we were pip squeaks at the time, even though we've gone public, we were a tiny company. You know, there are 100 big players that if they kind of thought about this the right way, just the power of staff and capital to step into that space. If they saw what the endgame was. This episode is sponsored by skill jar dot com. Founded by ex Amazonian, Sandi Lynn and Jason Stewart skill jar is the leading customer education platform. Innovative companies such as Tableau linked in and hundreds more rely on skill jar to onboard engage and retain their customers at scale to learn more, visit skill jar dot com. S K I l l j A r dot com. Hi, I'm dave chappelle and I'd like to welcome you to the invent Like an owner of podcast where I talk with the amazonians who helped build amazon dot com into one of the world's most valuable companies. This weekly podcast is for entrepreneurs, future business leaders and all students of history,

not to mention people interested in getting hired at amazon. The goal of the podcast is to capture the amazon creation stories and create a historical archive. On that note, my guests are recalling history as best they can. It's possible some of the details are fuzzy or just plain wrong. That happens. It's not intentional. I invite future guests or commenters on the website to help us get the facts as straight as they can be now on with the show. Today, I'm thrilled to be speaking with Joel Spiegel, who I was surprised discover only worked at amazon for four years from 1997 to 2000 and one because in my personal experience, Joel loomed very large. He made a huge impact on the organization, Joel exuded ownership and customer obsession and raised the bar for everyone. Today, we'll be talking in depth about goals,

impact on the organization, especially connected to the transition from a retail centric business model to a marketplace driven business model. Welcome Joel. Hey there. So yeah, when I was going through your linkedin, it's crazy. You were at again, I may get a little room but visit calico, visit corp, the parent company HP Apple Microsoft and then amazon it's quite a run. Yeah, it sort of surprises me. It was interesting. Who recruited you? Was it Jeff?

Where is it, you know, a typical loop? Like how did that all work? So it's a funny tying to that list of companies you just made because during my time in Silicon Valley, you know, there's a corp, Apple etcetera. Hp. We ran across a tech recruiter named Vicky helms and she was sort of legendary, one of the highest emotional intelligence people. I know she wasn't particularly a technologist, but she was incredibly networked. You didn't find her, she found you. And later on in my time in the valley,

I got a call from another recruiter named Susan Shea. And anyway I got to know Susan and fast forward a bunch of years when I was a Microsoft actually running a group doing sort of doing some search technology at the time, I got a call from Susan should I have a perfect job for you. And my response was, I'm happy at Microsoft. She said this job has your name written all over it, it's in your backyard and if you don't go talk to these folks will declare you stupid and never talk to you again. Not a typical right from a recruiter. And it turned out it was amazon and Jeff and Jeff and I went and had a breakfast and a very interesting conversation. So it was sort of like, you know, the connection chased back years and companies as to why I got the call. But you know, it felt right and Jeff convinced me there was the right place to be and there we were. I mean, at that point where they're even interview loops or was it just you talking to Jeff,

you know, and getting hired? No, I mean, I spoke with some of the other folks at the company as part of the process, but it was much smaller, less structured place to be sure, you know, the first person I spoke with interesting when he was only had breakfast on a saturday morning, he took me over to the offices in the Columbia building, which I'm always inclined to point out is quite different than the luxurious Colombia tower 15 16, 2nd half. Yes. And you know, the first person I met Was Sean Haines was running the associates program and I had worked with Sean Spouse Mary at Microsoft. She's in HR and Sean was was excited because you know,

give a sense of scale the associates program and just hit 6000 members. So it's quite, quite small relative to what's going on today. And when you were brought in, were you brought in for a specific problem? I mean, your title was VP Engineering right out of the gate or was it a different role when you joined? I think we settled on VP of Engineering essentially. I think Jeff had, you know, you'd be the better one to answer this, but he had been looking for somebody run technology and you come to the conclusion that there were two slices of technology that you couldn't find in a single person. So he needed somebody to handle sort of the transactional I. T. Infrastructure but also somebody who had sort of what I would call the Silicon Valley Apple user interface experience, you know,

which I had obviously picked up at places like Apple and there's corp I was brought in to address the the sort of scaling the user facing technology side right out of the gate. And I think there were a dozen or so technology people in the company at the time, half of them were sort of on what you might call the user facing technology side of the house. Although everybody was kind of a jack of all trades to some degree. And then we very quickly jumped into what matters to customers. What could we do, What should we do? It got down to Besides scaling which was a nightmare from day one. We had issues of inventing recommendations that were useful search technology that was relevant and did a good job, et cetera, et cetera. That the state of technology was essentially, Shell Captain had built the whole thing from scratch with the help of one or two other people. There were a few others who had joined prior to my arrival, but it was really for something that was serving a very large user base in the company is about to go public is very, very tiny engineering organization or set of organizations.

So scaling was a big deal. I ended up, you know, you may have heard me say it, that you know, really, I was just an over glorified recruiter because if you don't have the right teams, you know, it just doesn't, you can't do anything, you need, you need the people, the management and leadership structures that make it work. Who were some of those key first, I mean I talked to kim Rock miller a few days ago, she mentioned that you recruited her,

were there some of those other people that jump out at you as sort of the big contributors from early on that you brought in, you know, it's again, it all ties to sort of networks in a sense. So you know, Dwayne Bowman, Gene Pope, did you work with jean at Apple? Actually was hired into Apple by Gene Pope. I contacted Jean the first day of his retirement and I said, yeah, forget that retirement, you got to come to amazon. So was his first day of retirement from Apple? You recruited them right back into the madhouse, He had been out of Apple,

He has run, I think engineering for cork and retired. I want to call him at cork and the folks at court said, oh, he doesn't work here anymore and you know, I knew he obviously must have retired to the ranch in Montana and contacted him, said you have to come. But the other thing that's funny, you know, you mentioned, we got kim, is kim had been working with Erica lock at a startup and when I contacted Erica who had worked for me at Apple, you know, kim became part of the conversation as well, and Ruben Ortega came in to work with Dwayne because of his connections to Erica etcetera. So a lot of sort of very real life social networking was a key part of very rapid hiring processes.

What is it about when you spoke to Jeff that got you excited to join amazon, you know, you were at Microsoft is probably a good job, like was it Jeff, was it his charisma? Was it the scope of the problem or the impact you thought it was going to have on? You know, the world, like what is it that made you make another big jump? That's a long conversation in its own right. But we had breakfast and you know, the economics of selling on the internet made sense to me in a strange way, I had set up a personal little, you know, personal passion projects and when I started getting into managing technology teams, I want to learn more about business through a strange set of circumstances.

I ended up selling kayak paddles on the internet specifically because I wanted a very well bounded little business, I had an opportunity and at the time the internet was not the web, so wrecked boats. Whitewater was my vehicle for doing things, that it became clear to me, sort of what the economics of selling that way were. So it was incredibly fast conversation with Jeff. The first thing I thought is, wow, I'm really stupid. How did I not think of starting this company? Because I knew the core economics of doing those kinds of transactions via the Internet. So we skipped what were apparently a lot of the business explanations that Jeff went through with folks he was talking with along the way to explain why this made sense at a core business level. And immediately jumped into sort of the interesting scaling problems, people, problems in terms of hiring and recruiting and building teams,

things that led to some of the leadership principles, etcetera. So, so it was obviously Jeff, he was clearly super smart, he was clearly committed, but the problem space was fascinating and it was a green field opportunity, right? There weren't a lot of, there were no major e commerce companies out there. So it just sang as a problem space, you know, because like, so we're gonna talk about marketplace quite a bit because you made such a big impact there, but in your first month or months, do you remember some of the first projects that you were sort of tasked with helping to wrangle?

Obviously hiring was probably a couple hours of every single day, you know, interviewing. And do you remember any specific projects that were big sort of indoctrination to amazon recruiting was as you know, it a huge, huge part of it, probably more than ours measured in days of week, not hours a day, you know, a lot of that bore fruit, you know, we, fortunately after many cycles of conversations, etcetera, we brought in rick, Dalzell,

etcetera, but those were not fast or easy processes. But early on for my arrival, there was already some work on recommendations. We knew we had to take friction out of the buying pipeline. Those were big things, Jeff early on had the notion of wanting to do a B, testing my knowledge, he's the first person who thought about doing that. And honestly, the first version of that we built, I sort of misjudged something's, so we took another run at that and he was right about the flexibility and decentralization of how to do that. But the key thing was we wanted to keep things small and fast. Jeff made me sort of do a handshake contract with him that we would target projects could be done in three or four weeks. I don't have to be done perfectly,

had to be the right projects and proof of concept. And the notion was if they bore fruit, we would then, you know, scale them, enhance them, etcetera. So it was pretty interesting, for example, on recommendations, there was a project underway, we got it under control, Jeff Marches in one day and said, hey, there's a simpler approach. Well,

I don't want to disrupt the schedule. And he's like, well, I really appreciate it if you would and I'd like you to do that. And it turned out the simpler version was highly, highly effective and the thing that I would have bet on was not, and so by keeping these things to relatively short development cycles were able to do experiments very quickly. Would you say? When you got there, kim described the code base was sort of a monolith, right? It was one big code base than, you know, and over time it got separated more into, you know, unique.

I don't know what the correct terminology is, but what were things would you say? Was it more like a minimum viable product at that time, in sort of today's language? Or was it pretty complex even at the stage when you joined structurally, I'm not sure I use the term minimal viable product, but I mean, again, when you're talking about a system where, and I don't think Shell gets enough credit in the public mind for building the store, and when the company was tiny, there wasn't the kind of budget that I had, where I would, you know, or that Dalzell had, when we go out and talk to a vendor,

we had a budget, we had leverage, you know, there were a lot of dimensions to it. When Sheldon needed a search engine, he wrote it right, and it was very clever and it's scaled up to a point and then it didn't, and so, you know, it was very it was a very monolithic system by today's standards, but that's how you get things up is classic for software projects, years off schedule, and multiples of budget. And so there's something to be said for, get it done and make it work and we solve a lot of scaling problems. My first winter there,

I just signed pOS for more bigger hardware because we were trying to do, we scaled something's there was work on splitting the database off, but it was an incredibly monolithic product and there are so many things that we did then over several years that today you just wouldn't do that way because we have, you know, things like AWS we have infrastructure, we have Knowledge and, you know, code libraries that just didn't exist five years ago. Yeah, it's definitely not a criticism. It's more of a fact, like if somebody who's 25 right now thinks about what Amazon was like, you know, as a product or program, you know, back then,

it was quite different because that was the way things were built in. And it was also getting it out in a hurry and testing and breaking things and all those things and so yeah, it's definitely not a criticism. It's reality, you have to right size things and there's no point in using the data centers worth of servers for a feature or technology or approach, you don't know, little work or not. And so it was a very long arc of balancing. Can you scale by writing a check for raw hardware and today you would just do it by increasing your AWS account limits, you know, in structures in terms of your use of AWS. So yeah, it was a fascinating set of business problems every month. Do we apply software engineers to scaling? Do we apply them to new features to bug fixes? And it was just a constant loop in that regard.

Today, I wanted to talk a lot about the transition from amazon from retail only business where you know, they're selling it and where we, or they were selling it, packed, purchasing it, put it in a warehouse of filling it to a third party seller marketplace business. Can you explain maybe put into context for people like how that came to be? Or was it a Jeff idea? Was it to you idea? Like where did it come from? This idea of, hey, we don't need to be the seller all the time. If you hadn't talked to Jeff Blackburn, I'm sure you will.

This recollection is but one october he approached me and we were just every year early on the christmas scaling nightmare began in october and ran right up through christmas and Jeff Blackburn who was doing business development, wandered in last week of october I believe it was and said he wanted me to help with some due diligence with an auction company. He was thinking of purchasing for amazon and I kind of pushed back and said, look we're a retailer, we have incredible scaling problems. I don't have any time to spare. It was like I really need to do this due diligence. I committed I think I think that was on a Wednesday, forget whether that company was coming to do a presentation thursday or friday spent half the day with the company was not impressed with their technology or their approach. But talk to Blackburn a bunch more. And then over the weekend he had gotten me thinking because as part of this evaluation, I'd take a harder look at Ebay than I had and it triggered some thinking about the, you know, when I had been an Apple, we were focused on our products and control of those products. And when I went to Microsoft,

there's a lot of philosophical thinking about platform, right? The dynamic of markets, building up the value of your products and you know, today the world has changed a lot. People forget that in that time frame Apple was on the verge of going bankrupt, right? My wife is a landscape architect as you know, and I remember telling her coworkers how Dead Apple was, you know, and that they were, you know, it's going out of business and it's you know, now, now $30,000 of hardware later, I'm firmly,

I'm firmly wrong, you know? All right. Exactly. So the things that Jeff Blackburn made me look at, got my brain going over the weekend. I spent the whole weekend hunker down over Ebay and thinking about platforms versus product and maybe to sort of zip backwards in a sense, anyone who is around amazon at that point in time was familiar with the terms selection price, convenience, right? That was a customer proposition. We took it seriously. It wasn't sort of just like random branding. We really internalize that. How do we make sure we are delivering selection price, convenience.

And over the weekend, I became convinced that we alone as a retailer could never lead the way with selection price, convenience if we were the only seller of goods on our systems. And so as a simple example, we were selling books, but there's always somebody who's going to get remainders or a bankruptcy close out from somewhere that could sell a book title we had for less or there would be a collectible collectible. There would be other things that we couldn't get. Somebody would choose not to sell on amazon. I always said it's like if I'm going to want to read some beach books, they don't need to be hard covers. They could be paperbacks with the front cover ripped off because I'm just going to read it. My hands are wet, they're going to be sanded it like you want different types of products or different qualities of products at different times. Exactly. And then there would be things where we just couldn't buy them. For some reason,

somebody had a self published book and they wanted to be involved in the transaction. And then of course the convenience proposition fails if people have to look at 10 sites for what they want. So it just got my gear spinning in the phone monday morning because I think the first monday that november, I grabbed Blackburn and said, we just have to talk to Jeff, you've got my brain going. And I think it changes the whole value proposition for the company. So Jeff Blackburn and I went into Jeff Bezos office a little bit before nine, if I remember right. I remember very vividly looking at the clock when we wrapped up and it was 9 20. Right? So this all happened in about half an hour and the three of us had a conversation about what would be to be a retail platform, a selling platform rather than to always be the one, taking the risk of buying the merchandise, trying to price it competitively, fighting the whole world as opposed to engaging the whole world and all of the points of failure that that would engender if we were fighting against everyone else all the time.

And it was very interesting, Jeff Bezos had to jeffs in the room, so to be clear, a little whiteboard in his office and he scribbled on the white board and we talked, and if I remember correctly, media metrics, the internet marketing analysis stuff still came out in binders. And uh, Jeff Bezos pulls out the Media Metrix binders and looks at a bunch of stickiness and re visitation things with respect to Ebay. And by 9 20 or so, he pretty much said, you're right, we have to do, you guys are right. We have to do this because it's just a compelling proposition. We cannot fulfill the brand promise without making this happen somehow.

You know, at that time we already discussed pricing mechanisms, etcetera. So there were a lot of reasons why we talked about auctions and Z shops and what not. But early on literally from the beginning, the notion was to be a selling platform and to have different pricing and shipping mechanisms, etcetera. It is a very long conversation possible about this. But before 9 30 Jeff Bezos said, make it happen, be responsible with the shareholders, resources and assets don't break things too badly, which proved to be a very challenging thing to try to do this. You know, look through those challenges that Jeff Bezos alluded to, that those things.

I think in the 2016 shareholder letter, he told me, he said pull together senior management meeting for saturday. We have to start socializing this and getting the company used to the idea that we're going to do this. Did your role change with sort of driving on marketplace? Or was it just one more thing on your plate at the time? So at the time I did this, I was the general manager of the store. I was reporting to David Risher at the time who I just love to death and I want to give a pitch for his uh, nonprofit organization world Reader, you know, he's doing wonderful global work with that. But, but no, I was general manager of the store and I originally thought we would hire someone to run our platform efforts and it just turned out to be a really challenging recruiting problem and I was convinced it was super important. So I ended up shifting into that role and focus 100% of marketplace for a couple of years.

There were 100 things that made marketplace difficult. Like maybe talk about why did we decide to start with auctions rather than, you know, maybe that's all we had because we saw Ebay and we said we got to do that and probably a big mistake maybe, but we'll talk about, you know, how we made the decision or how you and people at the top made the decision to do auctions first and how quickly it spun up because I remember it spun up fast, I remember getting pulled in as a product manager and you know, I'm like, I don't think, I don't even know what auctions were at the time, you know, I think why auctions had to do with two things, it was a desire to step into and learn about variable pricing and because there was an inevitable collision with Ebay and dressing that sort of competitive situation head on as I think part of it too. But again, I don't recall those discussions exactly,

but we're very quickly decided variable to go for variable pricing, which in the form of auctions and as you know, we have a lot of discussions about how to structure that, and even the pricing models of auctions themselves, but immediately, you know, we said about looking at a fixed price model, which led us to purchase steve Leslie's company, We bought live bid, which was matt Williams and Sky Cruise Company. So, you know, we we approached this both by developing inside doing a combination of talent acquisition and sort of tech and intellectual property through buying these various companies, like people didn't know for a long time amazon had two or maybe even three different buying pipelines. I don't know if you remember when People would put something in their shopping cart. And one item was a marketplace item and 1.1 item was a retail item.

They were essentially checking out on two different payment platforms. If I'm correct, yes, I don't even remember the mechanics and they changed over time. But we initially started with our marketplace mechanisms as a separately build platform from the main line retail stuff. And that was a function of really just pragmatics, I think, you know, and you couldn't break a retail stuff. And also we didn't understand one of the things that happened that Monday morning was a discussion about learning. It's one thing to go out and try to compete with a company based on you have a slightly more efficient technology or, you know, you can do it better or you can address a slightly different market. There was no best practices for being a global online retail marketplace, right? Just there was no guidebook and we actually discussed the fact that we would have to have a lot of fortitude and conviction in order to make it happen.

And we took a guess, and it turned out to be about right. It might take three runs at the problem, but we had conviction because again, when you went back to that selection, price, convenience analysis, it just had to happen right. There was just no avoiding it. And so we had to find a way to make it work. And the discussion was very clear on the fact that we would have to accept pain and failure in a sense. Along the way, I could have done with a little less failure over those couple of years. They were psychologically difficult years until we figured it out. They were for all of us. Yeah,

they took a tall, you know, I think it's in the 2016 again, shareholder letter Jeff talked about sort of fundamental misalignments and marketplace did not align readily with retail, you know, as you know, probably better than any human on earth the review and reward structures for product managers who owned product items. They had to worry about stuff they bought, selling that inventory, how fast they sold it, their margins, etcetera. As you know, we generated no great love with those guys when we allowed other people to sell on our site and even put links from their product pages into 3rd Party seller pages that would undercut them and make it harder for them to sell their product, that the price that would have liked to have sold it, etc.

So it's hard. I look forward to talking to the people that sort of launched customer reuse. For instance, there was another example that was not popular at the time with the powers that be, you know, and we believed it was the right thing to do. It happened before I got there, maybe even before you got there but we believed it was the right thing to do for customers. And so we needed to figure out, you know how to make it work for let's say authors for instance in that case or or the manufacturers of a product that was getting one star reviews. We said well long term this is going to be a good thing because we're going to help people buy the best product and return the items less often, you know, and and and but in the short term there's a there are some eggs getting broken. Yeah. I mean another example of exactly that was when we introduced the bestseller list because the new york times bestseller wisdom from you know, again a lot of years out but it was based on Cell in and so you know,

something could become a bestseller just because Barnes and Noble but a ton of copies, not because the public was consuming it. And we were doing hours based on cell through what was the public actually buying. And two things happened first a lot of authors and publishers got upset because it was sort of a paradigm shift for that and you know, I forget which 11 of the major ceos of one of the major publishing houses in new york apparently saw what was going on and through his phone across the room back when phones were wired to a lot of authors went through a transition from getting upset at that because there is the new york Times list to actually realizing, oh This is actually really reflects important stuff, I can monitor this in behavior around this. And then we also sort of as a along the lines of the way, marketplace sort of had to deal with seller and buyer behaviour. We had basically, you know, people starting to game it. There are some folks in Wisconsin that has self published book that they had friends ordered 10,000 copies of your overnight, the best celery. Yeah, there's also those great stories about authors sitting there and refreshing the page,

you know, waiting to see if their, you know, their books moving up and forget if that was done hourly or we did it daily. I can't remember, you know how that worked at the time with marketplace and maybe not just specific to auctions because maybe if you can give a short description of the difference between auctions and Zshops and marketplace for people that maybe don't even know what we're talking about, like with the shortest example of marketplaces. If a customer goes to the website and you go to a new harry potter, I'm dating myself there there will be used copies for sale and those used copies are sold from anything like big bookstores too, you know, which handle used books two, you and I listing the product on the site, But can you give a short summary of the evolution from where we started with with auctions and how we got to marketplace and why it was such a big deal. We started with an ocean, maybe a selling platform and buying platform. And for a lot of reasons we've probably in hindsight,

you know, it's hard after all these years think about where we just didn't understand where we were limited by technology and staff and the ability to move quickly because if we were going to touch Obidos, which was the core selling platform for amazon is retailer, you sort of got tangled up in timing and that particular database structure and whatnot. So somewhere along the way, we decided to sort of partition it. So we would do auctions, which was variable pricing, we would do Z shops, which was fixed pricing. We did an auction's partnership with Sotheby's as well, which was its own beast. And it was interesting because it was all geared towards becoming a selling platform and figuring out how to meet the needs of buyers and sellers effectively. It was interesting. We never thought of it as, oh,

there's an auction's thing that's going to stand alone forever and there's a fixed price thing that's going to stand alone forever. It was all part of this in a sense journey. And we referred to it as marketplace internally very early on and we very much didn't refer to it as marketplace externally, because we were pip squeaks at the time, right, you remember? But you know, even though we've gone public, we were a tiny company and you know, there are 100 big players that if they kind of thought about this the right way, you know, just the power of staff and capital to step into that space, if they saw what the Endgame was. And so we very deliberately did not talk to use the word marketplace externally for a very long time, and we didn't know what the end game was. So,

you know, we talked if you remember, we hypothesized about single detail page and unification of of inventories, but you know, that came after I was already gone. It was a big deal. I mean like in hindsight it's very easy to say what we shouldn't have tried, but at the time, you know, we were figuring out as we go, but one thing that was always a mess and we knew it kind of from day zero was the 575 different detail pages for Moby Dick, you know like the used version and we'd have 500 different listings and we're like this doesn't makes no sense like they're all the same thing. You know I would say that half dot com like we definitely had the idea before we saw half dot com but half dot com was an epiphany because literally we had all of that described, you know we're gonna fix price shipping for sellers, we're gonna put these listings right on the pages and that was the big change was like going from the 5000 copies of Moby Dick on the auctions of the Z shop site, all eight cents off from each other too,

Putting 457 copies available, starting at 35 cents right on the detail page. And that's sort of close to what we see now. But back then, that was the biggest part of the change, I think people take things for granted, right? Even with Amazon is just its own cellar of merchandise. There was a lot of work done on title authority for books and different even just different editions. If you're reading Lord of the Rings and you finished the second book, you want to third, you'd probably want the third in the same paperback edition, the same hardback edition, the same, you know, anniversary edition.

You don't want this motley collection of things. It turned out that figuring those things out was hard when you actually had ESPN's and industry information. Now you have people listening things, some of which are Moby Dick, the commodity thing versus like you said, a collectible Moby Dick. The first edition or the Herman Melville signed Moby Dick is a very different proposition than just listing it as you know. Oh yes, along with the paperbacks kind of a thing. So these are strangely hard problems. Did we refer to that as item authority even back then? Do you remember that is my recollection? I think I recall I could be wrong, but the catalog folks and on the tech side Rebecca Allen, I think maybe scott Northrop may have worked on that. Rebecca was very deep into that.

She actually solved a very tough nut by inventing the assassin which was compatible with ESPN. So it saved us, you know, it cracked a very tough problem for how do you migrate from selling things? I did by I. S. P. N. Two, How do you sell kind of anything? And for the listener, if you go to any product on amazon and you look at the U. R. L. I don't even know if S. I.

N. I don't know if that's in the U. R. L. Anymore. I think it's DP. For detail page. But A. S. I. N. Is amazon standard identification number I think. And so we basically had unique essence for every product in the catalog. But to joe's point you can have 15 different Mobley dictations that are, you know, it's the out of copy are out of trademark version.

It's the leatherback version, it's the paperback version. They're all the same words or mostly all the same words in the product. And so item authorities are really difficult problems, especially later on when you get things like clothing where you not only have unique items but then you have attributes like colour and size and sex and, and, and, and so that's a different interview at some point. You know, when you think about how difficult it is to go, you know, from books to music was probably pretty simple versus going from books, music and video to electronics, which was a totally different product, but getting off topic. So when you look back at marketplace and I've never asked you this like we should have just not done auctions right?

I mean it's nice to say we learned something, but like in hindsight, if we hadn't done auctions and we had done fixed price or something closer to single detail page out of the gate, like it was a long, painful learning curve, wasn't it? You know what I mean? Because so I want, I always wanted to ask you that I've felt that way. I'm like, it was way too much time focusing on Ebay and maybe because we do nothing about being a marketplace, but have you thought about that to a certain extent and this gets down even to the business environment and money, you know, one of the things you have to deal with is when you step into a new space, you're going to miscalculate certain things. So for example,

just to get to the nuts and bolts, I got budget based on what I estimate it would take to convince our retail customers to try out the marketplace side. See shops auctions, I grossly miss estimated that it was doable, but they were largely different buying bases. There were people who really, really Trusted Amazon but were hesitant about third parties and if you were involved in the 80s he guarantee and all of those things to sort of Put Amazon's brand trust behind the buying from 3rd parties. But you know, we got to a point where we actually had a pretty good understanding of what the platform characteristics were for auctions, for fixed price, etcetera, what it took to get a customer. And that was right during sort of the time when we had a lot of pressure from Wall Street, we were cutting people again, they forget it was not always a smooth, clear road.

And we were in the process of doing layoffs and freezing budgets and so on. This was a challenging moment. And, you know, I had some some discussions with our CFO or in Jensen because I figured what I thought we would take to make auctions successful and I wasn't getting that money. So you could argue there's all these points with focus on a fixed price, might have been smarter, spending more on auctions. Despite the pressure, financial pressures we were under might have been smarter because again, I felt at that point I knew what the economic dynamics were, but it's a real world, right? There are trade offs, there are hard things, so I have pondered a lot of people think auctions was a mistake.

I go back and forth as to whether it was a necessary stepping stone or not and I might not be the right person to answer that because I build a lot of my own blood, sweat and tears on that. You know, the funny thing is in the long range the one that still should exist is Sotheby's like that was a really nice site and if that were still on the amazon site today, it's probably where I would spend most of my time just out of interest, you know, like what, you know, N. F. T. S and all these things going on now. But yeah, again live and learn along the way and the Sotheby's thing was I mean there are a lot of things that made that more complicated and problematic and yes, I wish that had been done and Jeff and I had a few ideas that honestly would have been interesting to do and might still be interesting to do someday. So I won't enumerate them.

But there's some very interesting things you could do globally. But the crux of it was you can't be the best that offering selection, price, convenience without being a selling platform is what it reduced to one big shift is we had to build tools for sellers. So we had our own tools for things like buying and interfacing with suppliers. But can you talk a little bit about the that change, you know, where we had to build, basically seller central, I think, was called out of the gate and you know how that forced the organization to sort of change its way of thinking that we now had a different customer that we had to serve. There was no obvious easy sort of knowledge basis in the company. I'm sure there were people out in the world who would have immediately said, oh, you need this. And our sellers told us very quickly,

we needed to give them a way to connect to their inventory, to tie into their bookkeeping requirements, their compliance and tax requirements with their products, etcetera. So yes, there was a whole new space. You know, we had to start from ground zero. It sounds strange, but was that something that was almost like we knew what we had to build on the front end and then like eight hours later, like, oh jeez, we need to build all this stuff on the back end as well, like, because I kind of remember it like that, I'm like,

oh, I'm not just the product manager on auctions for the front side of it, we have to start thinking about all the tools and the sellers are almost as bad as consumers in their requests, their never ending and like, no matter what we built or how fast we built, it wasn't good enough. So yeah, I mean that's the sort of interesting thing, we bootstrapped ourselves and my recollection and again, there are probably other people who remember in more detail is exactly what you said, because we didn't have experience with sellers on the platform. We really had not internalized their needs and their wants until they kind of let us know in no uncertain terms what those were. So yes, it's sort of like we bootstrapped ourselves in the way that amazon's sort of inventory management systems early on, when I joined were incredibly, by today's standards primitive,

but they made sense for the scale amazon was at, so when I interviewed, the first time I went down to the, it was called the warehouse at Dawson. When you placed an order, you gotta space allocated on a shelf and as the parts for your order came in, they got put in your space. And when the system moved that, all the parts of your order, we're finished, somebody went and grabbed that and shipped it to you. And uh, you know, we went very at the time I joined, they were just finishing up what was called random binning and picking. And anyone who had done large warehouses earlier would probably have just said,

oh yeah, this is how you do it. So there are a lot of places where we hired UNknowledgeable talent, but other places where we just were like, oh, we need to learn this and understand and listen to our customers. It's funny, I loved working in the warehouse, but you know, it's probably just different mentalities, but it was just awesome seeing that sort of last touch before the customer got it and sort of magic. Again, that was just constantly evolving the software to make picking more productive, speed up throughput. It's basically a big ops problem and I loved it. It was fascinating.

The one thing that was really cool, even though it was not scalable about the original system, as you could walk a little company, you can kind of go anywhere. You can walk through the stacks of shelving and see orders coming together and visually see personalities and professional interests and people's professional development just by the stacks of books in there in an order. It was just a number on a shelf. You didn't know who it was or where they were, but it was just fascinating to see what went together. I do remember that too. I remember I didn't graduate from NBA like I remember thinking about how quickly you learn who the person is by. The types and books were perfect because it's expressions of interest. And you could just see how the computers were going to be able to make better recommendations over time. And you know, make smarter decisions about new categories, like which new category would be interesting to this pot of people vs. That part of people.

It was really exciting. So you left in 2001, were you there through the launch of marketplace, the successful single detail page? Or did you leave before we got to that? So, to the best of my recollection, you know, I left marketplace in the hands of Eric Ringwald, then Mike George took over. It was a couple of generations of this, of, of learning of normalizing this throughout the company because as you know, there was a tremendous immune system reaction to this part of the company, marketplace part of the company. And again,

I refer to the entire arc of this is marketplace competing and causing pain for the retail side. The retail side of the business did not care for the marketplace side of the business. We had a lot of resources. We were generating little revenue. And then when we launched single detail page, Joel, it was the example of three year overnight success or however many years it took like I remember that first day, it was a significant percentage of units soul. And suddenly all these people like marketplace that didn't, you know, for many years. But if I weren't working on marketplace, I wouldn't have cared for us much either. You know, when I'm starved for resources and I'm generating all the revenue, you know,

and we had these discussions at one point, you know, our revenue was so low when I offered from a accounting point of view to the product managers, I said, here's the deal. I got a nascent business. But how about I just credit your side with all of our sales and that didn't sing well and our sales were quite small. So it's like all you're doing is inflicting pain without returning enough to make that worthwhile. And our customers, I said, I'd originally miss estimated our customer acquisition costs into marketplace. But interestingly they were quite a bit less than the customer acquisition costs when I joined amazon. People forget fast forward a couple of years and people are used to, we have this business, people are buying, here's our cost director.

And there was actually a point where CFO Warren wandered down and told me he wanted me to shut down marketplace with that word because that was internally, I was referred to at the time, you said your spend is too high and your return to the business is too small. And again, it took a lot of corporate conviction and discussion and debate to stick with this based on the premise that the issue wasn't would it work? The issue was how did we figure out how to make it work? And again, you were a key player in that it was a long process of iterations and understanding how do you price for buyers have your price for sellers? How do you make it work? How do you even get adequate product in each category to retain buyers and sellers? Yeah, it was a big chicken and the egg. Like do we focus on the seller? Do we focus on the buyer? Like ultimately, I think the answer is you get sales for the sellers,

They stick around no matter the pain. You know, they give you a window to improve the tools and make things better. Which is what happened with the third iteration, the first two, they weren't getting enough sales because we didn't have it cobbled together well enough for buyers to find their products. And, but once we did, once the sales were flowing, we had the oxygen, if you will. And the time just constantly improve things for sellers. And uh, the flywheel got turning as you know, we had a nightmare of trying to get sellers and get buyers early on, especially in the Z shops days.

And we had a very active and great group of people recruiting cellars and cellars would come and get with the half life was, but it was measured in a small number of months. They would leave because they wouldn't having a buyers. And at one point Jeff had wanted to do a project where we would basically by one of everything on the planet to list in the catalog and everybody in the company pretty much hated it, including me. And we were talking about this chicken and egg problem and he said, well, the answer is back to my pet project in effect. And, and I was like, I got annoyed because I said, look, why do you want to compete with marketplace? We need to achieve this via marketplace. And said, I'm actually trying to help marketplace.

And it was simple but brilliant insight into the chicken and egg problem said it's really hard to get people to have faith and conviction, but it's easy to if you have faith and conviction to put your resources and your energy behind it. And so early on in the conversation, you started going back to this pet project of his that everybody pretty much didn't like including myself. And I just said like step back, explain why you're pitching this, Why are we having this conversation? And he said, because you have this chicken and egg problem and if you just bring in cellars, whatever category and they're not, buyers will go away. If you bring in buyers, they can't find it. We fail the convenience part and they go away. And so he brought up recalled. It's just an amazing conversation so we can crack the chicken and egg problem by stocking stuff from all the categories that we're going to recruit sellers for.

I think he said, he talked about fishing rods and horseback riding saddles and said, you know, if somebody is a custom fly rod manufacturer, they try to sell on amazon today. There's no buyers, they're going to get frustrated. So if we can develop an inventory of fishing rods so people want to buy fishing rods come, they'll be here and if they see a great custom fly rod, maybe they'll buy that. And now our sellers have a marketplace and that really, I think it took a long time. And then there was the tech side and going to a unified detail page downstream after a few years. But that was a really important realization that we had to use the things we could control to crack that chicken and egg problem. When you step back and squint and get away from the details about auctions. The shot,

like what broad lessons do you think future entrepreneurs can learn from the big amazon retail to marketplace transition because it was a big, big scary bet at a time when we were not financially secure. You know, so again, stepping back from all the details like what do you think the sort of lesson is here for future entrepreneurs? You know, it's easy to have conviction because you want to believe Yeah. And people drive themselves off cliffs all the time because of that. When you figure out something matters, it's an imperative. You have to have conviction and you have to make it happen and you have to find a way to get the resources to make it happen. And that's not always easy. I think being adequately capitalist, if you're going to talk to entrepreneurs under capitalization is like a death knell because you can't go out and have repeat failures and we tried to fail smart. But whether it was on building distribution centers was warehouses and distribution centers and has no known as foot fulfillment centers, we acquired enough capital.

We could move fast and learn and same thing here, even though we went through a bumpy time in terms of that 2000 and one dot com bust thing and that made life hard, We had capitalist ourselves very well and that gives, it gives some navigating space and then again, just you have to listen to the universe. We didn't stick with auctions as they were, we didn't stick with the shops. We acknowledged internally where we were getting it right and wrong, Jeff engaged and other senior folks engaged in really important conversations about what are you guys getting right, what you're getting wrong and why? And again, is thinking about inventory and using our resources to create a cellar pool. You just have to think of these as multidimensional problems. I don't think there's an easy, simple nut to hand out, I guess I'll end with,

who should I interview next Or who are some people that you think have interesting stories to tell? I know you've shared some before, but anybody as you're thinking about this comes to mind as having made a gigantic impact and their stories should be heard. Let's see, Gosh, we've talked about, you've got calling Bayram janu Frankie sleeve Wendy wolf, J J. Absolutely. You know what was JJ's biggest, everybody has their claims to fame. What did she work on early on? So, JJ letter, our initial efforts with respect to recommendations. So her team,

including Greg linden and we ended up with their advancing on that team, you know, she was also just really helped with sort of what I would call engineering social cohesion. We had a number of people who were very good at that. So she was very good at pulling people together to work on the right things at the right time, which mattered. So yeah, she was great, You know, and again, the work we did on recommendations, there were no good templates, there were no touchstones, no examples to follow. And so she pulled a lot of that work together. Artificial intelligence wasn't the hot term at the time.

It was, you know, I look forward to talking to Duane and Ruben about the early bottega boxes and oh my God, that was. And again, that was part of flexibility, Dwayne, I just resolved a bunch of scheduling issues with Jeff the next morning, Dwayne and Rubin showed up and said, oh, we have an idea, we'd like to blow all of our schedules. So we did right. And they made the case, I took the case to Jeff and there we were. But yeah,

compared to machine learning today, this is also so much of this is very, very primitive. But again, it was, it's a step at a time. These were the original building blocks for doing this kind of work. The other thing about JJ is like was with a lot of the other folks we had, she was Jack of all trades, right? She unique system administration. When we realized we needed seven by 24 coverage on a bunch of things. She camped in the office for a long time and worked on to do double duty doing systems work and software engineering work, and management work etcetera. So yeah, I mean there's just an astonishing set of people and I just,

I don't even know how we ended up with such an amazing group of people because that was really critical to surviving into making things work. I agree. Final thing to put this into context for people Marketplace. I was googling earlier today. It's more than 50 of units sold. I don't know if dollars, it's hard to always compare some of these things, but more than 50 of units sold on Amazon and I think this was a few years ago they passed the 50 threshold. Our marketplace driven, you know, so they're basically third parties are probably a lot larger ones than the ones we started with working with in the early auctions days. But you know, it includes small sellers because I still sell things from time to time on the site. So it's a gigantic part of amazon. If you haven't ever noticed it before, go take a look at the pages you can list as a seller and maybe changes the way you think about the platform from amazon is a pure place you buy stuff to a place where you know,

hundreds of thousands of people make their business selling on amazon and Ebay and other platforms on the web. So Joel thank you so much for being a guest on the podcast. I think we could probably do a few more hours on different topics. So I may reach out. I personally found it really interesting and I'm positive other people will too. And as usual, it's great to catch up. I've told Joel this directly but Joel made a huge impact on my life by giving me way too much responsibility to either screw up or succeed with. And so and he also invested in my company after leaving amazon. So huge debt of thanks to you, Joel and Karen for the audience. Thank you for listening to the podcast. If you'd like more details about what we discussed today or want to contact me with edits or suggest things we got wrong, which is totally possible. Please visit invent like an owner dot com to sign up for my weekly newsletter and be sure to subscribe to this podcast to get all the future episodes. That's it for today. And remember no sniveling. I don't know. Mm Yeah.