[Auren Hoffman] Welcome to world of DaaS, a show for data enthusiasts. I'm your host, Auren Hoffman, CEO of SafeGraph. For more conversations, videos and transcripts, visit safegraph.com/podcasts.
Welcome, Steven. Welcome to the World of Daas podcast, really excited to have you on the show.
[Stephen Orban] Awesome to be here on. Thanks for having me. Always a pleasure to have a discussion with you.
[Auren Hoffman] Alright, so let's dive into the AWS Data Exchange. I think it's like a super interesting play for AWS, just give us like, the brief backstory. What's the history? How did it actually happen? It's just some idea where you guys brainstorming on the whiteboard or was it very obvious from the beginning? Or was it just kind of a non obvious manifestation that some customer came through or what happened?
[Stephen Orban] The story is, is very similar to, in fact, identical in a way to any service that AWS or Amazon really starts. You know, we do everything by working backwards from the customer, and what they actually need, what problems they're telling us how we can make their lives easier. And in this particular case, I started at Amazon six and a half years ago, I was leading our enterprise strategy globally, basically talking to big enterprise customers about about their just overall move to the cloud. And one of the things that we kept hearing was, hey, it's great that the cloud makes it cheaper, better, faster for us to develop our applications, analytics and machine learning models. But a lot of the times these applications, analytics, and now machine learning depend on data that we don't already have. And hey, it's really weird that I have all these awesome tools in the cloud. But when I need to go get data from somewhere else from a vendor I work with, or a website or whatever, I'm still stuck pulling FTP sites, some of our customers still ship hard drives to each other, or, or having to hit you know, dozens or if not hundreds of different API endpoints. So we decided to study that problem. And we did. But you know, I went through the same process, we call it the working backwards process where we write a press release. And instead of accompanying FAQs around what we would launch what what the customer problem is, and what we would launch to solve it. And I went through that process for a long time. It's a it's an onerous process where you're talking to lots of customers, you get lots of feedback, you, you know, see, could
[Auren Hoffman] You like build, like, you have to like, go get budget from someone else, you have to like build a business plan internally as an internal VC that you go to like you get your series A or something, or how has it work?
[Stephen Orban] So when you're writing it, so it depends. There's a bunch of different approaches you could take in this particular case, in my approach, since I was in I wasn't in the product and services area for AWS, I was in our field organization talking to customers. So I wrote this pier FAQ, which ended up including a business case, on not just what the customer value proposition was, but but how is it going to be? You know, how is it going to be beneficial for customers? And then and then what's what's it afraid of us to and so you know, all of that ended up in a fairly comprehensive in this case, it ended up being a set of documents that we reviewed over a period of time before Andy Jassy who's the CEO of AWS currently on his way to becoming the CEO of Amazon. Overall, as everybody probably knows, basically gave it the thumbs up, and we got our funding to, to start the business and launched it in November of 2019.
[Auren Hoffman] The way I think of the data exchange is, is, is almost similar to the way people might think of Amazon as a store, like it's a store to go find, discover and buy data, you'll learn about the different types of data that you want, just like you might learn about a charger you might need for your laptop or something like that. on Amazon, you think of it similar.
[Stephen Orban] I mean, there are a lot of parallels to data exchange, both to the Amazon retail marketplace, but also to AWS has a marketplace for third party software.
[Auren Hoffman] Which you now are starting to run, so congratulations on that. That's a big, that's a really big deal.
[Stephen Orban] So it's a little bit of an aside, I started the data exchange businesses, as we said, And anyway, I'll kind of come on to that in a second. But we ended up putting the data exchange business next to AWS marketplace, because we knew, and it was true that there was a lot of API's and processes and mechanisms that AWS marketplace had built to help ISV partners been third party software to our customers. And so we use a lot of that plumbing and infrastructure to start AWS data exchange as well because of the similarities that you pointed out and I'm going to very fortunate position a couple of weeks ago, my boss David McCann, who ran the marketplace, and who I worked for to start data exchange has decided to downshift and play a part time role. And Amazon moving forward. And I was asked to oversee all of the marketplace as well as data exchange and a couple of other services we have.
[Auren Hoffman] One of the interesting things about marketplace and also, the AWS data exchange is kind of like how you how you guys have figured out how to incent your own internal sales people and customer success people to sell other people stuff, right. Because when you're selling, when you're selling AWS, there's there's a certain margin that you can play with. And you it's a little bit easier to figure out how to you commission your salespeople when you're selling the data. And your cell and a US marketplace, the the margins compress? Because it's other people's things, I presume, right? So you have you have less to play with. So how there was kind of like a real innovation in AWS marketplace on how they commissioned their salespeople, which I think you borrowed for the AWS data exchange. Is that right? Can you explain that?
[Stephen Orban] Yeah, so there's, there's a number of things that are going on here, I think it's kind of useful to do to unpack several of them. So just like we work backwards from customers for how we develop product, using the working backwards process that I mentioned, pretty much our entire sales strategy is helping a customer be successful, do something I need to be us, you know, by working our solutions, architects or professional services, teams, whatever, whatever the sort of way, we're helping a customer. And then we then celebrate that this customer was able to achieve x with their business outcome, which is better for them for these reasons. And then we celebrate that both with the rest of our sellers across AWS, but also just more more broadly as part of our as part of kind of our marketing because because we believe there's there's no better person place entity to tell the story of how customers can be successful than then customers themselves. So we built the marketplace and data exchange in a very similar way where the best thing to motivate the sellers in our field teams is to see how a few things work. And then how they could take those sort of wins. And what a customer did to customers that looked like them. And kind of have a similar conversation. So just to give you an example, we had a number of customers in the food services industry, who were using location data, for example, which I know safe graph is specializes in and has helped partner with us for some of these customers. And they want to know as COVID and the pandemic has gone through its kind of lifecycle over the last couple of years, what foot traffic was gonna look like in, in their establishment. So we had one, food distributor and food services company do that. And we celebrated that. And we took that to the account teams who work with other food distributors, and they had those similar conversations with our customers. And the great part about that is not only are we able to then tell the story through the lens of of customer success, but those customers tend to then use that data in more analytics to drive you know, even more differentiated capabilities for their business intelligence or whatever, which of course, then drives additional AWS adoption as well.
[Auren Hoffman] What I one of the things as an outsider, I find that super fascinating about Amazon as a company is yes, it's incredibly innovative. It has all these great technical innovations, and then and then and then it's kind of known for its like customer centric innovation, like there are things that just won't will never change. Customers always want low prices, they always want speedy delivery like that will never change. But But there's also all these like Organizational Behavior innovations that Amazon has. It's got like the two pizza rule, kind of like it really like champion micro services. Again, like the way the sales people are kind of done is in some ways very damaging. I think a lot of people are copying all these things throughout it, is there something about like the water and Amazon that it's constantly creating all these like, because it's a fairly big place. And as big, big places are not known for innovations, especially not organizational behavior, innovations.
[Stephen Orban] Yeah. So like, I'll give you I'll give you the synthesis that you might hear, Jeff say, Jeff Bezos, our founder and and then I'll give you sort of some of my own kind of opinions around that. Jeff will say, three of the things that have kept Amazon successful and as you know, been able to grow our business, the way we have is one to be customer obsessed, and constantly listening to the customer and what they want. And, you know, customers have this amazing quality of being perpetually dissatisfied, they're always going to tell you what they want you to do differently, or more or better, or whatever, you know, you can take that listen to it and continue to build, you know, a pretty valuable business for that. The second thing he would say is that we're not afraid to invent and pioneer, even if that means sometimes it's inventing some sort of overlapping value proposition with something else, we have it at Amazon. And I think that's a little bit unique when you were talking about some of these big companies who, you know, might not move as quickly or whatever. And oftentimes, they were worried about cannibalization, or inventing something like we just, it's just not part of the way we think about things if it's a customer need, and we think we can solve it in a unique, long lasting value creation way. We'll do that. And the third thing is the long term focus. And being patient is the way he kind of describes it. So the sort of combination of those three things you said, Is there something in the water? I think the combination of those three things is, is is, you know, the way he kind of drives the company. And then my from my perspective, my personal opinion, in terms of like how we're able to scale so many of these different businesses, we have these things called the leadership principles, and there's 14 of them. And each of them are very well thought out. And we're, you know, I wasn't around for this, but we're, you know, hotly debated, and literally every word about them. And I think of them as, as tenants or beliefs on and how each of us as leaders in the company operate with one another. And the sort of magic if you will said, Is it the magic in the water, the magic with the leadership principles is if I have an initiative, and I need to drive something from another team, or take a dependency on another team, or need to reach out to somebody about an opportunity somewhere else in Amazon, and maybe it's somebody I've never worked with before, never met, we've never done anything together. I know. And the way I'm approaching them, asking my questions, getting answers and collaborating with them, the rubric by which they're evaluating the opportunity, and how they're going to work back with me, because we have these leadership principles that are written down.
[Auren Hoffman] Like a common language, a common formula. So you, you it's almost like an API for people. So you know how to interact with one another.
[Stephen Orban] You know, nobody's ever put it to me that way. But that is a very fascinating, and I think appropriate way to, to put an API for people. Yeah, I mean, look. So I was a student of Amazon long before I came here. Yeah. And in 2007, or eight, I guess it was, I was leading engineering teams at Bloomberg and I heard about AWS launching s3. First, I got fascinated with it, because my wife started to buy diapers and paper towels and everything for the house from this online store. And I didn't understand that. And then, you know, they started to watch cloud services. And I was like, Who is this company, I started to learn about the leadership principles, and I just blatantly stole some of it into my own leadership philosophy. And it made me a better leader without fully understanding it, that I got to Amazon six and a half years ago. If I'm being honest, you know, when you first get here, we do doc reads for every meeting, and everybody sits silently for the first 1520 minutes of every meeting while we're reading a doc. And then we have a sort of a healthy debate around it. And these leadership principles just had an API for people, they get thrown around the room like frisbees, and in the beginning, it's like, wow, this is this is this is pretty intense. This is pretty serious. But it didn't take me very long. I was like, wow. as as as sort of strange as that may seem, for a newcomer, it really works. And it has like this crazy alignment property.
[Auren Hoffman] Or let's let's, let's double click into the data. And then like, I later want to just understand, like, more of the mechanics of like how other data companies can be successful because a lot of data companies are listening in how they can be successful on the AWS data exchange. So one of the, if you think of like a data science thing, like it seems, it seems a good idea for our data science team, let's say your data science team at McDonald's or something like that, to first focus on like doing good data science on your own internal data that you're already creating. And then once you get to some sort of level on the curve of getting good at that, then you might want to bring in external data is that the way you think of it, and obviously, you'd want to get all the way through where there's an asymptote, where you're not getting anything out of your internal data anymore. So we're on the curve should like an organization like McDonald's start Investing in external data?
[Stephen Orban] Yeah, I certainly agree that investing in data science practices in today's day and age is a no regret move, whether it's for your own data and better business intelligence around it or something that you're going to augment later with third party data. And in fact, I would maybe be a little bit stronger than call it a no regret move. If that's that could be a little bit. I don't know if this is controversial or not. But I think if you're not as a company investing in it, you better be careful because your competitors are and they're looking for it angle that they can to better compete. And data science is certainly going to be a growing area and spot where they're going to be able to do that. So I think investing in it and building muscle around it, and understanding your customers and their behavior. And your partners and their behavior is just like that's just going to become increasingly important. As you do that, you will find more opportunity to go get third party data sources that you might not otherwise have used advertising as a very common example in the advertising industry has been advertisers have been augmenting their own data with who they want to be able to target ads with, you know, for a long time. And that's not new. And then I would sort of say that there's there's other bucket of use cases where a customer is in some position, or companies in a position where they just have a question that they need to answer because there's like a systemic risk to their business or because they have a hypothesis that they need to vet and they may or may not have any data science practice, and they may or may not have any first party data to be able to go answer it. And even if they did have the first party data, oftentimes this data is siloed away and organizations and there's some it gatekeeper who sees over it, and it's too hard to go get. So they go out to tender and or to the market anyway and go find something. So then there's the very sort of specific use case driven data sets. And that's one of the things that we've seen, particularly again, around the pandemic, where all of a sudden, there was a huge shift in everybody's behavior across pretty much any dimension you could possibly imagine. And every business was trying to understand what that was going to mean for them. So, frankly, I was super nervous because we launched the business and three months later, yeah, at least in North America, that pandemic hit, and I was like, Oh, my God, what's going to happen? And a lot of the status quo stuff we would have done, particularly around like financial services, kind of, you know, was put on hold, but all these companies needed to figure out what the pandemic meant for them. And that was their catalyst, data science practice, or not, to sort of get started with it.
[Auren Hoffman] If you're a data buyer, now, you're your McDonald's or your your, your Domino's or whoever, and you're coming on, you want to you want to, like, there's so much data on the data exchanged? Like how do you even find what you're looking for? How do you? How do you know? And I yes, their search mechanisms and stuff, but, you know, how do you end up like, with a good experience getting the right data? that's right for you
[Stephen Orban] Yeah. It's a great question. And and if I'm being honest, I think that's an area where we're not just data exchange, but just generally speaking, the industry has a lot more work to do to kind of make it easier to get to insights from a huge treasure trove of, of data. So you're right, we have search capabilities. And there's a number of different ways you can facet our catalog by searching through various industries or our terms or what have you. On top of that, we have what we call data exchange, customer advisors, and particularly for customers who are either heavily using the service or or we believe over time are going to heavily use this service, because they're, they're predisposed to using lots of third party data. We work directly with them, and will understand through our account teams, we're usually already working with them. Because you know, we benefit from having millions of customers already worldwide using AWS services to understand what data they need and why for what use cases to power.
[Auren Hoffman] Imagine there's like a collaborative filter that happens like if you buy this data, you should buy this other or if you use these tools in AWS, you should you might be more interested in it. You can do imagine something like a little bit more sophisticated happens in the future.
[Stephen Orban] I do. I do. I think, you know, we already see that happening and a little bit more of a scrappy approach and you're kind of describing and I would just kind of tie that back to what I was saying with the with the use cases when we see a customer really successful. One of our customers built a dashboard to help them optimize their supply chain, in times of the pandemic, for example, and they serve as both supermarkets and restaurants. And the beginning of the pandemic, all the restaurant traffic shifted supermarkets. But it was very dependent on location. Because Florida versus New York versus California versus Texas, yeah, populations there took very different approaches to social distancing. So they couldn't take a uniform approach. So they use a lot of location data among a number of other data sources to sort of build dashboards around that we took that we packaged it up. And we celebrated that. And then we saw other customers do similar things. And then we had another customer in a completely different industry here on consulting group. And healthcare and Life Sciences build comparable dashboards, but for different purposes, to help their customers who were like the mercy ambulatory centers and other health care provider institutions, to help them kind of optimize staff and shift changes. So we do kind of see that it's a little more kind of learning as we go in iterative, then then maybe some master algorithm like you've described, but, you know, we're gonna keep getting better at it, and iterating on it and make it you know,
[Auren Hoffman] Better and better over time. As you join data sets together. It's not like one plus one equals two, there's some sort of like exponential value as you join these things together. And they really only become valuable as you join them together. So if you just bought data, and it just sat there, it doesn't give you as much value as if you can integrate it into your own systems. And one of the ways of course to do that is like some sort of join key where you can like join them together. And we've been involved at safeguarding an open source initiative called place key to join data about places together, but there's many other types of things like how do you
[Stephen Orban] Have I would say, you have a particularly rich background in this space. Before safe graph, as well. But yes,
[Auren Hoffman] How do you guys think about like these join keys? And how do you Because obviously, like, the more we join these exist, the better it is for for for, for em, for AWS, both, because they'll buy more data through AWS, and they'll probably consume a lot more services, because they're bringing in much more data and they're able to use it more. So you're, I'm sure you're you and the customer will probably get more value. So how do you think about like, which join keys you try to promote which join you? How do you think about this whole kind of thing through the through your own lens?
[Stephen Orban] I think I'll start by saying this, we took a decision or a tenant an operating belief, when we started the service, that data providers just fundamentally know their data and their customers and the use cases for their data by customers better than we ever could, particularly at the sort of scale we're going forward. We want to have any data set that our customers want on the service. So it's very industry and domain niche, use case specific. So there was no way we would be able to predetermine schemas taxonomies formats of data, we wanted to be as flexible as possible, because we just fundamentally believe that customers want, you know, the best choice. So we're not opinionated on how providers sort of pick and package and what join keys if at all. They use any now that leads to the conversation with customers and data lead with what you're describing, which is okay, this is great. But I need to marry this with my own first party data, or I want to pull these four datasets together. And how do I do that? You know, I think one of the reasons that this is an interesting business for AWS is because we have another 200 plus services in our portfolio that help our customers do stuff with data. Yep. So we work with customers and our account teams, our solutions, architects, help them take the data that they're getting from data exchange, and then use it in the rest of AWS, which oftentimes comes down to join data. And I think you will see us continue to innovate in this space where data providers can can can have a little bit more knobs on how they're describing their data as they publish it to customers, so that it can be more easily and quickly. I'll call it activatable. And sort of air quotes, as Yes,
[Auren Hoffman] It's certainly one of the biggest one the biggest hindrances to join to bind is like, I know this day is gonna be valuable, but it's gonna take me 50 hours to do the whole ETL to get it into my system. And I've got, I've got a lot of other engineering priorities. I don't want to take the 58 now if it took, like, if it took 50 seconds to move into my system, I'll buy I'll happily spend the money. So it's not it's not the money is that I just don't want to spend the money and have this data sit on a shelf. I want to actually spend it and get it useful right away.
[Stephen Orban] Yes, that is one other. There's one other sort of area worth exploring here. So we're also working on partnerships where there might be a ISV tool ISVs independent software vendor like a tool that a customer already uses. to reason about their data, and we are plugging our API's into more and more tools, so that when a customer subscribes to data on data exchange, they can very quickly use it in a tool that they want. Whether that's an AWS first party tool, or one of our partners tools, and an example I would give you, Deloitte, who many of you know is that the big consultancy and auditing firm, they have a business, which is neither of those things. It's actually a software business, and it's called converge health. And it's a set of software tools that they licensed to healthcare and Life Sciences companies, namely the ones we're involved in our pharma companies. And pharma companies have a lot of data scientists and modelers who are looking at all sorts of data around what they call real world evidence, how drugs are actually used in the wild, how long they're here to buy patients, how often doctors prescribe them, what their efficacy rates are, so on and so forth. And so this converged health business has a tool that these data scientists within these pharma companies use to model and ask questions of data that they have within the pharma company. Well, oftentimes, that data scientists will be searching for something. Let's say they're looking for comorbidities between, you know, diabetes, and Coronavirus, for example. Yeah. And they might not have the data already, for whatever reason. So they've built an integration with us. So they can also search the data exchange catalog, click a button to send off a procurement workflow that goes to their compliance department to then kind of procure the data through data exchange, once it meets all their standards and governance requirements. And then assuming that it does, they're a click away from that model within their tool
[Auren Hoffman] That's even ever an extra barrier, because it has to be HIPAA compliant has to be done in the right way. And, you know, these patient records are incredibly valuable. And so you have to figure out a way how to move it in an A de identified way. That's super, super, super, super, super interesting. Now, for like a data owner. So like, like a company like safe graph or data owner, we put our data on AWS data exchange, we want, we want potential data buyers to be able to find our data. So like, what like, what advice would you be like how to listed? Or how to SEO add on it? And how to use the right keyword? So it's found? Or are, you know, do we do we work with like, the AWS raps or to help educate them? or What advice would you give to a company like us?
[Stephen Orban] Yeah, I would say all of the above of all the things that you said, we, we do give some best practices for the listing based on the stuff that we're constantly learning, by the way, I'm going to services year and a half old. So I wouldn't say we have all the answers yet. But, you know, as we continue to find ways to optimize listings, they are all indexed by search engines. So if you search for AWS data exchange safe graph, you should find the listings that you have on us, and if you if you put certain keywords in there, they will come up as well. And then again, I'll kind of come back to that, how we how we help our sellers worldwide understand is that we kind of get some wins together. And we work Yep, to help them solve some of their use cases. And, and get that out there. The other thing I would say is, you know, people are really willing to use free data sets, obviously, it's not easy. Yeah, any money and then the more expensive it gets, the more sort of scrupulous and the longer the sales cycle is going to become. So I think it's really valuable to have a number of whether you consider it a free trial or a sample or Yeah, data providers use different words for for the way they kind of describe that but, but having something where customers can kind of get a feel and understand how they're going to join it with the other data sets or how they're going to even if it's just like the schema with 10 rows, for example. Those sort of data sets lead to the conversation for some sort of like larger deal. And the idea that you would be able to list a data set, whether it's on our data exchange, or list anything anywhere, in my opinion, for six figures, or more, which a lot of these data sets are worth Yeah, and somebody's just gonna come and click a button and buy it is like, that's just doesn't happen. So
[Auren Hoffman] How does the bind motion work just like, is it somebody comes, they see some data they download, as you mentioned, you know, a small number of rows and, and kind of check it out, and then they hit like contact sales and then and when they hit contact sales, are they talking to the company that sells the data, or are they talking to other AWS wrapper? How does it work?
[Stephen Orban] So the answer is a spot somewhere. Yes, yes. A lot of it often starts with that with with the with the free trial and testing it out, there's there's this, you know, at a couple $100 or a couple $1,000, we do drive, self service subscription. So at those levels, if the listing is clear enough, the value from the data is, you know, well articulated enough, we do have some customers who will, of course, just just buy another AWS invoice when it starts to get higher value. It's very important, I think it's one thing that everybody leaves with, we are facilitating the sort of transactions and hopefully accelerating the sales cycles for our data providers and ultimately getting customers what they want, but they still maintain a commercial relationship with the data provider. So it's a requirement for all of our listings, Where, where, where everybody has a support, email and channel that they that they're active on. And we test that you can't just like throw up some support email and never answer we that's something we monitor in our catalog, because we want our customers to have a great experience. So the customer in practice, they do both, they reach directly out to the data provider sometimes and sometimes they just reach out to their AWS rep, because maybe they have a great relationship with the AWS rep, and then AWS rep will reach out to my team, and we'll start the dialogue with the data provider. But when you're talking about some of these higher value data sets, that will often then go into Okay, well, I wanted to try all but your trial didn't have what I wanted. So we have a number of features and capabilities where you can package up private products, for whatever fee or commercial terms you want, that aren't listed in the public catalog, you can still do all that directly to the service in sort of a hand to hand way with your customers. And we support you know, and everything that's needed to follow that sales cycle through to the end.
[Auren Hoffman] Now one of the things is, to me, it's interesting about what you do, anyone who's ever used AWS knows that you can commit to a certain amount of spend. And if you commit to this certain amount of spend, let's say for a year, you get certain discounts and certain higher levels of support. And the more you commit to the more you get, which which makes sense. And I know that both for AWS marketplace and for the data exchange that at least at least at some percentage of those dollars goes toward that commit, which, which to me kind of blew my mind and was really cool. Like, how did that decision that that's kind of sounds like a small thing, but it does, it does seem just really start to drive a lot of behavior for the end customer.
[Stephen Orban] Yeah, so we have a bunch of years ago, AWS started doing what we now call private pricing agreements, where were our customers starts to get, you know, bigger and bigger and using AWS in a more meaningful way. And in exchange for some, you know, usage commitments over a period of time, they get a number of benefits, you know, seven to service discounts among them. And, you know, this, this decision actually predates me for marketplace purchases, as you said a portion of of the spend, that goes to the marketplace, lands towards that commit. And then when we built data exchange, it was only natural to kind of follow on the same motions, because it's such a similar business, as you got to write it out in the very beginning. So we follow the same suit. And, you know, that can be very beneficial. Like, the cool thing about that is we get to have conversations with more than just the kind of data buyer persona, because then it's also the procurement professionals at some of these big enterprises whose You know, they're they're incentivized that optimizing just overall corporate spend, and they're used to dealing with AWS in the invoice. So it becomes a benefit for them to where they can, you know, they can make improvements to, you know, their overall financial spends, by starting to combine some of these concepts.
[Auren Hoffman] Now, now, this might be a little bit like long term, little crazy thinking and stuff like that. But you can imagine, so you have these, like data science, machine learning models that are being run on AWS tools for a lot of your clients, and they're trying to optimize something, right? optimize whatever revenue or whatever they're trying to do to optimize, they've got some sort of optimization function. And, and you can imagine a scenario where you could just like, they all they care about is optimization of that data science model. They don't even care what the data inputs are now, and then you could help them like bring in the different data that will help optimize that more and be better and then they're, they'll gladly just pay instead of paying for the data, they'll gladly pay from going to like, point eight accuracy 2.85 accuracy or you know, or several, whatever metric they're looking at in the models.
[Stephen Orban] Yeah, I mean, we'll see what the future holds. I will say this, in addition to data exchange, We actually launched before data exchange, we have a machine learning model marketplace. It's part of the AWS marketplace family, which is, which is also now part of my remit. But we've got several dozen machine learning models that customers can then that run in their own secure container inside of their accounts, where the vendor who is bending the model doesn't see the the data that the customer is inputting into the model that the customer is just getting the inferences that were pre trained by the vendor who listed it. So I don't know if it quite sort of gets to the gets to that Nirvana that you were describing, where it's like, Hey, here's my business outcome that I'm looking for. Help me optimize to that. And you could imagine that they're just going to be paying by a basis point, every base, yeah, you improve it. I will pay x. I think that's an interesting idea. And maybe one that we can explore together.
[Auren Hoffman] You know, AWS, really kind of like pioneer on this, this data exchange, but you can imagine, its competitors like Azure and stuff, like also opening up a data marketplace. Like, could you how do you see it like playing out over time? Will they'll be like, multiple data marketplaces? Do you think there will be a marketplace of marketplaces? Like, how, how do you think it works out?
[Stephen Orban] Yeah, it's an interesting question. So Far be it for me to opine on what another company might or might not do. In the same space, I will say that we continually hear from customers that they want more data to make better decisions. So we got enough signal that suggested it was important for us to do. So I would assume that other companies who might look like us are also getting similar signal, I, you know, I can't, I can't predict that they will, what I can say is, you know, we're gonna keep listening to customers on what we've built, and what they want us to do with it. And not just with the data that we're offering through data exchange, but what they want to do with it. And all these services, you just mentioned, the use case around model optimization. And, you know, we're, we're pretty fortunate that we get to be involved in some of these conversations, which may not have anything to do directly with data exchange, it's more of an analytical tool that we need to, you know, build to better serve a particular use case. And, you know, we then either incubate those ideas in my team, or we talk about them with other service teams around AWS, who have the sort of the right portfolio of services for it to fit into. And, you know, we're gonna keep doing that forever.
[Auren Hoffman] All right, I think our thing our listeners who run data, businesses or consume data, I find this super, super interesting. Look, let me just wrap up a few personal questions. I'm always interested in the personal side. So we you wrote this book ahead in the cloud, it's got like, 220 reviews on Amazon, like, what like, my guess is like, 93% of our listeners want to write a book, like, what advice would you give to these people? Because you saw publish this right.
[Stephen Orban] But I did self-publish the book. Yeah. So the story behind the book, when I first started at AWS six and a half years ago, my role was go talk to technology executives from the world's largest companies, and help them understand the change that is required to modernize their it. It not just their infrastructure, but their operating model their people their approach to, you know, moving from plan, build, run, or ITIL, to DevOps, or whatever. Having done it before, as the CIO of Dow Jones. It's how I landed in that role. I decided to join AWS to help other big companies through some of the changes that I have all the scars to prove, leading all the mistakes I made to help other companies maybe not make them again. And when I went out and talk to customers, for years, I would learn a lot, they'd asked me lots of questions, they would say that their assertions or assumptions on something are x, and I was like, Oh, well, maybe you should think about x a little bit differently. And here's why. And here would be your approach. And so I started to write all these blogs. And the blogs were based on conversations I was having with customers where I felt, maybe there's enough of either misperception or or or meat on the bone or some idea that I could write a small piece I fast forward a couple years, and I had dozens of these blogs. And I had kind of planned with, okay, if I can write enough blog posts that are when you put together make a sensible narrative, maybe I can blog my way into a book, which is I understand that other people before me have done as well. So I kind of got to that point. And I thought I had enough content for a book I had written a series on the stages of adoption, as I call it, and seven best practices for enterprises moving to the cloud. And I was like, Okay, I'm going to write a book. I talked to a bunch of people who have written Adrian Cockcroft is what I would mention, he's also at AWS, he's a fairly well-known figure in the technology space, having spent a bunch of years at Netflix and sun, and elsewhere. And he's like, every page you write turns into 10. And it's way harder than you think it's gonna be. And I'm stubborn.
[Auren Hoffman] You’re kind of a more engineering background, not a journalist background. Did you hire someone to help you edit it?
[Stephen Orban] Yes and that was very helpful to organize my thoughts and structure certain things, but you know you only get one reputation in life. At the end of the day, I wanted to make sure it was my voice and thoughts coming through.
[Auren Hoffman] Are you writing it between 10pm and 1am every night?
[Stephen Orban] I thought because I was starting with all of these blogs, I off to such a good headstart and I was 80% of the way done. This is what Adrian was trying to tell me, no you’re not. And I didn’t listen because I’m stubborn and when people tell me something is hard, I want to do it more, and I went ahead and did it in a way that I thought would be a 100 hour exercise turned into a multi-thousand hour exercise.
[Auren Hoffman] Oh my gosh. Ok, I’m not writing a book then. That 93% just went to 23% in the last few sentences.
[Stephen Orban] It’s not for the faint of heart. And look you know, I am very happy with the way it has turned out. A lot of people who have read it have given me lots of kudos. Even so, it’s not perfect. The negative reviews that I do have say you can tell it’s a collection of blog posts. That’s outfavored by people who said they learned something so that’s why I am happy with the results.
[Auren Hoffman] Two more personal questions. You live in the Westchester area, I happen to have done an extremely rigorous study of the best places to eat near Mamaroneck. So the most controversial question I am going to ask you right now. What’s better -- Sal’s Pizza or Walter’s Hot Dogs?
[Stephen Orban] For me, that’s easy. Sal’s.
[Auren Hoffman] Alright, here we go. Sal’s right here. Sal’s Pizza. The best pizza in the world. Anyone who doesn’t say it’s the best pizza just doesn’t know what they’re talking about.
[Stephen Orban] I agree.
[Auren Hoffman] Ok, I am definitely with you. Walter’s is the best hot dog in the world. So if you want a good hot dog. But I just think pizza beats hot dog. Walter’s for sure is the best hot dog. And then Sal’s is the best pizza.
[Stephen Orban] So we moved to Larchmont Westchester 8 years ago. My wife brought Sal’s pizza home our second week here and she’s like you’re going to love this. It’s the best pizza. And we had moved from New York City, and of course I was like whatever.
[Auren Hoffman] Yeah, exactly.
[Stephen Orban] I ate the pizza, and I was like oh my god, this is the best pizza and we eat it once a week.
[Auren Hoffman] Alright, I think after this podcast, the net sales of Sal’s and Walter’s is going to go up at least 15% each. Ok, last question that we ask to all of our guests. What would you tell yourself from high school or college to save yourself time or money or emotional wellbeing.
[Stephen Orban] History can be a predictor of the future.
[Auren Hoffman] Oh. Let’s dive in. What do you mean?
[Stephen Orban] Yeah, so you said this a couple of minutes ago. I’m an engineer. When I was a kid, from a very early age, I was always interested in computers and I was good at science and math and I hated everything else. To the point where I was lucky I got out of high school and went to any kind of college. I went to a SUNY school of NY and today the same SUNY school wouldn’t have let me in with the grades I had. Largely because I ignored all of the social sciences, history and writing. Frankly, that’s why writing a book was even harder. It wasn’t something I focused on until my late 20s. And my wife Megan, I married a history teacher. And the little that I have tried to educate myself over the years, when I realized that was a regret, I learned a lot and a lot of those things are applicable to what I’m doing today, even in a tech environment. If I could say anything to myself and have a hope that I would listen, would be to actually pay attention in history class.
[Auren Hoffman] Awesome, this has been great. Stephen Orban thank you so much for coming on World of DaaS podcast. This has been so much. And congrats again on your huge promotion to run AWS MArketplace. Congrats on that as well.
[Stephen Orban] Thanks man. It was a pleasure.
Stephen Orban, GM of the AWS Marketplace, talks with World of DaaS host Auren Hoffman. Stephen previously served as the GM of the AWS Data Exchange and CIO of Dow Jones. He’s also the author of Ahead in the Cloud: Best Practices for Navigating the Future of Enterprise IT. Auren and Stephen cover the launch of AWS Data Exchange, Amazon’s role as an industry-wide innovator, the advantages of Amazon’s leadership principles, and what the future of data discovery could look like.
Hilary Mason, co-founder of Hidden Door and data scientist in residence at Accel Partners, talks with World of DaaS host Auren Hoffman. Hilary previously co-founded Fast Forward Labs, which was acquired by Cloudera, and served as the Chief Scientist at bit.ly. Auren and Hilary explore how data science has progressed in the past decade, the role of data science in an organization, and data ethics.
Auren talks to Jack Dangermond, CEO of Esri, about his philosophies for building and running the world’s most successful geospatial software company. The two also discuss how software can be used to store, represent, create, and share geographic knowledge for solving problems more holistically.