DigitalOcean | Interview with its CEO – Ben Uretsky
In New York (NY), we meet DigitalOcean CEO Ben Uretsky. He shares how he came up with the idea for his fast-growing company, how the current business model works, as well as provides some advice for young entrepreneurs.
The transcription of the interview is provided below.
Martin: Hi, today we are at New York City interviewing Digital Ocean. Ben, who are you and what do you do?
Ben: I am Ben Uretsky and I am CEO of Digital Ocean.
Martin: Great. When did you start this company and what is your background?
Ben: I started DigitalOcean in 2011 and my background; I have a Bachelor’s Degree here from Pace University in information systems which I basically got to show my dad that I can finish college but I have been working since I was 15 years old in high school. So my background is I thought myself Linux and networking and servers and then ultimately joined a bunch of Tech Companies starting out into the voice over IP space, and then transitioning to hosting, and then ultimately starting my own company.
Martin: Okay and in parallel you did study.
Ben: Oh, yes. I did study, I finished. So when I started working in high school, I graduated from there and then went to University and was able to actually start my first business at 21 before graduating from University.
Martin: Great. Can you tell us a little bit about some of the similarities of those business models because most entrepreneurs have some area which are very like, and they tweak around and have some different business model in that kind of area.
Ben: Yes, that’s definitely true. The business model that we use is a recurring subscription is really the heart of the engine. So the first company that I started in 2003 was called ServerStack and it managed hosting, just taking care of customer environment online. So let’s say you generate some revenue because you are a retailer, so e-commerce or you drive some sort of a subscription network maybe it’s video ads or advertising in general, in other words you are generating revenue online, you want to make sure that your website stays active. And so we monitored and took care of that entire environment, provided a 100% uptime SLA, took care of back up security, capacity planning, network expansion, just everything that falls within IT management and helped our customers grow their business. And so what was great about that business was that as you sign up that first customer, more often than not, they continued to grow their business, so each month, each account produces more revenue and the goal is to get as many more accounts as you can.
Martin: And how is the current business model of Digital Ocean working?
Ben: It’s similar. It’s interesting because we are a utility cloud provider. So customers can use as much or as little as they want. Our billing today works on hourly basis but the majority of our customers run what we call sticky infrastructure. So once they set it up, they won’t tear it down, it will continue to exist for months and essentially even for years. Although they have the luxury if the demand changes or they need to make a modification, they can pay by the hour and make those changes in real time. Nonetheless, each customer continues to spend more money every single month because they create a business which finds traction and attracts new users and so they need to grow the server foot print to keep up with demand. And so even though it’s not a traditional subscription model, based on that sticky usage, they wind up paying more every single month but that’s tied directly to their success and usage.
Martin: And did you transition DigitalOcean directly from your former company because the business model is quite similar but you are coming from managed hosting to cloud hosting?
Ben: Yeah. We had an opportunity to step back in 2011 and really think about where our first business was headed. And we had some concerns about dedicated servers, many other providers came out with virtualized solutions where you can purchase these systems online. Some of them sold the fraction of the server, so it seemed like price was substantially less but that’s also because you’ve got substantially less resources. Nonetheless, a consumer, the person who’s actually going to use those recourse, didn’t pay as much attention to the unit economics, they just thought, ‘Oh wow, I can spend $20 bucks a month, instead of two or three hundred.’ So in general customers really began to adopt cloud based servers and wanted to make sure that we stayed relevant within the market place. We had a good grasp on business operations as well as the technical sides, so the servers and the networks. Where the first business didn’t find traction was actually in its unique value proposition. So we were selling support like many other companies and it was hard to differentiate ourselves and stand out in that crowd of market.
So as we began to analyze what the new business would be all about. We started reading a lot of books on marketing, on positioning, there is actually a book literally called, ‘Positioning’. I think it’s Al Ries and Jack Trout if I remember correctly. It talks about finding a unique value proposition that fits within the customer’s mindset. So if you have to re educate them, it’ll be a very difficult proposition and you also can’t compete against any of the existing values or opinions that they have. So you have to try to find a way to squeeze into all of that and still stay relevant.
And so as we started thinking about building Digital Ocean the other concept that we really kept in mind is brand extension. And sometimes that works for example, perhaps you make a toothbrush and you decide to extend into the tooth paste business. It could work but it could also create some confusion. And what we see so many times in our industry is companies create brand extensions and so first they are dedicated server provider, then they become a managed service provider and then the cloud comes and around and all of a sound they offer virtually cloud offerings. But as the customer evaluates that front home page, they are confused, it’s like, ‘Can I buy this server or that server’? It’s a mixed message. So what we realize is, we could not extend our original business to provide cloud servers.
So we decided to start a company from the ground up with a single laser focus on providing the best cloud server experience that we could, that levered on nearly a decade of running the first business, going through a lot of growing pains in helping our customers achieve scale. So we took that operational experience and we created a product. And that’s another huge distinction between the first company which was a services based business and essentially there, I like to say you tell the customer, ‘Yes’ and you then you go home or you go back to your office and you try to figure out how to deliver that service. In a product based business it’s the exact opposite. You build a product that has certain characteristics and that’s all that you can actually deliver because product developing takes a much longer time to actually achieve that new product version.
So that really was the basis for Digital Oceans, all of these ideas floating around in our heads. We had access to couple hundred customers in our first company but we also did survey the market. And what we recognized is that no one was really focused on the developer. Everyone was focusing on larger scale businesses and the way that they thought about their companies was through technology lens. So they said, ‘Hey, we can deliver and build these features and this is what you guys will use’. And for a larger size business that actually works well because they understand their requirements, they have the necessary teams to manage that complexity and so access to this scalable infinite infrastructure is a great way for many companies to build their business. But if you take a look at the opposite end of that spectrum it’s an individual user, a developer who is just getting started, maybe they are learning a new language or they are pushing their application for the first time into a beta phase, trying to find that initial traction, create a start up that’s ultimately going to scale, that user was neglected. We decided to focus digital ocean on providing the simplest cloud infrastructure experience possible catered to the developer rather than to a scaling business.
Martin: Ben when somebody is thinking about ‘what type of server hosting I am using?’, you said that it’s quite hard to distinguish the unit economics, can you walk us through the unit economics for the different type of segments like dedicated, managed, cloud hosting?
Ben: Yes. So dedicated servers are fairly straight forward because you can a la carte pick and choose your processors, memory, storage and all of that gets bundled into a monthly price tag. And then typically you also pay extra for bandwidth and you have a lot of customization on dedicated server side. So anything that fits into the chassis you can probably get from a provider.
Once you go into a cloud environment, it becomes a little different. Most cloud providers bundle a specific set of resources together, so you will receive one or two CPUs with a couple of Giga bytes of memory and some storage space. Now, some providers charge separately for bandwidth. We actually include an allotment of network transferred bandwidth with that original plan as well, so that customers don’t have to worry about, how much will I actually pay at the end of the month. The majority of our users fit within the allocated bundle and so they know exactly how much they are going to pay in advance. And with cloud, the business unit economics have become much more granular. So we see companies nowadays that are doing not only hourly billing but even per minute and I think we are even starting to see some come out that are gauging by the second. So that’s great from environments where you have high volatility. A great example would be a news website, some event happens, everyone rushes to the website, you get 10x or 20x of traffic and then a day later you need to return back to that original amount. So in those cases the elasticity of cloud is a great fit for that use case.
Martin: Okay great. What experiments did you do in order to understand how best to reach your customers? So what marketing channels did you test, and what have been your hypothesis, and how did you test them?
Ben: The hypothesis for the most part we developed behind closed doors because we had almost a decade worth of experience working in this industry and understanding what our customers wanted. So in some regards it’s almost 10 years of continual customer development that gave us a really good foundation to work from. But as we created that initially hypothesis we then ran some very focused user groups. So we brought in one person at a time so that you don’t create a group think environment and sat them down behind an alpha version of our product and ask them to use it. And we were looking for two things:
- Can they successful launch a virtual server?
- What are the adjectives or how would they describe the experience of using our product?
And I think every single person successfully created a server which is not the case with all providers. Sometimes people get stuck in that creation process, so we hit simplicity there. And two is the experience, they were actually quite pleased with what they say. So very intuitive straight forward control panel language that isn’t confusing, keywords that resonate with their understanding of the market, and they also pointed out a bunch of mistakes or oversights on our part. Like you can only log in and sign up from the home page none of the other pages that we had provided those links. They helped us with a little bit of the user navigation to create a smoother flow. But that I think is actually very important. And we felt extremely challenged when we first said, ‘How are we going to get people to come into our office and actually provide this feedback?’Because you don’t want to take a stranger from the street who might be doing grocery shopping to evaluate a cloud server business. So we placed an Ad on Craigslist and decided to pay $100 for this 1 hour session and the criteria was that you would need to use a competitor service to prove that you are a potential customer for our product. And I think that that experiment thought us a lot about where our customer’s heads were at, but the fundamental framework was really developed over that decade working in the first company.
Martin: How did you evaluate whether the customer will use your service? How did you try to estimate the willingness to pay for example?
Ben: I don’t think we really estimated that, it was—
Martin: Just taking competitive prices, or so?
Ben: Yes. Although our pricing we actually developed in a very interesting way, we are a price leader in the industry today and we approached it bottoms up. So we calculated how much it would cost to deliver this basic unit of compute and we created a healthy margin with which we could run the business successfully. It just turned out that price point was substantially less than our competitors. And that’s a great question because why are we able to offer such a different price compared to the industry veterans?
- We do have a late comer advantage.
- We are extremely focused, so we only provide cloud servers.
That means we don’t have to worry about legacy business or competing interests when we are able to really optimize the entire operation to deliver that as cost efficiently as possible, plus we have the industry of experience from the first business to understand the right relationships that we should build with vendors and the proper financing terms to makes all of this happen, so it was an unfair advantage.
Martin: When I am look at this business model from the outside, I see mainly three steps:
- the one customer acquisition,
- then this type of technology which means hosting, infrastructure, etc., and
- then the last part this kind of customer service.
In what parts do you think is your completive advantage?
Ben: Yes, I mean our competitive advantage stems from the actual customer. Most infrastructure providers care more about the infrastructure more than the customer. So you see them leading with the “what“, like we sell this amount of resources with these features at this price. We like to lead with “why“. Why are we doing this is to create a simple infrastructure experience that developers love and who do we focus on, we focus on people rather than the technology. Everyone has access to virtualization, to the same silicon that is used to build the microprocessor, RAM, storage, network, we are all plugged into the same internet. Those are the commodity components. But what we try to do is create an experience based around the actual needs of a person to empower them with these tools so that they can build and easily manage their infrastructure and dedicate it and really focus their attention on the application. Because that’s what creates their business, it’s not the server underneath. It’s like if you turn on the light switch, you expect the electricity to work so the only time you actually think about the infrastructure is when it doesn’t work. We try to really keep that analogy. We want to get out of the way our customers so that they don’t have to think about us and can actually focus on their business building heir applications, getting user tractions and growing their companies on top our cloud.
Martin: Imagine a young developer is currently hosting a Bluehost or somewhere else. Now you want to target this kind of person in using your service. How do you minimize the switching cost, is there some kind of automated process where he can just say I’ve got a blue host account or Go Daddy account, whatever and then just click this button, you take care of the rest, etc.?
Ben: Well we are working on the “easy button”, it’s not ready just yet. What we do is slightly different, we lead with the education approach. So we have written almost a thousand tutorials to date that talks about how to use the latest and best open source technology, but also educate users on best practices in the industry. So a portion of topics go towards migration, so how do you take a database and move it from one server to another? How do you migrate a web server? Or how do you actually migrate an entire configuration across multiple machines. And a savvy developer will recognize, ‘Hey I can actually use this information to migrate my service from Bluehost or any other host to DigitalOcean’ what’s great is, it’s not proprietary, it’s not locked information it will be applicable to any provider on any server. And that’s really something that we see lacking in the industry today.
We have some great community sites like Stack Overflow but they limit the conversation to, here is a problem that I encountered and the community will provide solutions or answers but there is no open ended conversation around, if I want to build a highly available low balanced, fault tolerant website, what are the technologies? How do I go about actually building this environment and then maintaining it? So you would have to dive deep into Google, find a couple of blog posts and use the spare piece of information to try to glue this together. And that’s what we are trying to create with our DigitalOcean Community, is to be seen as the place where developers can go to conversate about these topics, to accelerate and improve their infrastructure, whether they are a host with us or not.
Martin: And by doing this you do two great things:
- one thing, you generate trust between your potential customers, customers and you and
- second thing is I assume you lower you customer acquisition cost because there is some kind of network effect and SEO as well.
Ben: Yes, absolutely. We are ranking in the top 5 or 6 results across many different terms on Google. But what’s great about it is it’s validation from our customers because they succeed at deploying those applications and installing that software, it is also validation—what does Google try to do? They try to surface the best content and we fall into the top 5 or 6 because we do invest a lot of dollars into making sure that the content that we produce is top notch. That it is both technically accurate but then it’s also written in a way that, someone new to the technology perhaps they are coming from a shared environment where they did not need to install all these different software packages, they can actually make the switch relatively easily based on the information they find in our community.
But yes, it also attracts a lot of people to our website that learn about DigitalOcean, Ultimately go on to try it but the community is just one part to our success. The real reason why we have this phenomenal rate of growth is when customers use DigitalOcean for the first time, they truly encounter a unique experience that they can’t get anywhere else. And it’s exactly what they expect. It’s really quick and easy to use. And ultimately it leaves them with the feeling of satisfaction that they want to go and share it with someone else, whether it’s a friend, another developer or even just write a blog post about their experience using DigitalOcean. And that speaks to the original mission that we set out on, it’s to delight our customers, to simplify infrastructure because it’s not complex, at the same time it is growing more complex everyday. And that’s where we really want to use the community, as new projects come out, 4 west being a great example, a new product release that we just did last month. We are able to talk about it, educate our user base, and teach the best practices on how to deploy that in the cloud.
Martin: When you are looking at your product and every entrepreneur that has some kind of vision and says ‘Okay’ this is where I want to go’. What do you think is still missing in your product that you say, ‘I want to do that!’
Ben: Quite a few features are not present in the product just yet. So the challenge that we have is, we decided to write the software in house and that has created the experience that we have been able to deliver to our customers. Without writing it and owning the code there is just no way that we would have been able to deliver the intuitive interphase that we have. The other side of that sword is that we are responsible to develop the rest of the features as well. So I know our customers desperately want additional storage capacity. So it could be something like an object store, especially if you have a user generated content so someone that uploads pictures or media. You need a place to store that data and we provide only a limited amount with the droplet which is our virtual machine today. So there is no easy way to scale out storage. And the other component, I think that a lot of customers are asking for is more advanced network services related to high availability. So as you develop a production website and want to ensure as close to 100% uptime as possible, taking advantage of low balance and to create full tolerance, or truly private and secure networking with access list and firewall policies. Those are the features that our customers would love to see us develop as quickly as we can. And a lot of where our investment dollars are going is into the engineering team and the efforts behind pushing those features to market.
Martin: Ben, let’s talk about the market development and the hosting segment. Imagine some kind of matrix where you have on one side the small companies and the larger companies. You are one of I would say the smaller companies, then you have something like Go Daddy or so on the larger spectrum. And then on the other hand you have the different kinds of hosting products that we talked about before like managed hosting, cloud, etc. What kind of trends can you identify in this? Is there some kind of migration from a lot of hosting users from the big ones to the small ones or the other way around? Is the hosting market in general growing in terms of total sales or is it more constant and just migration between the different services, what trends can you identify?
Ben: Well, I think definitely the market is growing every single year, more users are coming online, more businesses are being built online and you might start out and host it from your apartment but as you gain revenue you quickly realize that’s not reliable. So you look to a data centre so that your application can live in a true 24/7 uptime environment. So there is no doubt that the market is growing.
I think that the cloud providers probably represent this—one of the newer trends that we are starting to see is: before 2010 we may have only had one or two people in the space, today as we look around we have nearly 10 cloud providers that are trying to compete for business. And the majority of them are not able to differentiate clearly and I think that customers at that point then begin to look at things like price and convenience. The other trend that we are also seeing is the tools that are becoming available are actually starting to become much more complex, that complexity is really hitting away. So you have these frameworks for how to deploy larger applications so things like Docker. Platforms that are being open sourced so how do you run a Heroku inside of your own environment that allows you a quick and easy way to deploy a reels application.
And I think in general we are starting to see that users care less and less about infrastructure and servers and more about the application that lies on top of it. So how can I really deliver my application as quickly as possible to the market and then ultimately maintain it with the least amount of resources and management that goes into it. So that’s certainly a trend that we picked up on and are trying hard to stay ahead of. You can almost phrase it like: ‘By using DigitalOcean, you can delay the need for people that work in the developer operations segment, the SREs to ensure that your infrastructure is online. So we pour a lot of dollars into building a highly redundant environment from the start. Other cloud providers take a slightly different approach and say, these instances are ephemeral and they are here today, they could be gone tomorrow, so you really have to architect and application that can withstand a high level of failure. And we are not immune to issues and so we would always encourage customers to build availability into their applications, but we also have a better basis from which for them to get started. So in the early days when they are still relatively small they don’t experience that issue as much, obviously when you scale to thousands of nodes failure rates, even if it stays at relatively small percentage still creates a lot of issues.
So I think that’s another trend in general is that the software is finally beginning to catch up to these much larger distributing environments. The database has always been one of the biggest issues there with running a huge monolithic application on a single server that you scale horizontally. First you start out with two CPUs and then your business grows and you need a new chassis with 4 CPUs, and then it gets tricky after that. And some companies have actually been forced to re-write their application for months or a year to get out of that dependency and now what we are seeing is distribution and availability built into the software from the start so. NoSQL is a great example of that and many distributed Databases that have come as a result of it which inherently are a great fit for cloud providers because of the large number of nodes over disparate machines and resiliency to failure.
The last thing on the trends, I think as a result of, I don’t want to take too much credit here, but DigitalOcean coming in with a great price point, an easy to use interphase and the education. I think we have enabled a whole new set of users to come online, so people from countries that are really developing nations and are beginning just to have their people get on the internet from Asia, within Europe, south America as well, are able to take advantage of DigitalOcean because of it’ price point and ease of use and finally start building something online. Whereas in the past they would need to wait much longer, develop on their home computer or laptop and perhaps never even take the step to publish their application live. And I am very proud of the tens of thousands people worldwide that have access to the server resources as a result of our company. We have also enabled thousands of students whether it’s through hackathons or—we even sponsor our high school that I attended here in New York City so we provide them with a ton of resources every year for students to do their final term project and we do that across the nation and even internationally. These education facilities can actually request resources and sponsorship from DigitalOcean. We want to enable the next generation of developers and we don’t think there is any better way than getting your hands dirty, partially that’s how I learned, so I would love to see other students start out the same way.
ADVICE TO ENTREPRENEURS
Martin: Ben over the last 15 years can you share what things have you learnt about how to start a company and what’s better not to do?
Ben: Yes. I chased lucrative opportunities, so I started when I was 15. My first job was in a appliance store and I worked there for two weeks before I quit. But at that point I had saved enough money that I could buy a server online and from that server I started to generate revenue online, initially with advertising then obviously in the infrastructure space and today in the cloud space. So I think these opportunities present themselves and that’s what I love about New York City is the exposure to different industries. You should follow your passion and surround yourself with people, join a company or business that you are really interested in and let the business create itself. An opportunity will strike you, you will recognize that this is something that you feel passionate about and chase that dream. But I would say don’t quit your day job. How can you use a lean model to interate quickly? Try to build a business on the side. When that responsibility becomes too great for you to manage as side project you know you have something that’s worth investing into and I would recommend making the transition at that time. I have had a number of employees though out the years look up us as role models and say, ‘Wow, it’s fantastic that you are able to start this business, I am going to quit and I am going to go ahead and start my own company’. And I always tell them the same thing, I’m like, ‘Are you going to quit and disappear into a vacuum to try to create something or you are leaving because you have an idea that you want to pursue?’. And unfortunately so far those businesses haven’t survived. So I would say surround yourself with the right environment, learn from it, see what you become passion about.
Martin: Great. Thank you very much for your time, Ben.
Ben: Yes. Definitely.
Martin: And the next time you think about starting a company, you should think about whether you need some external funding and if that’s the case maybe then you should quit the company and then start your business, or if you don’t need very much money and you don’t need to talk to clients all the time but having an online business then you can think about bootstrapping in parallel to your employment. Thank you very much.
Ben: Yes. Thank you, nice to meet you.