Sumo Logic | Interview with its CTO & Co-Founder – Christian Beedgen
In Redwood City (CA), we meet CTO and Co-Founder of Sumo Logic, Christian Beedgen. Christian talks about his story how he came up with the idea and founded Sumo Logic, how the current business model works, as well as he provides some advice for young entrepreneurs.
Martin: Hi, today we are in Redwood City in the Sumo Logic office. Hi Christian. Who are you and what do you do?
Christian: I am the CTO and Co-founder of Sumo Logic. And we’ve been building this company for the last five and a half years. We started in early 2010. We now have our own office here in the beautiful Redwood City. Having two full floors full of people, it’s very exciting. We’ve got the sales crowd, the engineering crowd, you know, all the in-between folks that you need to make a great company.
Martin: Christian, you have a super interesting life story. So, starting from Germany, how did you come to the U.S. and how did you move from start-ups to your own start-up?
Christian: Yes. I came to the U.S. in an airplane that was flying into Seattle in late ’98 and I was extremely anxious and I really wanted to turn around and go back. So, why was that going on, and why was I going to Seattle? Well, I was a student at Fachhochschule Brandenburg and so if you go to a school of applied sciences, in Germany anyways, in the 7th semester they’ll kick you out and you need to basically do an internship. So, I ended up, and this is really kind of a convoluted and long story, and through a lot of hoops, etc., I ended up with a bunch of guys who got acquired by Amazon in order to sort of bootstrap Amazon.de that was called telebuch.de back then.
And the acquisition had just happened and, as an intern, they basically just said, ‘Hey, why don’t you come with us to Seattle, and we’ll build something, of some sort?’ And so I was actually at Amazon then in late ’98, in downtown Seattle. They didn’t even have proper offices yet. They were like spread all across the city. I remember, I was like walking down 1st or 2nd street with literally compact servers underarm. It was ridiculous.
So that was basically a six month thing, where it was super interesting and that was the first time I programmed, I guess, as a quote unquote professional, having done some stuff in school before. So, anyways, from there we helped, me and another intern, who is actually also, he’s not here today, otherwise I would introduce him. His name is Stephan he’s another German guy. And I’ve been working with him ever since. He’s our chief architect here. So, basically, we helped this founder guy to work on one of his ideas which was sort of a German to English translator and vocabulary service. Fast forward, that really didn’t go very well. There were some differences between people.
But we had met this lady who came over from Miami to Berlin to help out, because now this was actually a two continent story and but then people kind of fell apart, as what sometimes goes when you mix too many people too quickly and all of this pressure and what not. So, she had this idea that people should just be able to store their files on the internet, you know? Like upload them to the web server. People really didn’t have laptops yet. And, certainly, they didn’t have USB sticks, right? So you would go from another café, to another café, and you’d get home or to the office and sharing your stuff, even with yourself wasn’t super easy. So, in other words, essentially, what you know today is Dropbox.
So, ‘It’s a great idea,’ we thought, ‘let’s do it.’ And we started a company together with her in Florida, in Miami actually. She was responsible for running things and bringing the money and Stephan and I started building the product and engineering and so forth. But, we had no idea what we were doing, and we ran it into the ground. Timing was not great, it was like in 2000, but that’s not really an excuse. It was really us.
We then ended up going around, sort of making the rounds to a bunch of PCs, both on the east coast and the west coast. Of course we came to San Francisco, Silicon Valley, nobody wanted to give us any money, but my resume was kind of was left behind. There was this VC who ended up at the same time who actually had just incubated this company called ArcSight. This was basically in early 2001. Somebody got my resume, they were looking for technical people and they interviewed me and flew me out here and I got to show up and was like ‘This is very cool.’
To this day, I remember interviewing with them, and they completely blew my mind. They were, by far, the smartest people I’d ever met. And I was like: ‘Oh wow. There’s something cool going on here. This is definitely a cool place.’ So they paid for the relocation, me and another German guy, so we threw all our stuff in a big moving truck and we hoped on a plane. So then, in early 2001, I was starting to work at ArcSight in Silicon Valley, living in San Jose, and so I’ve been here ever since.
Martin: And how did this story unfold until you really started Sumo Logic? How did you come up with this idea?
Christian: You know, if you are working in a particular field, and in my case, ArcSight was a company that was dealing with a lot of security information management so ArcSight basically did log management for security. And so, I was actually there for nine years. The company was successful, it was actually a really good story. I was an early employee and there was lots of opportunities for growth and I became the chief architect and all these kinds of things.
And so you kind of get steeped in the domain to some degree, right? And then at some point you start realizing that ‘Hey, I might not know a whole lot, but when it comes to say in my particular case, log management, there are probably not necessarily another million people out there that know as much as I do.’ And I’ve observed the goods and bads about how to do this product for nine years. So did a bunch of other folks at ArcSight who have been working with and so Kumar, who is now my Co-Founder here at Sumo and I, basically started putting together this idea, that there’s a lot of good things, and there’s a lot of bad things, ultimately nine years in, about how we approached the product and stuff at ArcSight. And if we can improve on the things that didn’t really work so well, given there’s a lot of good things and there’s a huge target market and so forth, and there’s expansion capability, maybe we have the hypothesis behind a next generation company. And that was basically it.
We basically were trying to improve on the delivery model. ArcSight was still enterprise software, so you would give the software to the customer and it just turns into a shit show. As a developer or as a vendor you lose control, and without control, you know, helping a customer and making the system really fly is hard. Unfortunately, I felt that it also sort of ended up kind of stopping the innovative capabilities of the company because, with enterprise software, you know once software has left the building, it will never die, right? And there will always be other people that still use it and they want fixes and patches and then suddenly you spend all of your time with every new version it gets worse because the old versions never die and all of your engineering and time that you would use to build new cool stuff gets kind of soaked up in, you know, backporting and supporting existing customers.
So, we wanted to fix that, and so that’s why we said, ‘Hey, this has to be a service,’ So the cloud stuff was just coming up. Of course people had done SaaS before, that’s clearly not our idea. We saw the applicability for it for our field, I guess. As usual when you do stuff like that other people say: ‘Oh, you can’t possibly do that. No way you can do it.’ And I’m like- ‘Well, I think we can do it.’ You have to be a little bit stubborn, I guess. And you’ve just got to be high on your own supply, I guess. A certain amount of arrogance is warranted. You’ve got to really believe into it.
And so we ended up having a very good relationship with a VC from Greylock and his name is Asheem Chandna and he helped us along tremendously because was really the only person we had talked to that had sort of a similar idea that came to him based on his understanding of the market and the way that things were going. Most of the other VCs didn’t even know what log management is. And so, they ended up funding us in early 2010. We did a proper series A, we raised $5 million dollars. And, essentially, overnight, more or less, put the company on the map. And we started going through all the things that you needed to do as a company.
Martin: Christian, can you please describe what was it like the first 3-12 months? So what have you been focusing on and when did you talk to the investors?
Christian: We basically went through a little bit of sort of just getting to know people. And when I decided to quit ArcSight, it was basically at the end of 2009, Kumar, who was at Mint.com, and Mint.com just got acquired into Intuit and he’s really not a big company guy so he wanted to get out of there. So we basically were both sitting there with nothing to do and we started really gelling on the idea and ramping up our relationships, talking to VCs, going to all the various places in Palo Alto where you have to go to meet VCs.
But through the whole things, Asheem really stuck out as the guy who was the only one who really understood what we were all about. And then, basically, give or take a 3 months period to Series A. So we got very lucky there because we really believed in the idea and who felt that we were equipped to execute on this, at least as a kernel of a company, considering both of our histories (Kumar was there for the same amount of time). Just basically domain experts, I guess. And so, I think this is probably not a very classic story. Because sometimes you have to go out and convince people and ultimately VCs make decisions in very interesting ways. I understand this better today.
If you have somebody who almost felt like: ‘Hey, this company should exist.’ And if you show up and you basically have to say ‘my idea’ and you offer yourself because VCs can’t do the company themselves, there are always more people involved. So in many ways, the stars aligned really well for us. We got very lucky, I think, and we had a great mentorship and all of that. So that was the story.
It happened pretty quickly, actually, we didn’t have anything built. We had no product. We barely had slides. We got a lot of mentoring and they had to put the slides together and from Asheem, and he ran us through the partnership and made sure we check all the boxes. And ultimately here as a partner meeting. I think it was the last week of March on a Monday and we went there and I think we didn’t totally blow it. We didn’t totally blow it up. We were sitting at a beach and having some shitty lunch and we got a call they said, ‘Yes. We all agreed to it.’ It’s pretty much like a rocket, kind of. It’s like being strapped to a rocket, really.
BUSINESS MODEL OF SUMO LOGIC
Martin: Christian, now let’s talk about the business model of Sumo Logic. So, who are, basically, your customer segments and what type of value proposition are you delivering to them?
Christian: In many ways, the people that we’re selling to are working in IT, in general. So they might be IT operations folks. They might be developers. They might be this new kind of unicorn that’s called DevOps. And they might also be people who worry about information security both from the compliance angle, as well as, to some degree, from the hunting angle, like security analytics and those kinds of things. So product really takes logs, which is the fancy term for what is machine data, but it’s really the stuff that gets dumped on your disk and is over written and you lose it. The insight there is that this is very rich information and if you can find a good way to process it, it will add a lot of visibility.
And so our main audience, it really sticks with folks who have to run applications. And that includes the entire stack. It’s not just the application code. These things all plugged all together, from they write some codes themselves and there are some libraries and frameworks. Sometimes it goes all the way down to the machine often, of course, people today do it in a cloud] and they don’t really know what a machine is anymore, which is fine too, I haven’t seen one in 10 years.
There’s a lot of diagnostic information and random information and metrics that comes out of that, that you can choose to ignore, or you can choose to actually make it work for you. And Sumo provides a way to put all of this stuff that you already have to work. So, folks use it when things blow up. It’s a very easy way to go to one single data class to figure out, as a troubleshooting exercise what’s the root cause and restore service.
Of course, we would also like to help our customers, not always waiting for shit to blow up, but get more proactive. So there’s a lot of capabilities in the product that can take the logs, which are not really well structured, and transform them into something that you can ultimately aggregate and usually trend. You can wrap a learning around it and so hopefully you catch the problem next time before it happens, or at least a little bit. You know, all of these things.
And then to go all the way to the proactive side, we also, and this is another tactic we’re taking as a company, apart from being a service which is quite differentiated still, we really wanted to take… So when you have a system like this, in many ways, you are building a system that is sitting there waiting for your user to ask you a question. And you need a really slick interface to get that question expressed through the tool and then the tool is supposed to come back in the fastest time possible with some sort of answer or some sort of supporting evidence that lets you figure out what actually happened. And that’s great, because people are good at asking very good questions, but at the same time there’s a lot of data that’s there that probably never gets touched by your questions.
And there are a lot of questions that you can derive from that data, that’s otherwise sitting there, by running things like machine learning and statistical inference. In many ways, learning aspect like big data aspect. And sometimes that’s as simple as just counting, and sometimes that includes a rather complicated and algorithmic stuff, etc.
And so we provide a bunch of pretty differentiated functionality there, all the way to automatic detection which can identify relatively complex anomalies events across multiple types of logs, etc. and can guide you, ideally, into a place where it either already gives you all the clues you need for an answer or at least points you into the direction. Causality is not something that machines can do quite yet, apparently until like end of 2030s or something, we’ll be there. That’s probably either going to be good or really bad. But humans are very good at adding causality and the machines are very good, however, at correlating large amounts of data that humans are notoriously bad at. And so we’re basically trying to put the two together in terms of just making this really slick and having a great tool for answering questions, we also try to fuse or augment the human intellect with what machines can do.
Martin: How did you acquire the first customers? So when you got the 5 million dollars, started the product development, and when did you start acquiring and looking for customers, and how did you do that?
Christian: Yes. That’s a good question. So we started putting the team together, in engineering, and so forth. I think there must have been around, June 2010. We came out of self and we did the go to market thing in a launch of the company in January 2012. So, let’s say, that’s basically 18 months. That’s the time that it took to build the system. The first version, I guess, of the system. There’s no versions of the system anymore. It’s all this living thing that keeps evolving.
I think we had our first beta customers nine months in when we had some things strung together that, sort of, barely worked. How did we find the beta customers? There was a very important person that got added to the team, right after the funding and that’s our VP of Product- Bruno. He was the business guy. Both me and Kumar are the technical guys. The funding was like- ‘Yes. It’s okay. We’ll find you a business guy.’ So we went out and we found Bruno and he’s awesome. He kind of comes from the space as well. He has the background. And Bruno did a lot of customer discovery. He has a MBA background so he’s much more of a rolodex than either me or Kumar and we did 80 or so interviews with people. We played the idea off of them and waited for them to tell us that we’re fucking crazy. And most of them didn’t, actually. The obvious objections came up: ‘What do you do with all the data in the cloud? How do you make it secure?’ etc. and we started developing more and more reasonable answers for that. And out of that discussion eventually, through some connection, came somebody who was willing to actually be a guinea pig and then, of course, we kind of swamped him. So we spent quite a bit of time on all of that and then it kind of went from there. But it was basically relationship based.
And the lesson there is, and I guess this is pretty obvious, at least in hindsight now to me, no matter how good you are as an engineer, or you might think you are, or probably you are, who knows, that doesn’t necessarily mean that you’re good a running a company. You need more people. Especially if you’re doing it the first time around. I think if you’re on your third start-up or so, and you’ve observed, and absorbed, and learned, you probably can probably impersonate certain roles. But, especially if you are a first time founder, or maybe even if you just don’t give a shit about this other stuff and you just want to build the best possible system and you don’t really want to talk to 80 people on the phone, because that’s not really fun, maybe. Then you need to find people that can come and join the team and become part of the whole, that can actually do all of these functions. It turns out that even people who can write code, can actually be pretty decent. So that’s kind of what we learned.
So Bruno did a fantastic job there and a lot of credit really goes to him for getting us into the beta phase and all these things. And we’ve gotten questions before about like: ‘So, 18 months seems like a long time.’ And it’s like-‘I don’t really know.’ This is a big heavy-lifting kind of system. We are now at a point where we’re processing 50 terabytes of new data per day. We have many, many petabytes of data under management and this is a really infrastructure heavy data processing play. We have built everything in the AWS and of course there is infrastructure that gets taken care of. So at least we don’t have to worry about power & pipe. But still, this is a large distributed system, and so it takes a while.
So I think, in retrospect, that 18 months is not that bad. Of course, today you think that, depending on what you want to do, you can launch it in three weeks and all that kind of stuff an MVP there. But we knew exactly what we wanted to build, there was an existing market, we knew, basically, we had a very strong hypothesis we started with. There was no MVP business going on. And this might be our arrogance. I don’t know, the company might be in better shape today if we had done more experimentation. Or not, I don’t know. But it’s in good shape right now.
Our sense was that it was pretty clear what we had to do and we set out to do that and it took 18 months and it has never stopped since. And this is the trade off that you make. When you ship software, it’s a false trade off actually. Because when you ship software you think: ‘Well, now I’m done with it,’ when, in reality, you never are. As a service, now, you have the control, but you also have the responsibility for running it. Which is still a positive sum game and I’ll do it again. But it’s much harder to duck. If your shit is down, it’s down. And that, in many ways, makes us very similar to all of our customers, and that’s actually good. And at the same time we can also use the product for dog fooding which is also almost perfect.
Martin: We’re heard some different opinions of people. Some people say, ‘Do a MVP and in one month, and then ship it.’ Some people say, ‘Okay. Work on it for 18 months, and then ship it.’ From my point of view, in the end it boils down to, can you differentiate based on what you’re delivering to the market and if you can differentiate through some other kinds of solutions out there in one month, ship it. Of course, if it takes 18 months to be like 5x, 10x better, then you should invest 18 months.
ADVICE TO ENTREPRENEURS FROM CHRISTIAN BEEDGEN
Martin: What type of learnings did you generate over the last years at Sumo Logic and other companies before that you can share with other first time entrepreneurs? Some dos, some don’ts?
Christian: So one thing that I learned was, you don’t have to think that you’re an entrepreneur and you don’t necessarily have to wear a t-shirt that says, ‘fund me’ And you do not necessarily have to be on Angel List all day. That’s one way, but especially if you have established a little bit of a record in terms of your work history, etc. And, even more so, especially if you have managed to be, and this was by luck, even in my case. If you were actually part of something that was successful, and if you were actually part of that and people will reference you like that, you don’t have to be the CEO, but you were an integral part of making this company successful to the best of your abilities, then, from the investor’s perspective, that counts an awful lot. At least a lot of them. Maybe others are looking differently at it.
My point is that Kumar and I had established a reputation without even knowing that we had one, if that makes any sense. And when we got the VC connections, the right VCs took us seriously in a way that totally blew my mind. Because I don’t have a personal brand. I don’t have a blog. I’m not on Instagram, or something like that. And that works for some people, but that’s just not who I am. But that doesn’t matter. If you actually can talk intelligently and if you have a reasonable reputation, especially if it’s attached to a successful company or if it was a very heroic unsuccessful company, because that sometimes also happens, but if you’re real. And that’s really what they’re looking for. At least I think that’s what they’re looking for. That’s what worked for us on our part. Because we know what we know and we know there are things that we don’t know. We don’t know which those things are. But we don’t run around. We were basically pitching something that was a logical extension of what we had done before. And we had this moment when we were sitting there and we were like: ‘Maybe we should build a mobile photo sharing app.’ and then we decided not to do that and that was probably a mistake. I’m just kidding. I don’t think people would have taken us seriously, to the same degree that they took us seriously because we pitched something that was so clearly coming from personal experience. So that blew my mind, actually.
The other thing that I’ll say is that when it comes to VCs, of course there is a huge amount of information asymmetry, etc., but, you have power too. Not from the perspective that there has to be a fight, but don’t underestimate that there’s a lot of money to go around and the money needs to be put into people that do work. And that gives you power because if you can present yourself as someone who can take that money and do something reasonable with it, chances are they’re going to give it to you because it’s still better than the money sitting there and not doing anything because they still need to return funds, etc. That applies to anybody who has expertise in a certain area and who might not even be aware how much that counts. So, for me, that was very interesting.
The amount of how important your reputation is, and how badly you can mess that up if you’re basically being not nice to everybody, and people will remember that. And the VCs, and the good Vcs in particular, they will make a ton of reference calls. And they will find everybody. They’ll first call your references, but the real work goes into finding the people that you’ve worked with who have not given you a reference, for whatever reason. And getting that 2nd degree or 3rd degree kind of opinion matters a lot. So generally being decent is actually not such a bad idea. I guess that’s one way to do it. You can also be a total d… to everybody. But, the way that I saw it was like: ‘Your reputation really matters.’
Martin: Christian, thank you so much for your time and for your insights. Next time if you’re thinking about starting a company and you don’t have a clue which problem you want to solve, maybe it’s a good idea to build a reputation and knowledge at some other cool innovative companies. Learn from that, get into some specific area where you have awesome knowledge, and then maybe you’ll find a bigger problem in this area and start a company then.
Christian: Thanks man.
In this article, we discover the topic of secondary market research. We start with 1) an …