The SparkPost Story: George Schlossnagle

In this episode, we welcome George Schlossnagle, a mathematician turned tech entrepreneur, who shares his fascinating journey from academia to co-founding one of the world’s leading email infrastructure platforms. 

George discusses the challenges of shifting from theoretical mathematics to building a scalable business in the internet age, the trials of managing email infrastructure, and the pivotal decisions that led to the creation of SparkPost. 

He also delves into the impacts of transitioning from on-premises software to a cloud-based service and the intricacies involved in that massive operational shift. 

Tune in for an insightful discussion on entrepreneurship, technological evolution, and the balance between personal life and professional ambitions in the tech industry.

Listen here:

Watch here:

Read the transcript:

[00:00:00] George: I think anytime in a business, when you build a product and people love the product that they build and sell a product and love the product that they sell, it’s hard to change your mindset around like I. What it does, what it’s valuable for, and to fight that urge to say like, oh, they want to do this.

[00:00:18] George: Yeah, we can, we can do that. And it’s like, we don’t want to do that. Like we could do that, but it’s, we won’t be able to build a scalable business if we engage in that type of behavior.

[00:00:35] Ken: George Snaggle, great to have you on the show. You’ve got this fascinating background in mathematics. Uh, you have a math degree from Johns Hopkins, uh, a master’s from Brown, and then you went on to revolutionize email infrastructure. Um, tell me how does a mathematician end up building one of the world’s biggest email platforms?

[00:00:59] George: Well, um. It all starts with a love story. Um, so, uh, you, you know, when I, when I was in graduate school, I, I went to graduate school for math. Uh, I was intending to do a PhD in, in mathematics and I, I wanted to be an academic, uh, ’cause it’s fun. It’s cool. I, I really enjoyed the, the subject. Over the course of time, uh, and especially, so I took a hiatus from that and was in the, was in the Peace Corps and, um, while I was there, did a sort of, a lot of soul searching around, um, you, you know, what I wanted to do and.

[00:01:42] George: So I had two, like, two conflicting goals. So one was, you know, I wanted, I wanted to be an academic, um, uh, and do theoretical math. And then I wanted the woman who would be my wife to, to, to agree to marry me. Um, and. I think one of the big challenges in academia, at least if, if you’re serious in it, like in and seriously pursuing research, is you have very little control over where you end up.

[00:02:12] George: Hmm. Um, because, you know, you, you end up in these like hyperspecialized fields and if there’s an opportunity to go and work with, you know, one of the seven other people in the world who do this specific thing that you do. You really have to, to, to do that. So you, you have very low, you, you know, as well as like, you know, if you have a, a strong tenure offer for some from somewhere, um, or like a strong postdoc from somewhere, you, you really need to, to do that, to be successful in that field.

[00:02:43] George: And I, I was pretty convinced that like my, my relationship would not survive that and. So ultimately I, I chose, uh, I chose, uh, love over, over math and Wow. Uh, left, left graduate school, and at least at that time, you, you know, I, I think kind of like the three things that you, or maybe four things that you could do as a, as, as somebody with, you know, a couple advanced math degrees were be an actuary, which.

[00:03:16] George: Definitely didn’t wanna do, uh, go work in the defense industry, which also wasn’t super excited to do go work in finance, uh, as like a quant. Um, and I. That was interesting, but like, also like, not really what I wanted to do. Uh, you know, you interview at these places and you meet people, you meet your future coworkers and you know, there’s some people that you like and some people that you don’t.

[00:03:44] George: And, and then the fourth thing was to go work in what at the time was, uh, kind of the emerging field of, you know, software on the internet. Mm-hmm. Um, and. Met some really nice people that I, that I really [00:04:00] liked. There were exciting problems. I, I like, I like exciting, challenging, uh, problems. And so that, that was sort of what stuck the, the, the hook in that.

[00:04:09] George: And I, I moved to New York City to work, uh, at a company on Silicon Alley. And, um, that. Brought me close to where my wife was, was, was living at the time. And, uh, she wasn’t my wife at that point, but, um, but my master plan ended up succeeding. Uh, and yeah, that’s, that’s sort of how I got into technology. And then, you know, my, my brother.

[00:04:33] George: Was also, uh, working in technology. So he had sort of his own track where he was doing a PhD in computer science and doing some consulting work on the, on the side. And then, you know, eventually ended up wanting to prioritize doing the, working over the, over the studying. And there was a opportunity. To join him, uh, and help him, help him grow that business.

[00:05:01] George: So I did that as a side gig for, uh, a couple years as like employ, you know, 1.5. And then actually after nine 11, um, we had an opportunity to, to sort of really grow the business. We had a consulting customer who parted ways with one of their other contractors. There was a big opportunity to like grow our relationship with them and.

[00:05:25] George: It was the right opportunity. So I, I, I, you know, left what I was doing in, in New York and, and moved back to Maryland to help him really grow that business, which, which became Omni, which was omni TI. Mm-hmm. Um, and. You know, one, one of the, one of one of our consulting customers, our, our big consulting customer that kind of spurred that, uh, growth.

[00:05:50] George: Um, a amongst other things, uh, was probably definitely one of the top five, uh, B2C senders in the US at, at that time. And we were responsible for, for managing, managing their email. Um, which, uh, which was run on, uh, we were running on exim. Um. Yeah, had all sorts of challenges with that, which ultimately led us to, you know, look at how can we solve this problem.

[00:06:21] George: Uh, and both like improve the quality of experience for them, but also like, uh, quality of our own lives. Because you, you know, we were spending, there were two and a half of us at, at sort of that time managing the email and, you know, we were spending. 60 hours a week, you know, between, between the two and a half of us kind of managing it.

[00:06:43] George: ’cause it was very hard to do that, uh, back in that daily. 

[00:06:46] Ken: Yeah, I mean, I remember, uh, in, in the early two thousands, I mean, you were talking about actual physical servers that would go down. Yeah. And, you know, these things weren’t fast like they are now. You couldn’t take, you know, certain like basic things that we take for granted now, were real challenges.

[00:07:02] Ken: Back in, back in the day, no, there was a very limited automation of deployments. For example, you know, I, I remember actually, you know, crawling around a data center, hooking up cables for the early stuff that we did online. And, you know, the data center was some disused office building it, it didn’t look professional like things do now.

[00:07:22] George: Yeah, I mean, it, it’s, it’s, it’s interesting. So there are all those challenges and then, you know, a lot of software challenges as, as well. Um, really when you look back at the history of it, all of those open source MTAs, so you, you, you know, send mail and X and Postfix were. Really designed for, like, they were not designed for, for for large scale male.

[00:07:49] George: Mm-hmm. And while they generally worked, when things were. We’re, we’re working well, um, you, you know, also the internet wasn’t as reliable back then as it is today. And you know, what I used to say is like, you know, if when the internet has a bad hair day, um, it can become a real, it become, become a real issue.

[00:08:11] George: So, Hmm. You, you know, all of those systems. Um. Just as a for instance, uh, don’t, uh, don’t like, have an in-memory map of their, of their queue. So if you need to know what’s in the, what’s in the queue on disc, you have to go read the disc. And if you, you know, because you’re sending a lot of mail, happen to have.

[00:08:34] George: 10 million males on your, you know, 5,400 RPM Scsi drive. Uh, it, it can take a, like excruciatingly long time to, um, to, to, to, to even understand where your issues are, right. And where you’re having problems. So you, you, you know, like one of the core. Um, issues that we were trying to address was, was really, or I guess two of them were like visibility.

[00:09:06] George: Mm-hmm. And, and, and like maintainability. So be because the, the, the real world problems at that time were a thing would go wrong and you would know because your system wasn’t working right. Like you weren’t able, like, crawled to a halt, uh, and was like IO bound. But it was very difficult to understand you, you know?

[00:09:27] George: Is that. Am I having an issue delivering to a OL? ’cause a OL was the largest receiver back then. Uh, is it, is it something else? Like, that type of visibility was actually like, shockingly hard to get. Hmm. Um, and then being able to do things about it, so, okay, so like, now I do know what the problem is. What do I do about that?

[00:09:49] George: How do I isolate that problem so that I can at least contain it to the service with which I’m having issues and like, let the rest of my life carry on? And um, you know, back then those were like actually really, really hard problems. 

[00:10:04] Ken: Wow. Yeah. Fascinating. So the early years were, uh, you know, uh, leaving mathematics to pursue the love of your life, which worked out.

[00:10:14] Ken: That’s awesome. Uh, and, uh, and then getting into the, the world of the internet and, uh, sending mail for a large client, uh. So now, you know, fast forwarding a bit, uh, for most of us, we remember 2008 as the year of the great financial crisis. Um, and, and it was not a great time for tech companies. It was, it was like a really terrible time for tech companies, and yet it was in 2008 that message systems was spun out of, of Omni ti your consulting business that you started with your brother.

[00:10:49] Ken: What was the motivation for splitting message systems out into its own business unit? 

[00:10:55] George: They really had their own like clear different trajectories that they were, that they were on. Um, you, you, you know, we, so, so Omni ti was, was, was a hundred percent bootstrapped. One of the aspects, as you, as you know, as somebody who bootstrapped a company, um, like oftentimes the very beginning is, is is really slow.

[00:11:14] George: You can have a, a slow number of years. Mm-hmm. You know, as you grow literally from zero. Um, but. As we started to get traction in the space, it, it became clear that, um, that the, the businesses just had different goals and, and trajectories. Mm-hmm. And we loved them both and wanted them both to be able to be, you know, successful in their own right and to pursue their own trajectory.

[00:11:40] George: And the cleanest way to do that was to make them into, into separate businesses, uh, so that you’re not competing for resource or time. 

[00:11:49] Ken: Um. Yeah. Yeah. Very co. I think that’s a very common. Thing that happens, uh, especially with consulting, uh, oriented businesses, you know, you’re, you create a product for some use case that one of your clients has, then you realize that there’s other customers out there who have the same problem, and then pretty soon you’ve got a product group, but it’s like not profitable, you know?

[00:12:11] Ken: That’s right. It’s demanding all these resources and it, and it has a, you know, generates a lot of tension with the consulting side, which is, uh, eating what it kills, you know, one thing at a time. 

[00:12:22] George: A a. Absolutely. I, I, I mean, I distinctly remember having conversations with a former partner at Omni ti. Hmm. Uh, who w was adamant that we should like, kill the bu, you know, kill the product, kill the bu it’s not comparable.

[00:12:39] George: And, um. You know, I mean, I think it’s just symptomatic of exactly what you described, right? Yeah. Like product, businesses and consulting businesses are, are, are, are very different product businesses like kind of have to start at a deficit and then Yeah. And then, and then work up from, from that and then, you know, ultimately end up having, [00:13:00] you know, hopefully if you do it all right, like really good margins and really good, really good growth, you know, scale like.

[00:13:07] George: You know, very independently of the, of, you know, from the cost basis generally, um, where, whereas consulting companies are kind of the opposite, right? So like, if they’re not profitable from day one, you’re probably not doing something right. Yeah. Uh, and, and, and so just the, um. The way you need to look at them is, is, is, is different.

[00:13:28] George: And the easiest way to avoid that tension is to, you know, separate them when, when possible. Yeah, yeah. We, we were lucky at that time. So in 2008, we were able to stand on our own as a cashflow positive, uh, cashflow positive entity. Um, yeah, so, so it, it, it, it made, 

[00:13:47] Ken: it made good sense. So, uh, around 2010, there was a lot going on in the messaging industry.

[00:13:53] Ken: Uh, I remember, you know, I, I would go and visit, uh, various companies in Silicon Valley, you know, periodically to, you know, because we had partnerships going on or we were exploring partnerships and I. Your name message systems came up frequently, uh, on those trips. One thing, uh, one thing that happened in the industry, uh, that probably almost nobody knows about is, you know, there is this company called Cloud Mark that made a, a, a spam filtering.

[00:14:20] Ken: I. Software program that was, uh, in use by large mobile operators and telecoms and ISPs to filter large amounts of email traffic. Uh, and I know message systems had a partnership, uh, with Cloud Mark, uh, delivering MTA solutions to these large customers. And then Cloud Mark went along and bought a French company called Bazinga.

[00:14:42] Ken: Uh, and, uh, really, I, I, I just wonder at that, around that time, what was that experience like? You know, when you are, when a really significant partner of yours in the industry goes and makes an, an abrupt move like that and buys somebody else to effectively replace you in the equation. 

[00:14:57] George: Yeah. You know, I mean, I think none of that for us was unexpected at the, at the time.

[00:15:03] George: Um, uh, and. You know, just to be completely honest, and I’m sure some of this is like hindsight being 2020, but that receiver space market, especially for carriers, is a really challenging market. You have. Very, very long sales cycles. Mm-hmm. Um, you, you know, those groups, uh, generally are cost centers within, within the businesses that they operate in, uh, versus profit centers.

[00:15:33] George: So like, it’s really hard to get, to get, to get budget because they’re enormous. Um, you know, they have a lot of negotiating power around unit economics and, and so it’s, it’s just. It’s just challenging. I mm-hmm. Um, you know, we, we, we service that market and I think we, we serviced it well with a, a lot of high profile, uh, carriers.

[00:15:58] George: Mm-hmm. Uh, some, some great OEMs, but, but ultimately, you know, I, I think, I think getting a little nudge out, out of that market was honestly very useful. Um, interesting because I, I, I, I do think that. I mean, I know you, you operate in that space as well, so I don’t want to, uh, crap on your market too much, but I, I do think it, it is a, it, it, it can be a, it’s, it’s a very challenging, uh, place to, to, to be.

[00:16:30] George: Um, I. Hard to differentiate yourself. 

[00:16:33] Ken: Any, yeah. Any industry in which every sale involves some kind of RFP and layers of bureaucracy is not gonna be very comfortable. Yeah. Uh, you know, we, we didn’t, for, fortunately we didn’t get too deep into the telecom and ISP market. Mm-hmm. Uh, we branched out into web hosting and focused on email delivery as a problem area, which is, I think, a better problem area to take on with more scale.

[00:16:57] Ken: Yep. Uh. But yeah, that, that’s interesting. That’s interesting. So in, in a way that shakeup might’ve been helpful in, uh, in. Giving you a little extra encouragement to explore other areas that might have worked out much better, which in fact they did. Uh, yeah. And, and so in, you know, uh, in 2014 to 2015 after, uh, raising a bit of private equity in 2010, uh, you launched Spark Post as a cloud service.

[00:17:23] Ken: Uh. After, you know, many years of selling, uh, on-premise, uh, email delivery software, uh, that was owned by, you know, owned and operated by the customer. Now, you know, in hindsight, with the benefit of hindsight, I think we can all say now like, wow, what a, of course, that was an obvious thing to make a shift to the cloud.

[00:17:42] Ken: I. I’m sure that it was a difficult shift to make, uh, inside of message systems, you know, going, renaming yourself, rebranding, and then moving into the cloud with a whole different set of problems, uh, different economic problems to solve. What was that transition like? How did you make the move into being a cloud provider?

[00:18:01] George: Um, I, I, I mean, de definitely, definitely challenging. I mean, I, I would say cer certainly with hindsight, probably even in the moment. I think that. We probably should have gone a couple years sooner into the, into the cloud. I think from a competitive standpoint, that put us at, at a, at a disadvantage. Um, you, you know, I, I think that the, the, it’s.

[00:18:29] George: It’s, it’s interesting y you know, in some ways, so, so for me, um, personally, I, I look at it as almost like a return to the roots, right? Because, um, you know, I told you a little bit about the origin story of the, of the software stack we were operating that, right? Right. So, um, you, you, you know, long before, like the notion of cloud services existed, but you, you know, we were operating.

[00:18:55] George: Our on-prem software, you know, in a customer’s data center that we actually maintained. We were also, you know, ’cause we had a broad consulting relationship, we were also responsible for, for, for, for running, you know, their hardware and, and the number of their applications. And so. You and I and I, I think it’s a really healthy thing to do.

[00:19:16] George: Um, I think it really helps you understand your own product when you, uh, are responsible for operating it. You see how it actually works. Um, I. I think that you, you know, I, I think it’s a real, I think it’s a real advantage to, to, to, to do that. Um, and, and I, I think it’s really hard to, you know, have full empathy for your customers and what their experience is like if you can’t literally put yourself in their shoes.

[00:19:42] George: Mm-hmm. Um, that said, um, you know, for a lot of people who, who were at Spark Post or message systems, who you, you know. Didn’t, didn’t start in the late nineties. Um. You know, it’s very different. You go [00:20:00] from, uh, building, uh, software that you package up and, and give to people to having this kind of dynamic target.

[00:20:09] George: Um, and I, I think that there are a bunch of different, different challenges. A a lot of them frankly, are cultural challenges more than specific technical or, or business challenges. So you, you know, some of those are, um. Like, Hey, this thing like actually needs to operate all the time. Like we’re responsible for, for, for operating as a service.

[00:20:31] George: People are buying a, a service from us. Um, I, I think that there’s also, you know, one, one of the, uh, hallmarks of the momentum, MTA, the, the on-premise product that we sold was that it was extraordinarily flexible, uh, and had a really like. Robust, uh, modular API for being able to do all sorts of things with it made it a really great target for, uh, for, for, for carriers and for OEM partners who wanted to bake their special sauce into it in a highly performant way.

[00:21:05] George: I. But it’s hard. Like you can’t do that when you’re in, in the cloud. Like even if you can technically do it, you, you, you would be a fool to try and actually operationalize that because it, it’s not really scalable. Mm-hmm. And so, you know, you have to break, you know, the. I think anytime in a business, when you build a product and people love the product that they build and sell a product and love the product that they sell, it’s hard to change your mindset around like what it does, what it’s valuable for, and to, I.

[00:21:41] George: Fight that urge to say like, oh, they want to do this. Yeah, we can, we can do that. And it’s like, we don’t want to do that. Like we could do that, but it’s like we, we won’t be able to build a scalable business if we engage in that type of behavior. Hmm. Uh, and so I, I think changing the culture and mindset around those things is, is challenging.

[00:22:02] George: It’s actually harder. So not only there, but I think also within like corporate culture mm-hmm. It is oftentimes. Harder to change the orientation of people who, uh, have kind of only been in the ecosystem for a little while versus people who’ve been there for a long time. And I’m not entirely sure what, where that dynamic comes from.

[00:22:26] George: Probably part of it is that, you know, if you’ve been a place at a place for a long time, like you’ve already gone through a number of, you know, necessary changes, and so your appetite for change is like, it’s. It’s evolutionary, it’s not revolutionary. Whereas, you know, if if it’s the sort of the first big change that you’ve encountered, then um, I think that there’s a, a stronger motivation to be like, well, that’s not us.

[00:22:52] George: That’s not what we do. Right. Um, so it’s, it’s interesting. 

[00:22:57] Ken: Yeah. Like I, I remember the moment, uh, at which we made the switch from software to cloud. Mm-hmm. Uh, and. I was sitting in the office and I thought to myself, damn, you know, this stuff is working 24 hours a day. Like when I go home tonight, if something goes wrong, we’re gonna get called.

[00:23:22] Ken: I mean, we knew this, right? Yeah. But there’s that moment when you. Realize, and I think the rest of the organization realizes that there’s something really fundamentally different about running something 24 hours a day as opposed to just chucking software out. And, you know, you help people configure it, but ultimately it’s their responsibility.

[00:23:40] Ken: Uh, being a cloud provider, uh, you, you kind, it’s like having a kid, you know, you, you never quite sleep the same again. You know, 

[00:23:49] George: it, it’s true. I I think it’s also, you know, it takes an appreciation that for many customers. E email is like absolutely a mission critical part [00:24:00] of their business. Mm-hmm. And, um, if, if their email’s not working, um, you, you know, it, it truly negatively impacts their business.

[00:24:12] George: It can impact the, the, the career of the people that you work with, right? Yeah. So like, it’s not just like, oh yeah, it broke. Um, it. It can, you know, it can destroy businesses. It can, it can you, you know, like a good software purchase, good software relationship can make or break, uh, you know, your stakeholder’s career.

[00:24:35] George: And, and I, I think, I mean, I actually, I, like I would say that sounds overly dramatic, but it actually, I don’t think it, I think it’s actually a hundred percent true. Yeah. And, um, you know, it’s, it’s hard to appreciate that and take responsibility for it. Like people don’t live and die by it. Um, although we.

[00:24:52] George: Power some, you know, critical emergency notification systems that maybe you could stretch it to that. But in general, like people, [00:25:00] people don’t die if mail doesn’t go out, right. But it can seriously impact, uh, people’s livelihoods. And so it’s a big re it’s a big responsibility. 

[00:25:08] Ken: So, uh, you know, on that topic of, uh, powering emergency services, uh.

[00:25:15] Ken: Can you speak about any, uh, really challenging situations that arose in the early years of running the Spark Post cloud service? You know, stuff that just kept your engineers up all night. I know you might not be able to talk about some of them, but is there anything that you can talk about that, that was like an issue that really shocked you guys and you had to work wrestle really hard with?

[00:25:38] George: Yeah, I, I can think of two that come to mind. Mm. And I, I think that. Honestly, the, the hardest problem certainly that we ever incurred were ones that were, um, at least partially caused by, by others. Hmm. And I don’t say that to say that, you know, we had a perfect team, although. We had a [00:26:00] phenomenal team. Um, but, uh, I, I think it just makes things harder to debug when you, you know, when you don’t control all of the parts.

[00:26:11] George: So, you know, one, one that we blogged about back in the day, and I, I don’t think that that blog exists anymore, but, um, early in our. Spark post service life. We, we had a, a, a very mysterious like network outage that basically shut the service down for Oh wow. Uh, for, for multiple hours. And it was very unclear what was, what was going on, and like over a half a day of, of, of debugging things and working with our, um, with our, with our cloud provider partner.

[00:26:46] George: Um, it, you, you know, turned out that it was. Some undocumented, uh, limits, uh, around, um, around DNS querying, uh, that were designed to prevent, uh, DO sing inside the platform, um, like inside that cloud providers platform. And as I’m sure you know, like particularly for Alabama Mail can be a very heavy consumer of DNS services.

[00:27:14] George: Absolutely. Um, ’cause you, you know, like on any given day, like. Yes, like 80% of your mail goes to four domains, but the other 20% goes to a couple hundred thousand domains, right? Uh, and, and you can really hammer, hammer, DNS uh, especially if you kind of like take like a, you, you, you know, naive. Organic approach to it where like you, you know, you, you, you, you’re resolving off of each individual MTA, you know, and then you have a big cluster.

[00:27:47] George: So you take that big cluster, you multiply it by a huge number of domains and you can really slam slam DNS. And so, um, it took a very long time to figure out that. Effectively our usage pattern was causing our provider, largely unbeknownst to them, uh, to um, automatically put us into a, like, Hey, um, you, you, you know, there’s A-D-D-O-S, uh, going on here, and we need to radically restrict this traffic.

[00:28:19] George: That was super painful. 

[00:28:20] Ken: Right. Yeah. And you, you know, you have to, uh, reinvent the wheel while the car is driving down the road with DNS, right? Yes. Like, it’s not easy to, it’s not easy to just like do your queries in a different way when you’re making millions and millions of queries a day. 

[00:28:35] George: It is, and that infrastructure ended up evolving quite a bit over time.

[00:28:38] George: I, I mean, I, I think it’s like with anything you, you know, like if you have good visibility into what the problem is. That’s like 90% of, of, of a solution right there, you know? Mm-hmm. The, the real issue is like when you don’t understand why a thing is happening, because Yeah. Then you’re just shooting in the dark, so Yeah. 

[00:29:00] George: You, you know, I, I think it took us, I. Because it took too long. I think it took us longer to diagnose what the issue was than to to end up you, you know, put, put in a, a crutch to, to solve it. And then once you’re aware of it, then you get like hyper paranoid. So you know, then, then you develop like an extremely.

[00:29:20] George: Extremely robust, uh, DNS infrastructure to, you know, prevent that from ever happening again. 

[00:29:26] Ken: I think we might have gotten stung by the same DNS issue around the same time. Uh, yeah. ’cause we have. Very similar partner in the cloud space. And, uh, yeah, that one was tough. I, I remember, uh, it was not just making too many DNS requests, it was literally within the network equipment in the network that they were looking for particular signatures and packets and rate limiting.

[00:29:54] Ken: Based on those signatures. It wasn’t sort of hits to their DNS servers, it was literally packet control. Uh, and, uh, you know, so the fix was not simple. It wasn’t, you couldn’t, for example, just like run your own DNS servers because they were also really limiting those packets. Yeah. It was, you need like a hierarchical 

[00:30:13] George: structure where hierarchy, where you run a, a local caching resolver.

[00:30:15] George: Yeah. Yeah, exactly. Yeah, yeah, yeah. It, it’s, it’s, um, yeah, that was, that, that was a good one. Um, right. And I think a, like, probably the other one that, that really stands out was the, the Gmail apocalypse of 2019. I don’t know if that affected you, but that was, uh, lemme jog my 

[00:30:34] Ken: memory. What, what, what happened?

[00:30:36] George: So for a multi-hour period, I think GI, I think that the, the origin of it was that Google had an internal auth failure, but for a multi-hour period, a large portion of legitimate. Mailboxes at Gmail returned mailbox does not exist. Uh, like a 

[00:30:56] Ken: 500 error. 

[00:30:58] George: A 500. Like like a 500. This mailbox, this mailbox does not exist.

[00:31:04] George: Wow. And you know, like if you know anything about mail, which I know you do, that is a really awful situation, especially if you are. A responsible mailer, and we were a responsible mailer, and we en we enforced those responsibilities, uh, in a lot of ways on customers. So we, you know, we did automated bounce processing and things like that.

[00:31:26] George: Suppression lists. Absolutely right. Like we, we don’t wanna allow, we didn’t want to allow, uh, that type of, you know, bad behavior of like sending to. Poor quality lists, sending the lists that you don’t have the, the, the, the right to send to. Mm-hmm. And one of the mechanisms for, for doing that was to have automatic internal suppression around, around hard bounces.

[00:31:54] George: Right. And as a feature. Yeah, totally. But the nature of that error meant that. Like you and, and all of that, like the, the whole of that is conditioned upon what should be a really sound assumption, which is that Gmail will tell you the truth. Right, right, right. Like, Gmail’s not gonna lie to you if they tell you a, a, an address doesn’t exist.

[00:32:15] George: It doesn’t exist. Uh, and so this was a really pernicious error that was, um, like also had all sorts of cascading effects, right? So not only did it affect us and. You know, again, when like you own all the parts of the problem, it’s relatively easy to solve. So like we, we had good transaction records, we could back all that stuff out.

[00:32:37] George: That was fine. But you know, most of our customers were highly responsible mailers and also wanted to suppress bounces out of their own list. Right, right. Because right. You know, it’s a belt and suspenders type type situation. They were glad that, you know, we could do it, but they wanted to do it as well, um, because it’s the right thing to do.

[00:32:58] George: So, so then like you also put that problem’s been pushed out to you, you know, thousands of customers who, who now have this like highly bogus data, um, on, on, on suppressed users within, within their, uh, within their accounts and. It was a huge problem. Wow. Uh, it, it was, it was a huge, huge problem. Um, and like a, a good indication, like you, you know, one of the other things about, I guess probably just running a business in general, definitely running a cloud service is, um.

[00:33:31] George: You know, it kind of like, doesn’t matter whose fault it is, it’s your fault. Oh, yeah. Right, right. Oh yeah. So, you know, when our cloud provider, uh, breaks our network because of undocumented, uh, internal like network protections, I. It’s still our responsibility, which is, which is fair. And when Gmail breaks, like it’s still our responsibility because you came to us, you, you, you entrusted us with this service.

[00:33:59] George: And while we didn’t. While, while we didn’t cause it, you know, we, we definitely have to help clean it up and mm-hmm. And take responsibility for, for cleaning it up because we’re the experts. That’s what you hired us for. 

[00:34:12] Ken: The email space, uh, being any kind of 24 hour day email provider, whether it’s like offering mailboxes or delivering mail or.

[00:34:21] Ken: Cleaning up mail. Uh, it is such a mission critical service, and people are so sensitive to downtime and errors. Yeah. Uh, I think, uh, customers get angrier with their email provider than they might with other kinds of providers. I. Like even with their web hosting provider, if your website is down for 15 minutes, it’s okay.

[00:34:44] Ken: I mean, like, you’re gonna recover from that, people are gonna come back. But if email is down for 15 minutes such that people are getting non-delivery receipts, you know, like hard bounces, that is going to have ripple effects that cause damage for, uh, your customer and their users in, in a different kind of way.

[00:35:03] George: Yeah, I, I mean I think part of it is, um. It’s like the nature of email as this like asynchronous, persistent service that causes this. So like when your website goes down, like yes, you can’t access it for 15 minutes, but when it comes back up it’s like it never happened. Right, exactly. Like if you weren’t around for that 15 minutes then, then it literally is like it never happened where, whereas with with email, it’s different.

[00:35:32] George: Mm-hmm. Because. All those problems carry, carry forward. Like may, may go into your, into your list hygiene and management, uh, issues, you know, even on like an individual, uh, like recipient basis, like I. Now, now, like I tried to mail this person, uh, uh, you know, person to person, and I got a bounce. Like, do I believe that?

[00:35:55] George: Do I not believe that that bounces in my mail for, for, you know, my mailbox until I delete it? So like, you know, the, the, the, the visibility of the problems, I think, um, uh, like lasts, lasts a lot longer. It persists a lot longer. 

[00:36:09] Ken: Yeah. Uh, it’s email’s a challenge. Yeah. Let’s, uh, let’s fast forward a bit. You know, SparkPost obviously worked out, uh, you know, you guys did, uh, additional, additional private equity rounds.

[00:36:21] Ken: Um, eventually reached a fairly significant scale as a company. And then, uh. During the pandemic, uh, message Bird came along, uh, and, and acquired Spark Post, uh, how did that happen? I mean, obviously I know acquisitions, there’s a lot of confidentiality that, you know, things you can’t talk about, but generally speaking, how did that happen?

[00:36:41] Ken: Why did they, uh, 

[00:36:43] George: you know, I, I think that there was a period of consolidation in the platform as a service, uh, space. So shortly before that, um, uh, Twilio acquired Syngrid. Well, technically didn’t happen until after our transaction. At the same time, cinch was acquiring mail gun and so, right. I, you know, I, I can’t reach into the, the, the mind of the leaders of, of Bird or, or Twilio or, or Cinch.

[00:37:13] George: But, um, I, I think that, that, that, that consolidation, definitely, you know, these things happen 

[00:37:21] Ken: in waves. Yeah. Uh, you know, for somebody who’d been running his company since the 1990s, uh, and, and actually working with email since about 99, uh, what was it like to, you know, what was it like to sit down at the closing and sign that final document, uh, and, and hand the hand the keys over?

[00:37:43] George: I mean, honestly, it was, it was, it was good. It was, uh, we worked in that space for tremendously long time. Mm-hmm. Um, you know, through ups and downs. Um, and so, you know, like, no, no regrets around that. I, I think it also helps that. We had a comprehensive employee stock ownership program or stock option program.

[00:38:06] George: So, you know, everybody was a, a aligned. Um, so, you know, it ended up being a, a really good transaction for employees, for investors, for founders. And like what will kinda like, what more can you, what more can you, can you ask for? 

[00:38:24] Ken: Yeah, absolutely. I mean, it does seem, uh, like the numbers were great. It was, it was $600 million exit.

[00:38:30] Ken: Uh, uh, I know the multiple might be a secret number, uh, but, uh, certainly a really healthy figure that that left a lot on the table for employees to participate in and, and investors and whatnot. That’s definitely not always the case. Uh, yeah. And, uh. And it happened at a time during the pandemic when there was a real froth in the market.

[00:38:51] Ken: A lot of m and a going on, uh, probably the most fantastic timing. Um, so what, what’s next for you? What have you been doing in the years since, uh, spark Post, uh, got acquired? 

[00:39:03] George: Yeah. I mean, so professionally I’ve been doing a lot of, uh, advising and some board work. Uh, both, both of which I really enjoy a little bit of technical diligence work, uh, which I also enjoy.

[00:39:15] George: Um. You, you know, right now I’m loving that. And, but both kind of like the diversity of intellectual challenge that comes with like, learning about a bunch of new businesses and in some cases new spaces, new like parts of the email space that, um, you know, like wasn’t as familiar with as I was with, with sending, um.

[00:39:42] George: So that, that part is, uh, really great. Also, love that it affords me, uh, good freedom to do a bunch of stuff that I, I couldn’t do in the, you know, proceeding 25 years. So, you know, I’m, I’m a, I’m an avid landscape and nature photographer and I’ve been able to spend a lot of time, uh, doing that as well. Uh, and so I think achieving that balance has been really nice.

[00:40:07] George: Um. Also really love, like I can go grocery shopping with my wife. Um, oh 

[00:40:12] Ken: yeah. You finally get to spend time with her. Yeah. After all the entrepreneurial years. 

[00:40:16] George: Yeah. I, I, my, my daughter, uh, my daughter graduates from, from high school this year, been able to spend a lot more time with, with her than, um, you, you know, than I, I had been able to when I was, I mean, you know how it is, like when you’re, when you’re, when you’re working a, a startup, you know, it’s a lot of 12 hour.

[00:40:36] George: 14 hour days, a lot of weekend work. And you know, especially when you, when, when you, when, when you run it. I’m, I’m definitely a person of the opinion that like, I won’t ask anyone to do something that I wouldn’t do myself. Mm-hmm. And so like if, if, if that’s what’s, what’s needed, plus I also like, I understand like I have a dysfunction and, um, a [00:41:00] dysfunction.

[00:41:00] George: You, you know, I, I, I. Find it really hard to like let go of things. Mm-hmm. And I’m a bit of a perfectionist and so uhhuh you, you know, like, like also kinda like no, no regrets about where I was. It was the, it was the, the, the right and natural thing to do, but also enjoyed not, not having to do that now. So, you know, in terms of like operational roles, I don’t know.

[00:41:25] George: I don’t know if that’s, I, I don’t know if I, if I, I think it would have to be the perfect thing. And the perfect thing has, um, um, not crossed either my desk or my mind yet. 

[00:41:37] Ken: I feel like for most mortal people, you’ve kind of got one. I. You’ve got one startup in you, and if you’re fortunate, you get an exit outta that startup and then you can move on and do other things.

[00:41:49] Ken: But it is such a hard to appreciate lifestyle. Uh, you know, you speak about working, uh, 12 hour days and on weekends. For me, the, the impact is more, uh, the fact that you’re always thinking about it, you know, you’re always thinking about it. There’s never really a time when you can switch off. Your holidays aren’t really holidays.

[00:42:10] Ken: Uh, because there’s always that one call you have to take or just that thing that you’re thinking about that you can’t leave behind. You know, you’ve got this big thing running underneath you. 

[00:42:19] George: Yeah. I think you just have to make peace with it. I, yeah, I, I mean, I, um, I. You know, probably for the first f 16 years of, of running things, um, you know, I, I pretty literally never took a vacation.

[00:42:37] George: Yeah. Um, and, uh, and then, and then once, uh, I moved from the CEO role to CTO role. Mm-hmm. Brought in, uh, Philip Philip Merrick to be, uh, CEO and, and as well had like a, a broader, uh, like a, like a broader like strong leadership team, you know, started to try to be able to take [00:43:00] like, hey, like maybe I can do a backpacking trip, right?

[00:43:02] George: Um, you, you know, once every two years. Um, and I will say that honestly, like, again, probably part of a dysfunction, but. I found that traveling with my family as an entrepreneur was a lot easier if I didn’t turn it off. So if I turned it off, it kind of ruined it for everyone. ’cause I was very stressed, right?

[00:43:30] George: Like, you don’t know what’s going on. You feel out control, and, uh, and it, and it’s really, it’s really hard. You also like oftentimes come back to things, uh, that you didn’t expect. And, uh. And like it can be really difficult. And so for, at least for me, and again, this is not for everyone, but for me it, I found it worked a lot better if I just budgeted a little bit of time.

[00:43:55] George: So, like, you know, the rest of my family likes to sleep in. I, I’m a relatively early riser, so. If I budget myself an hour in the morning to like kind of go through email and you, you, you know, at least be aware of things. Mm-hmm. And figure out like, Hey, do I, you know, need to respond to this now Do, do I need to respond to this now with a, like, I am aware of this and we’ll respond to it later.

[00:44:21] George: You, you know, can I just like leave it lie? That for me was really. Uh, helpful. Like, it, it, it, it made, it made, it made family vacations. I, I think a lot more tolerable for everyone. Um, 

[00:44:34] Ken: yeah. Yeah. So I can, I can relate to that. You can’t really fake it. You know, if you’re, if you are, uh, an entrepreneur, if you have a, a company, you know, it’s on your mind all the time.

[00:44:44] Ken: Uh, being straightforward about that and, and, uh, and letting the people in your life know, uh, that you need to take some time for, it’s probably a better thing to do than, than, uh, pretending it doesn’t exist. So, is there any. Uh, is there any time in your life for mathematics anymore? Do you spend, you know, are you reading papers?

[00:45:04] Ken: Uh, no. No, no, 

[00:45:05] George: no. I, I, I, I, I, I mean, I’d love to say that there is, but No, not, not, not really. You know, it’s interesting. I mean, my son’s in college now and, um, you, you know, so it’s like. As much as you can with a 20-year-old, like try to talk to him about my own educational experiences and the you, you know, like lessons that took me a long time to, to, to, to learn.

[00:45:30] George: And you, you know, one of the things that I’ve told him and I tell other people is. I love my education and I think that it, um, uh, provided me a really solid basis for approaching problems. ’cause fundamentally, mathematics is about, um, you know, like understanding the conclusions from a given set of data.

[00:45:55] George: Like, that’s kind of like really at its heart what, what, what mathematics is, is all about. Um, and I think that that’s a fun feel, like a, like a really foundational, like awesome skillset. Also, I think in my professional career, definitely until I started, uh, like taking on ad hoc management of our data science team post the bird acquisition, like prior to that, I think I used like stuff that I, I.

[00:46:22] George: You know, I, I think the number of times I used anything that, that I learned after freshman year in college was May, maybe twice. 

[00:46:30] Ken: Yeah, yeah, yeah. 

[00:46:31] George: Like maybe twice. Like in terms of like specific, specific things 

[00:46:37] Ken: Yeah. Ordinary differential equations and stuff. Uh, theorem proving don’t really come into the picture very often.

[00:46:43] George: They, they don’t. Yeah. I think that the, that the, the approach to problem solving, like 100% applies right? And Right. Um. I, I think that that has served me well in general, in life. Um, I like to think of myself as a, as a, as a good, uh, debugger and problem solver. Mm-hmm. Um, and, uh, I think a lot of it is supported by the type of thought processes that, that you develop, uh, when you do that.

[00:47:12] George: But in terms of, yes, so like, like, hey, like you, you know, how, how can I apply? You, you know, real analysis to a situation that’s like, doesn’t, doesn’t happen. Yeah. And I wasn’t a statistician. If I was a statistician. Like May maybe. Right. Um, so like e even, even with a lot of the advances in, in, in, in data science, I mean, first off, like a lot of that stuff’s just like.

[00:47:40] George: Like really large linear algebra. Um, but, uh, but beyond that, to the extent that it’s more than that, um, you know, I think an applied mathematician, somebody with like a group graph theory, stats, background, uh, is probably better suited in terms of like actually being able to apply specific, you know, 

[00:48:00] Ken: specific knowledge.

[00:48:02] Ken: Yeah. Interesting. Yeah, I always think, uh, it’s hard to take the X out of the Y now, whatever your X is. Uh, it’s how you’ll live your life. So if you study, you know, if you did two math degrees, you’re going to be applying the principles you learned, uh, to everything. And I, I find that in my work, um, as a CEO over the years, I.

[00:48:22] Ken: I, I can’t take the, the programming engineer out of it. Like everything I do, if it’s some, you know, if it’s some task that, that requires banging through a spreadsheet, I’ll write a program to do that and, you know, to, to orchestrate calls to an LLM or something to solve some problem that is honestly very simple, but boring.

[00:48:42] George: But I think you’d agree, like those are like foundational skills, right? For sure. Like, that’s not, yeah, that’s not like third year CS stuff. No, that’s, that’s, that’s what you, I mean, honestly, like. Probably not ’cause we’re of an age, but, um, today it’s like stuff that you, that you learn in high school. Uh, yeah.

[00:48:58] George: You know, before you go and, and like, you know, that’s like the general, general programming that now everybody should learn. Right. Um, but, but I agree. I mean all like, we’re, we’re all a product of our environment and, um, yeah. I, I think we’re, we’re the, we’re the richer for it. Yeah. 

[00:49:17] Ken: Well, I think we should end it there.

[00:49:18] Ken: George, it’s been a real pleasure talking to you today. Thanks for coming on the podcast. Yeah, it’s my pleasure. Thanks for having me.