so incremental improvements in unit productivity are critical.

So can you—in in our fulfillment centers, you know, which are about seven—we have about seven million square feet of fulfillment centers. Each, you know, each one is about 700,000 square feet; these are huge buildings, you know, kind of corner to corner. They're half a mile on a side, and so how do you, uh, pick—you can order any two items.

We have a million—more than a million items in stock—which, to put that in context, you know, a Barnes & Noble superstore has about 125,000 items, uh, a Costco or Sam's Club has about 4,000 items, a big box electronics store has about 7,000 items.

So over a million in-stock items is a lot of items, and you can order any two of those items and, uh, before 6:30 p.m. Eastern time, if you, if you can—if you can order them before 6:30 p.m. Eastern time, we will ship them that same night.

So—and those two items might be on opposite sides of this big fulfillment center, so we have—there's a tremendous amount of work and, um, very sophisticated computer algorithms that go into optimizing the pick paths so that we can pick those items and get them into the same box at low cost.

It's easy to do that if you're not worried about optimizing cost; it's very difficult to do that if you are concerned about optimizing cost.

So those kinds of incremental improvements, taking those pick path algorithms and making them modestly better and doing that every day—and, and the key to doing that kind of invention is to make sure that you have small, separate, empowered teams that aren't subject to a bunch of dependencies in the rest of the organization.

They know what they're trying to achieve and they go about making those incremental improvements, uh, day in, day out, month in, month out.

They have to be able to answer the question—which it's amazing— you'll find that this is very common in the business world that people often cannot answer the question, "How do you know whether you're getting better or not?"

Um, at this particular thing, and the picking, the granularity of where you ask that question is the trick. I mean, it's very easy, you know, on the distant outputs. If you look at, you know, something like free cash flow or revenues, uh, those kinds of things, people know whether they're getting better.

But those things are not directly actionable; you can't, you know, assign some team and say look go drive a free cash flow. Um, it's just too easy; uh, I mean it's too easy to command them to do it, but they won't be able to do it.

I mean, it's a useless kind of thing to ask for.

Um, and then on bigger granularities—so one of the things that we also observe is that used products, uh, can have much more value proposition for customers.

So we invite third-party sellers to compete against us on our Prime real estate, which is our detail pages—that’s an invention that also not only adds a lower price but selection too.

So you go through all these things—selection. Uh, one of the things that we’ve done recently to increase breadth of selection at Amazon is again invite these third-party sellers in. This works in all the categories where we operate, uh, and in some of the newer categories like apparel, it's a huge part of the strategy.

Um, convenience—the most recent thing that we've done to increase convenience is for our heavy customers, uh, which is something called Amazon Prime.

And Amazon Prime is a—we only launched this a week ago, and it is a $79 flat fee membership. You pay $79—a year—and you get two-day shipping for free.

Uh, and that is a, uh, the idea here is to take what might be an indulgence for people getting next-day air shipment or second-day air—next-day air—next-day air is just $3.99 an item once you're a member and take something they might perceive as an indulgence and transform it into a fixed cost, and then once people can view it as a fixed cost, well, why not have two-day shipping on my items?

Um, and so for a certain segment of customers, this is a very attractive proposition, and it's for—it's specifically for that segment of customers where convenience is most valued.

So, uh, you know, there's another segment of customers who are, uh, who are completely happy to wait, you know, eight business days for the products. They'll use our super-savers shipping offer, which is free; all they have to do is meet a $25 order hurdle.

So they're perfectly willing to save a few things in their shopping cart until they can meet that order hurdle. They sort of manage that; it creates a little bit of mental overhead for them, but they're—they're okay with that.

And so that becomes a great value proposition for them. And there's another set of customers who think about this completely differently—they hate the fact that there's this $25 order hurdle.

Um, they want to be able to buy, you know, one thing when they see it, get it done, have it come—they want it to come fast. They don't want to have to think about it; they don't want to have to try and optimize shipping costs.

And, in fact, one of the things that happens for that kind of customer too is they create a kind of, um, cognitive dissonance or guilt that gets set up in their head. You know, it—they kind of—they, since they can get the shipping for free, they feel like they should be getting the shipping for free.

Um, and so, uh, even though, you know, maybe for them, uh, that's not actually a rational economic consideration.

Um, this is, uh, another convenience-oriented feature that's been on the website for a couple of years now; it's instant order update, and this reminds you that you've already bought something.

And we get feedback from customers all the time; this is something that we measured, tested very, very carefully, and it reduces sales.

Um, there's just no question about it that in the short term it's statistically significant way reduces sales, and we so we overruled that with judgment and said even though we can't do long-duration longitudinal tests in time like—like a drug company would do to test for toxicity and that kind of thing where they do like a ten-year test, that those tests are just so expensive to do.

Um, we're g—gonna make a—we're g—gonna make an intuitive bet that having this thing that reminds you that you already bought something will—will actually—will be so beloved by customers that they'll—it'll be one of the reasons they like Amazon.com.

And so, uh, and, in fact, we do get tremendous anecdotal response back from customers saying they love this feature—little, you know, thank you notes saying I was about to buy the same music CD that I bought a year ago, and this is very real.

I don't know; you know, some people are more organized and have better memories than others. I definitely fall prey to this. In fact, just yesterday I was reading a blog that mentioned a, um, a Walt Disney DVD on the back; in the late ‘50s or early ‘60s, I can't remember where Walt Disney, um, was very interested in space, and he hired, uh, Wernher Von Braun, and did a whole series of television specials about the exploration of space starring Wernher Von Braun.

And these things have been collected on a DVD. And I read about this on this blog, and I thought, "My God, this is so cool; I have to buy this." And so I went to Amazon.com.

Actually, they were an associate, and so they had a link handy for me, and I clicked right through to the detail page, and there was my instant order update. I had bought it a year ago, um, and I've never watched it.

I bought it, and I must have, like, you know, put it somewhere. And so now I'm going to go search for it.

Um, um, this is a real one too—Drew, who's sitting down here, told me that he almost bought these, uh, earrings for his wife a second time. And is that a true story?

Um, and, uh, so I think in that case we—that was really a valuable service.

Um, another, uh, convenience-oriented feature is giving people high-quality product information right on the detail page. And one of the great things about this is an exam—this is a piece of the detail page for the Segway, and it explains in great detail how the Segway works.

And, um, one of the great things about the online model that we use is that a lot of, uh, our best customer experience is a fixed cost.

So if you have enough scale, if you have, you know, in—in the last twelve months, we have these, um, you know, 47 million customers who have purchased from Amazon.com, and if you have 47 million customers, you get to advertise that, uh, the cost of all this content across that very large customer base.

And so there's a sense in which you can have your cake and eat it too. If you look in the physical world, at physical world retail stores, uh, the kind of high-touch customer experience stores cannot have the lowest costs, and the reason for that is that those stores—the things that generate high-touch customer experience in the physical world are variable cost things.

So when you double your sales, you double those costs. Our model, when you look at this kind of content or any of the software features that we provide at instant order update, it’s—you know, somebody has to develop that and maintain it and make it work, but it would cost us the same to do that development if we had a million customers as it does if we have 47 million customers.

So that—that—that aspect of our model that we get to transform customer experience into a fixed cost instead of a variable cost is really a key. It's an underappreciated aspect of, uh, of what we do.

Uh, now, uh, this is also—for any company that wants to be innovative, absolutely critical thing to do, which is to not be distracted when people tell you that this is stupid and isn't going to work.

Um, and people are well-meaning, and entrepreneurs—and I’ve used that term the most broadly. Um, you know, people who are building, people who are inventing have to have this combination of stubbornness and flexibility.

Um, and the trick, of course, is knowing which to use when. Um, but, uh, but what you cannot do is be distracted by the outside world; you have to stay heads-down focused on customers and not let external events distract you.

And Amazon is a very good case of this. Um, this is, um, uh, December of 1998, but—and actually, this is a phrase that, um, and I think Rob is in the audience, but this is not you, Rob; you didn't write this.

Um, um, but this is—this is a case of, uh, 1997, um, uh, Forester research coined this term, and Barnes & Noble had just launched their, uh, online store, and I thought that Forester had a very good point.

Um, they were saying, you know, Amazon is this two-year-old company; they've had this nice two-year run, but you know—s—now the big, the big gorilla has come to town.

And at the time, we had annual revenues of $60 million a year. We had 125 employees, and Barnes & Noble had, you know, 30,000 employees and at that time, I don't know, $3 billion a year in sales, and so it didn't seem like an even fight.

And, uh, they coined this term "Amazon.toast," and it was widely picked up. And, you know, the—and it did bother the—the—there's one feedback loop here that has to be managed, which is you have to make sure that you have some All Hands meetings and that you communicate to employees what they—you know, should they be worried about this or shouldn't they be worried about this.

And in our case, I thought this was a very simple situation because the, um, I didn't think there was anything we could do about what our competitors were going to do, and so, you know, we did have the All Hands meeting, and I asked all of our 125 employees to be terrified and to wake up with their sheets drenched in sweat every morning.

And, uh, I did—I carefully specified sweat, um, and—and—but that they should be afraid not of our competitors, but they should be afraid of our customers because our customers are the only ones who are going to ever give us money.

Um, and so that helps; it actually helps have that kind of customer focus because it helps you stay heads-down. You don't get worried about your competitors. They're going to do what your competitors are going to do; you can watch; you can learn from them, but you can't, um, let them set your strategy.

You have to do what you're going to do, and you also can't let the media set your strategy, um, and—and—and—but you do see that happen; you can't let Wall Street set your strategy either.

This is a year after Amazon.toast; this is 19 May of 1999. Um, and, uh, this, um, I've always been amused by this, but I'm I'm comically misunderstood by this. I have to tell you my mom really hated this; this really bothered her.

Um, she—she was so pissed off about this, uh, this just didn’t like that caricature; it really bothered her. So, but, uh, anyway, you just have to ignore, uh, those outside, uh, influences. You can listen and kind of try to see if there’s any kernels of truth that you can take and adapt.

There's no reason to have a siege mentality, but that's—that's the kind of reactions you see; you either see that people enter this sort of siege mentality with respect to external factors or they're, uh, or they're too responsive to these external factors.

Um, and the reality is you have to take it with a grain of salt and, you know, reexamine your strategy. Make sure you really believe in it; it's the right one, and stick to it. That's the only way you can invent.

Um, because invention always leads you down paths that people are going to think are weird. It's sort of obvious. This is the—this is that ACSI score, and it's actually four years running that we've had this highest, uh, score.

So one of the things I mentioned is with the carbon dating and what happens in the, um, uh, online is you—in the—well, in tech in general is you get so much change, and you have to sometimes look at the change and see what you can invent.

So one of these things is Moore's Law, which is remarkable in its variance for bandwidth and—and in— in storage. But if you look at disc space costs as an example, disc space is 30 times cheaper today than it was five years ago.

And so, you know, that's just what happens when things get twice as cheap every year for five years, and humans aren't good at thinking in those kind of, you know, exponential terms. But this is really, um, a big deal, and if—if you were to picture a physical store, it creates that when things change that rapidly.

And actually creates a problem—you have to figure out what are we going to do with all that disc space because it’s not—you know, you can't just pocket the cost savings; you don’t get enough value out of that.

You have to say, well, if we can use 30 terabytes of disc, um, and that's a reasonable thing to—to spend money on now, um, what could you do that would actually benefit customers that use this 30 terabytes of disc?

Um, and the, uh, uh, one of the things that we did was search inside the book—which uses about 30 terabytes of disc because we keep the full images of all the pages—and we now have, uh, over 200,000, uh, different books that are completely searchable, and you can view the actual images of the books.

And, uh, and that, uh, uh, it kind of—the good news about that is it’s very beneficial to customers, and it soaks up a lot of this now almost free disc space that you can buy out in the world.

Um, another example is, uh, A9. Let me just go here, which just launched again about a week ago. The A9 and Yellow Pages—we do a search on Optical—oh, this regular mouse works too, and it's a lot easier to use.

Um, and one of the innovative things that the A9 team did for their Yellow Pages is take 20 million photos of over a million different businesses in ten different cities. Um, and they did that in a very cool way.

I mean, this—in some—you know, here, if you go down the list here, this is Optical in Seattle, uh, but, you know, the map changes; the little numbers highlight as you go over, which I think is a nice user interface touch.

But if I go down to Optical Illusions Inc., click on that, it takes you to a detail page for Optical Illusion Inc., and you can kind of stroll along the street here. So here is, um, you know, and you can zoom in on this picture, and actually, this is kind of a cool picture because this is—you can see in the reflection of this window that's the vehicle that we took the picture from.

And, uh, I also—I like the way this guy is in motion. It's just kind of, uh, and then you can't barely see her, but there's a woman behind him; you can just sort of see her—see her shoes.

Um, we’ve only been up a week and, but, you know, but if you take 20 million—20 million photographs of a million different businesses, we've already gotten lots of contacts from people, people, um, you know, who are—they found themselves in these pictures.

And, uh, if they request to be removed, we remove them, but—but this is a, uh—this is a pretty cool use of technology, and again, it's looking at what's changed in the world because one of the most interesting things about this is— is how they did it.

Um, you can keep going, by the way— I can click here, and it'll take me, you know, down the next street. I can just keep walking down this street.

Um, okay, so how did they do this? Should I go to the back to the PowerPoint?

All right, here’s the truck. There—there—there are more than one of these. Um, that is a, uh, a camera—a digital camera mounted on the top of the truck.

And when they first started doing it, their first tests were done with—not with SUVs, but with regular cars, and, um—the camera was not in any kind of enclosure, but they attracted too much attention.

And people would like smile for the camera, uh, or—or do more profane things for the camera. Um, so they solved that problem by getting a taller vehicle and, uh, you know, putting the camera in an enclosure so it just sort of looks like some luggage on the roof, and people don't notice it.

Um, and the—the—so let’s see here’s one of the—you guys, if you've been to Times Square, you may have seen this guy; he's out there all the time. I think he's called the Naked Cowboy.

Um, he's been doing this for years and years and years. He was there—he's been there at least for eight years—maybe longer; I don't know. Um, and I think TGI Fridays has hired him now, um, as a sort of mascot.

But anyway, um, they captured the, uh, Naked Cowboy in Times Square. Let me—watch this. I should actually—let me pause this for a sec.

Um, this—let me give you a little setup. So this is some—some of the actual footage from this camera. This is a—just to give you this is—a camera that is connected to a laptop that has a GPS in it.

Um, and so the Jeep—the car just drives around, and the driver doesn't have to do anything except make sure all the equipment is running properly, and it just takes, you know, a bunch of frames per second, and—and—and the car knows where it is at all times.

Um, so this—uh, now we—we—one of the cities that we did, we did over ten cities, but one of them was Washington, D.C., and, uh, and—and the driver, whose name is Josh in this case, had a fairly interesting experience because the camera broke, um, and needed some adjustments.

It just got jiggled or something and needed to be fixed and remounted, so—which happens occasionally—so he didn't think anything of it. He stopped the car; he got out, and he started to fix the camera.

But unfortunately for Josh, um, he happened to stop right in front of the State Department. And, um, and, uh, unfortunately, just to make matters a little more complicated, it was the day before the election, and so Washington was at its highest, uh, state of alert.

And, uh, fortunately this entire incident was captured, uh, on video and it's kind of fun to watch. So there's Josh; he's getting out of the car, gonna work on the, uh, camera a little, but he sees somebody’s coming; he's reaching for his cell phone now trying to call Amazon.com legal.

Now, now there are two of them; they're very curious why Josh is photographing the State Department. Here’s the third one; here’s the—here's the fourth one. And if you can catch glimpses of Josh's face, you can see that he is a little stressed.

By the way, too, at a certain point here, you can see the woman in the red hair, um, who I think is the one sort of in charge—she's trying to calm Josh down, and she says something to him like, "It's not like we're accusing you of having a bomb or anything," which didn't really work.

Um, oh yeah, here's the—the great—they take a picture of Josh and the—and they—they actually were, um, very, uh, friendly to Josh and he explained the whole thing, and they made some phone calls and confirmed his strange but true story, um, about what was going on.

And all—the whole incident only took like 45 minutes, and—and Josh was sent on his way, uh, to photograph the rest of Washington, D.C.

Um, let's go back. Um, okay, but all of those things—oh, okay—all those things are—if you look at what all the problems that had to be solved to do something like get that, you know, the GPS integrated with the digital camera and get the those in trucks and then solve all these little point problems that people were noticing the camera and so on—that's what innovating is all about.

Um, it really is those kinds of problems. And the great thing about this is, uh, at Amazon.com is I do very little of this.

Um, and so, you know, we have a culture where people in small teams can go off and do these relatively low-cost experiments and try to build, uh, neat things.

If you want to have an innovative company, the single most important thing, even ahead of minimizing the cost of experimentation, is to make sure that you're selecting the people correctly on the way in.

Um, and so you need to hire people who, uh, like to build, who like to invent, um, and you need to make sure that they like to do that at all granularities. You know, sometimes you come across these people who are only interested in inventing at the grandest sort of whiteboard level, um, and they actually can't make progress in the real world.

You know, they're unwilling to figure out how to mount the camera on top of the truck, um, and mounting the camera on top of the truck turns out to be incredibly important.

Um, this is, uh, I've mostly included this slide to demonstrate my true pedigree as a geek—this is me in fourth grade. Um, and, uh, almost all of you in this room are far too young to recognize that device that I'm sitting in front of.

But I was—I'm one of the only people I know who had access to my age who had access to a computer—computer—in fourth grade. I went—the school that I went to in—in Houston, um, got this—a company donated this teletype and also donated some excess, uh, mainframe, uh, time-sharing time that they had, and so we got this thing one day, and none of the teachers knew how to use it, and nobody knew how to use it, but had manuals.

And that—that phone receiver, it's called an acoustic modem, and you actually put the, uh—a 300 baud modem—and the—all the programs were stored on paper tape, which you punch holes in.

And, um, uh, we—me and about three other kids started staying after school and learned how to, uh, program this, uh, this mainframe computer and store our programs and our very primitive programs on paper tape.

And that went on for several months until we discovered, um, that the mainframe was pre-programmed to play Star Trek, um, and pretty much that’s all we did with the computer after that.

We would stay after school—the teachers were like, boy, these kids just love computers. Um, it was this really cool Star Trek game, um, where, you know, it's all done on this—this teletype on paper, and it would print a little, uh, I think it was a 9 by 9 grid, and somewhere in that 9 by 9 grid, like, you know, an asterisk would be a star and a plus sign would be a Klingon ship, and you had to do—it was a simulation game, and you had to like, you know, you had limited resources; like you had only so much power and you had to decide how to allocate that power between shields and engines and phasers.

Um, and it was actually unbelievably fun. Um, and, uh, so, uh—but you got to find people who are passionate about building.

This is our first, uh, employment ad at Amazon.com; we posted this on usenet, uh, and we asked for exactly what we wanted, uh, and, and you know, the technology—computer science people and—and business people who are interested in technology have always been most important.

So you must have experience designing and building large and complex, yet maintainable systems, and you should be able to do so in about one-third of the time that most competent people think possible.

Um, and by the way, uh, as a total aside, you'll see that the name of the company at the time was Kadabra; that was the first name of the company, and Kadabra was a terrible name, um, because while driving across the country as my wife driving, I called an attorney in Seattle and asked the attorney to—if he could incorporate the company so it would already be ready by the time we arrived.

And he said, "Sure, there's no problem." By the way, this attorney was, um, my friend's divorce attorney, so it wasn't—this wasn't like Wilson Sonsini or anything, you know, this was—and—and, um, so he said, "Sure, you know, what do you want the name of the company to be?"

And I—you know, I was all proud of myself for having anticipated that question, and I said, "Kadabra." And he said, "Kadav"—and I thought I knew right then that my name choice was not going to survive for long.

I was like, "No, no, no, no, Kadabra! Like Abracadabra!" Um, he was like, "Oh, okay." Um, so he, uh, about three months later, we changed the name to Amazon.com, and that was chosen because it started with the letter A, and at that time, all of the lists online were alphabetized, and it was also—the domain name was available, and it was—it worked internationally, and it was short and easy to spell.

So there were a whole bunch of good features of that.