August 2, 2022
President of Pearmill, ex-Head of Product at Taplytics, ex-Head of Mobile at Frank & Oak. YC fellow.
Peter “Fish” Fishman
Co-Founder, Mozart Data
On this week's episode we have a pioneer who has helped democratize the modern data stack. Peter “Fish” Fishman, co-founder of Mozart Data (ex-Zenefits, Yammer, OpenDoor) built his company on the idea of simplifying and centralizing data infrastructure to help startups grow. And it worked! Mozart Data has fast become a staple product that many startups use to ingest and analysis their data.
In our conversation he defines what the modern data stack is in today’s world, shares his wisdom building and constructing data systems and teams plus how Mozart Data has helped scale some of the most notable unicorns and even a few decacorns since they’ve started the company.
More Highlights Include:
[00:00:00] Nima Gardideh: [MUSIC FADES IN]
[00:00:03] Seth Bindernagel: It's this really delicate balance between knowing your product from like a pure intuitive sense what the right thing to do is and then using data to support those things [SWEEPING SOUND EFFECT ]
[00:00:13] Sean Byrnes: If you're in a growth stage company there's things happening all over the place There's lots of distractions You can get lost in your own world rising above it And putting yourself back in that customer seat is such a superpower [SWEEPING SOUND EFFECT]
[00:00:25] Peter "Fish" Fishman: In the early stages and you might not want to hire someone dedicated to growth until you have some sort of repeatability demonstrated [SWEEPING SOUND EFFECT]
[00:00:33] Aaron Glazer: The industry is not staying static in terms of like what do you need to make these decisions And how can you grow [SWEEPING SOUND EFFECT]
[00:00:39] Scott McLeod: I always like the notion you're like a mad scientist are mixing in a beaker of like all your different variables You want to keep some constraints You want to keep some variables let's say a little bit more of this and you want to see how it reacts.
[00:00:50] Nima Gardideh: If you're new to our show, welcome I'm Nima Gardideh, full-time CTO and part-time podcast host. If you're an avid listener, you know that we bring on some of the most prolific founders and growth marketers sharing their biggest challenges. Growing the most recognized brands in the world and at hyper-growth speed. I had the pleasure of speaking to Peter Fishman AKA Fish, the founder of Mozart Data on this episode. B Is helping democratize the modern data stack to companies around the world.
[00:01:22] Before going to Silicon valley and eventually moving to New York. I started my first company in Canada, where the most common thing you would hear is that we should operate the way companies in Silicon valley operate. I always thought that this was bad advice because you should be building companies the way that makes the most sense for your culture. But there was some truth to it. There are a lot of processes, software and management structures that were built by successful companies in the valley.
[00:01:48] And then there's a lot to learn from. Fish and his team at Mozart Data are part of a generation of companies which are helping others have access to some of this innovation. In this case, it's the modern data stack.
[00:02:03] In our conversation he defines what the modern data stack is in today's world, and shares his wisdom building and constructing data systems and teams. Plus how Mozart Data has helped scale notable unicorns, and even a few decacorns. But first Fish shares how it all began.
[00:02:20] Fish: I'm from the great state of New Jersey. I grew up on the Jersey shore surrounded by a bigger family. I used to love to go to the racetrack, which is where I learned everything I know about data and watching a lot of horse racing and other sports are what I was passionate about it sort of peaked my interest in statistics and I was never really as accomplished on the field as I was at, maybe understanding it or analyzing it ultimately found my way to the world of economics and data before eventually becoming part of the data tech world.
[00:02:54] Nima Gardideh: I don't know much about the horse racing world. Were you racing yourself or were you betting on races?
[00:02:59] Peter "Fish" Fishman: When most people refer to being a fan of horse racing, they're typically not describing the role of like a jockey or trainer. They're typically describing somebody that is sort of a self described handicapper, which means someone that assesses the odds of almost every single horse in the race and says, what's the probability of each of these horses winning.
[00:03:18] There's a rich set of data that is made available to the betters every day, which is basically horse histories. using those horse histories, the better sort of come up with a collective belief about the odds of each horse winning, the track or whoever is arranging the bets, takes some out of the pot for themselves.
[00:03:36] And then you just try to beat the consensus view. So your perspective on the odds has to be better than the consensus view. This is literally what one does in data, in any profession. But it's a really crisp and clean profession because you get really tight feedback loops, you know, a race happens and then, you know, you get paid out or you don't,
[00:03:55] Nima Gardideh: Oh, this is super cool. it's quite fast, right? So the race will take a couple hours, I assume, or even shorter.
[00:04:01] Peter "Fish" Fishman: A horse race tends to be less than two minutes.
[00:04:03] Nima Gardideh: Oh, super fast.
[00:04:04] Peter "Fish" Fishman: It's about 12 seconds per eighth of a mile.
[00:04:07] Nima Gardideh: And how much time do you get to crunch the data? Do you get it the day of or?
[00:04:12] Peter "Fish" Fishman: The answer is some people have really advanced models that they can crunch the data, whenever, but there's a signal that also happens throughout the day. Like sometimes, the weather is not what was anticipated or sometimes the track tends to be a little bit biased towards speed horses or closing horses.
[00:04:31] So even great models really need to have somewhat real time updates in their beliefs. even the signal, the private signal of the betters should also play into some of that model. In general, there's 30 minutes between races, sometimes a little bit less. Basically handicappers are trying to assess what's going on in a short amount of time with a pretty rich data set.
[00:04:54] Much of which is already in the price of the horses and in the consensus view.
[00:04:59] Nima Gardideh: How old were you at this point?
[00:05:01] Peter "Fish" Fishman: You know, I started going to the racetrack when I was two, I grew up about five miles from a racetrack and it was where almost everybody in my family had their first job. So basically that's a very special place for me. That sport has declined massively in popularity over the last sort of century. I always kind of love the thought exercise and the sort of, all the thinking that goes on around trying to assess what is the probability of certain events happening. that's really translated into all parts of my career.
[00:05:32] Nima Gardideh: That's cool. So yeah, this is one of the questions I was gonna ask you and kind of going through it is what do you think you've learned or like internalized as part of growing up in that world? I was raised in a quasi Marxist family. So I end up having very different points of view on capitalism in general and how companies should be run because of it.
[00:05:52] I, And I'm very aware that a lot of the ideas were instilled in me as opposed to having discovered them from a first principle standpoint. And I wonder what do you think when it comes to your upbringing? This part of it also seems interesting, but generally speaking in the area you grew up with and your family that you've internalized in a way that is very deep inside.
[00:06:11] Peter "Fish" Fishman: I wasn't really planning on getting into a nature versus nurture debate this morning. But I will say that obviously a number of things sparked in my upbringing. I got very excited about what was one of my strengths, certainly growing up I was good at math and statistics in a way that felt rewarding to me and that I wanted to pursue and that it just connected with me.
[00:06:32] And I thought it was a great way to understand different elements of the world. Now horse racing is just a nice example just because it's fun and entertaining. and it has this sort of principle of you can prove out to be sort of right or wrong in your beliefs over a very short amount of time, right?
[00:06:51] So you make a bet on a horse, you say, okay, I believe that this horse has a better chance of winning than the rest of the public. one realization doesn't tell you whether that's true or not. [00:07:00] How that plays into, who I am today more than, you 40 years later it's hard to separate out how much is my upbringing, how much is, stuff that's sort of naturally connected and sparked with me, but it's certainly the case the way that my family growing up used to tackle these problems and love to engage with the problems of, Hey, what do we think is going to happen?
[00:07:20] And their prism on, not just horse racing, but many elements ended up being a big part of one my career and two just how I live how I think about the world and not just, things in, building a tech company, but also Whether it's you know, any of life's challenges, I purchased an apartment in San Francisco and when I thought about making an offer on that apartment, how do I think about that offer?
[00:07:45] How much should I offer how wedded am I to buying this place and all of that sort of stems from sort of a set of thinking that I developed, generally as a young child and then ultimately, matured over time. So it's not surprising to me when you tell me that a big part of your upbringing now trickles into how you think about the world. It's very evident in the way that I run Mozart Data, the company, but also just how I live my life.
[00:08:10] Nima Gardideh: Yeah, that makes a lot of sense. And I think I've just seen that over and over again. It's cool to see people being aware of it. And I think yours is quite interesting, cuz it kind of links all together all the way to what you're working on now. Did you end up getting into data? So strictly in school you went down that path, what did you study before you got your first job?
[00:08:29] Peter "Fish" Fishman: I studied math and economics, and I ultimately did a PhD in economics. So I was nearly 30 by the time I had my first job, my first sort of longer professional job. I anticipated being an academic and a teacher for most of my twenties and ultimately came out to graduate school in the Bay Area, fell in love with the Bay Area and the way to stay in the Bay Area as an applied statistician was to get a job in tech.
[00:08:57] And the tech scene was really booming the opportunities for statisticians were multiplying really quickly. Some really interesting dynamics were happening, there was, abundance of data becoming available. You could really optimize the number of businesses with that data.
[00:09:15] There were a very small handful of large enterprises that leveraged data in a meaningful way to their advantage. So there was just a sort of limited amount of jobs with a very structured sort of entry path. that really went out the window at the start of the 2010s, as more powerful databases and cheaper abilities to gather and consume data.
[00:09:38] And then more people with this skillset started to enter the space.
[00:09:43] Nima Gardideh: This is super interesting. I feel like we could spend the whole thing. Just talking about why you decided to do a PhD. But just to quickly hop over that, do you recommend that path for people going into academia and just working on adding to the tree of knowledge or do you think it's much better to go [00:10:00] straight into being more applied in the world?
[00:10:02] Peter "Fish" Fishman: It's certainly the case that if you wanna be a professor, there is one route to do it, which is you go into academia and you push the bounds of knowledge. Through pushing those bounds of knowledge, there are positions for excellent research that are made available, all over the world where you can work on whatever is intellectually interesting to you and try to, again, pierce the limits of knowledge.
[00:10:25] I think a lot of folks that go into graduate study and myself included, and this is a knock on me, were maybe good at school, their entire lives and wanted to just keep doing more school. I'm good at it. And that sort of has some like dopamine that, you get this sort of rush of a lot of accomplishment, ultimately graduate school and academia is not really about learning some things, taking a test and proving that you've learned it it's about really being able to research and do really deep research on these problems. And it [00:11:00] turned out that I actually wasn't that passionate about that. So I wasn't really that passionate about pushing an idea to its limit and really stress testing it and trying to find some science in it.
[00:11:11] Just like the horse racing story, I love coming to an answer in 30 minutes, that is, more often right than not. That's what I look for and the types of problems that excite me. I mean, In some sense, when I was 21 and applying to graduate schools, maybe I should have been able to have a little bit more of a deep understanding about myself.
[00:11:30] When I started one of my most important tech roles, I was the head of analytics for Yammer. I was on the engineering executive group. So there were a number of leaders in the technical discipline and I was one of them. And in that group it turned out not that I had more education than any of the people I had actually more education than the sum of all the people in that group which was a source of like great pride and great [00:12:00] embarrassment, which is to say that a lot of my peers were able to gather real world experience and be just as accomplished if not more than myself through really sort of getting their hands dirty.
[00:12:12] So when it comes to sort of the advice about whether or not to be an academic. It is certainly the case that if you want to be a professor, there is one way to do it. But, nowadays if you you're at an applied statistician like myself, there's so many incredible data sets that are so available, whether that's in companies or publicly, that you can answer really interesting questions in of shorter time horizons with different sort of focus, like the focus being the business outcome, as opposed to, again, stretching the knowledge.
[00:12:43] And if that's something that sort of appeals to you, I would say that you're better off honing your craft and your skills as opposed to necessarily pursuing a particular sort of end career goal.
[00:12:56] Nima Gardideh: Yeah, that makes a lot of sense. And I think that the thing that stood out to me [00:13:00] about what you just said. Knowing yourself takes a lot of work. And when you're that young, it's actually quite hard to know what you are meant to do or what you will be very good at doing that also makes you constantly interested and, not only gives you the dopamine push that you're sort of want to look for, but also, enriches your life beyond just being able to do something that you think is the right thing to do.
[00:13:21] Just because all these other people behind you have done it I wanna get into your Yammer experience. But one thing I'll mention about PhD is that I have a few friends that have started companies either during their PhD or after, and the pain in which they have experienced in PhD seems to be very similar to founding companies.
[00:13:39] So that seems like a very good precursor to starting companies. And we can, I'm sure you, you either agree or, or have felt this before.
[00:13:46] Peter "Fish" Fishman: I don't think that the purpose of pursuing a PhD is to get comfortable with discomfort. I would say that the extreme commonality between doing a PhD and running a company is [00:14:00] non-linear progress. Again, when I was in college you made effectively linear progress, which is you get a subject you're taught to learn certain things about the subject.
[00:14:10] And at the end of a month or three months or six months, you get handed in an exam that tries to measure how much you've learned on the subject. for me, and for many, they check that box. I learned it. And often I learned it pretty well. What is different about graduate school is that you can work on a problem for six months and truly make no progress.
[00:14:32] You could work for six years and make no progress. It's that non-linear progress that is particularly tough. I talk about my time in graduate school often, which is to say I got to work on whatever I wanted. I got to work with some of my intellectual heroes and I got to have a great set of colleagues. These weren't very stressful or high pressure days.
[00:14:56] And I never felt great, which is so confusing. I mean, [00:15:00] the description of the work sounds almost ideal. I worked on it in a beautiful place and everything was great except I didn't feel great. And I think for me A big part of my identity whether it was academic or professional steps, that it marked my improvement or growth.
[00:15:20] And you don't get to experience that in either graduate school or as an academic, you have to get used to that non-linear progress. And as a founder, that's the definition of your life, which is to say one, nobody, will hand you a degree you know, you might graduate your funding series.
[00:15:38] But ultimately it's essentially the quality of the company that you're building and you don't get these very obvious if I put in X amount of hours of work, my company gets Y percent better. That's sort of the big similarity. So it's not surprising to me that you know, whether it was B D so all book dissertation or, myself, I actually finished my PhD.
[00:15:58] But these folks [00:16:00] that transition from being academics into being entrepreneurs, that's a, it's not a seamless transition by any means. I think that there's a lot of, mutual aspects of a personality that enables one to be good at both of those roles.
[00:16:13] Nima Gardideh: Yeah, that distinction is interesting. The one I had heard which I personally relate to as an entrepreneur is this internal ability to wake up every morning and do things without anyone putting a measuring stick, which I think is, what you're alluding to is there's no linear progress in that there's no day to day
[00:16:30] rubric if you are getting all the right things done. You have to figure all of that yourself and, and sort of create your own schedule and your agenda and, and lead through uncertainty effectively. And that part seems super similar. walk me through, so Yammer was the first sort of let's call. Professional data team that you were working on. Can you just give us a snapshot of what the data infrastructure was back then? What was available to you guys? And what was the role of the data team within Yammer?
[00:16:59] Peter "Fish" Fishman: I had a few quick career stops before Yammer. So I worked first actually out of graduate school. My first job was actually at the Philadelphia Eagles. So I was doing data analytics in football at a time where there were a few sort of geeks in the front office. And then I had more of an economic consulting role and then ultimately found my way to play them, which was a video games company that built
[00:17:23] video games typically on top of Facebook. That was sort of a transformational moment for me, basically the data was enormous. We had hundreds of millions of users through Facebook pretty much anything that you wanted to experiment on, you could and get almost instantaneous results.
[00:17:39] It was very intellectually fulfilling. I used to be an experimental economist and I would work for months and months and months on running an experiment with maybe hundreds of observations and here you could get instantaneous sort of results. And it was really incredible and there was a big data engineering team that would help, clean and massage and create and move that data to a central warehouse.
[00:18:00] It was sort of that infrastructure that I had in mind that I brought to B2B SaaS. It was a very forward looking perspective of the Amer leadership team, David Sachs and Adam Poony. They said, Hey, we wanna build a B2B product, like a B2C product. That today is pretty normal.
[00:18:17] You see B2B products that are well designed, and a lot of the PLG or product-led growth products tend to have that consumer mentality. But this was a pretty unique perspective back then. the types of data infrastructure that you would see then is exactly what the modern data stack is today.
[00:18:35] The names have changed and the quality of these toolings have changed. But what you saw a decade ago it's not necessarily state of the art, but it is what I think of as the world class offering of today. So at Yammer we built a tool called avocado today which is literally Mozart Data plus Mode Analytics.
[00:18:56] So the Mode team came from Yammer. The Mozart team came from Yammer, it's the exact same infrastructure. It was extracting and loading data from tools like Salesforce and our database into a column which was Vertica.
[00:19:10] And today, the El tooling is rather than engineers writing scripts,
[00:19:15] we leverage essentially an El platform like Mozart Data or Fivetran. And then the powerful database which was for us then is now Snowflake. But beyond that, we created tools that helped data analysts do data modeling. Essentially what you would see in the transformation layer.
[00:19:34] Tools like DBT, and then ultimately EI tools. We had an EI tool that we called SQL, which was capital S, Q and an L. So it was a tool that helped savvy data people write sql queries and share and distribute them. So literally the entirety of. what I think of as the skeleton of the modern data stack we've been using for really decades.
[00:19:58] It just happens that today this infrastructure is available, like by snapping your fingers and it's up and running and you're off to the races. Back then you had to hire a bunch of engineers. You had to write very, you had to do a lot of custom work. You built not bought.
[00:20:17] So I think the huge transition is what used to take maybe millions of dollars just to get started today. You can get started with a credit card swipe and, you know, maybe six bucks. It's really an unbelievable difference. But the sort of thinking is really identical to where it was say a decade ago.
[00:20:36] Nima Gardideh: A lot of these things like there's open source versions of everything you just talked about as well. Right. Did Yammer open source SQL cuz I'm remembering that name?
[00:20:50] Peter "Fish" Fishman: So there's a lot of sort of sequel editors with like words that have SQ and L in it.
[00:20:51] Nima Gardideh: I guess so, Yeah.
[00:20:52] Peter "Fish" Fishman: So no, all of these tools were for our team and ultimately got used by tens of thousands of people a month at Microsoft. When Yammer was ultimately acquired by Microsoft. But a lot of the thinking played into my subsequent team at benefits, but also into the tool chain that the Mode Analytics team built.
[00:21:11] And then ultimately the tool chain that we started at Mozart Data.
[00:21:15] Nima Gardideh: We're used to being users of Mode actually. They've been focusing more on the analyst end of this. And I think there's now like multiple users of this ETL ecosystem. There's marketers and product managers and analysts and everyone needing the data, which makes a lot of sense.
[00:21:29] So after you went from Yammer to benefits, did you effectively rebuild some of that infrastructure? Platform layer stuff that you no longer had to build available at that point? Or did you have to
[00:21:43] Peter "Fish" Fishman: Again, the name started to change a little bit. When we were at benefits we were a Redshift shop, so we transitioned to a cloud data warehouse approach then actually became the first customers ever of five tra. So rather than hire a data engineer to extract and load data from, our CRM was Salesforce.
[00:22:05] So rather than try to move data from Salesforce to Redshift we thought of that as almost somebody else's problem that they could solve maybe at scale. And in fact, five one did solve that problem for us. They did a great job of solving that problem for us. And ultimately they did a great job of solving that problem for lots and lots of companies.
[00:22:24] So I think again the sort of thinking and structure of data needs to get to a central place. You need to combine data sets. you know, there's sort of a one plus one equals three. This is, that was a Microsoft term, but there was that mentality with respect to bringing multiple.
[00:22:42] Sets of data together. So a lot of data you can look at, gets generated in a single place, a single tool, and you can look at time series of that data. Turns out often that’s not as interesting as being able to cut it by a dimension that typically exists in a different tool. Generally data gets a lot of its power by being able to join it together.
[00:23:03] So the philosophy of bringing everything to a powerful central place is really sort of my new religion and certainly core to the philosophy that we have at Mozart is also a huge part of what we did at Zenefits.
[00:23:18] Nima Gardideh: Can we address this? A powerful central place where all your data lives. I feel like people have called that so many different names. Do they have distinctions in your mind when people talk about data lakes versus data marts versus data warehouses, like there seems to be, at least one of them has a distinction to me, but I was curious as to, if you think those distinctions matter or they're just different ways of describing sort of the same idea behind what you just mentioned?
[00:23:43] Peter "Fish" Fishman: I think the first thing is that when you typically visit the marketing site of almost any data tool, it almost sounds exactly the same. They can be doing completely different things. One might be selling peanut butter, the other might be selling bubble gum, but they sound exactly the same.
[00:23:57] Get yourself saded. And what that means is basically, yes, I do think that you bring your data to a central place and then do magical stuff with it to get insights and optimizations and better company actions. That's ultimately what every single data company is selling. And maybe they're not selling that directly, but a lot of their marketing messages start to look something like that.
[00:24:18] In fact, very often I get people emailing me saying, Hey, Fish, did you see this new company? They do exactly what Moza data does. And then you land on their page. And it sounds exactly what Mozart Data does and then you sort of go, you know, a half step deeper and you're totally different from what we do.
[00:24:34] So yes, so I'm a fan of the data warehouse approach for, mostly a, a function of the power and the newer cost models that make the data warehouse, the right answer for a number of companies within a certain sort of context. We mostly work with reasonably early stage, so like series A-ish companies. Very rarely does it make sense to have a very complex data strategy where you move data to an intermediate place before ultimately putting it in the warehouse?
[00:25:07] My general philosophy is one, I do think all of these things are ultimately similar, which is to say, what is the sort of environment where you are going to join your data together so that you can, humans do this really well. I just walked into the office and, you know, in order just to get into the office safely, I have to observe what's going on with the street lights.
[00:25:30] And I have to recognize whether there's a car coming. And I have to think about kind of, what's the speed of that car and think about the physics of it. You wanna do the same thing with your data, which is you're trying to get actually a richer picture. The only way to get really a competitive advantage out of the data is not to look at it in a very direct or obvious way.
[00:25:47] very rarely can you have a very simple strategy that just wins for you? So again, if I were at the horse races, I can't just say, alright, I'm gonna bet on every horse that's gray. Most horses are brown or some variant of that, but sometimes horses are gray.
[00:25:59] Maybe [00:26:00] I like gray horses. I'll bet on gray horses. I can't really do that. And necessarily profit because that's a very observable thing to do something richer. Maybe you have to look at some combination of the data in a novel way. And I think the same, very much applies to businesses, which is where, so many companies that are doing this.
[00:26:18] There are real practitioners who say, okay we have some solution that ultimately looks like empowering people to one get access to, and then ultimately use a vast amount of data. That's ultimately a very, trained practitioner who can put together and make a useful insight out of.
[00:26:36] Nima Gardideh: I assume you're talking about sort of data lakes being the more raw versions of the data and warehouses being more computed versions of it. exactly. Historically from a cost perspective there were a number of almost more complex strategies where you would put low value data in a certain place and high value data in say the warehouse, because Snowflake is very famous. They separated storage costs and [00:27:00] compute costs and that ability to sort of, separate, these basically makes the economics of your data strategy when you're not at crazy volumes very different.
[00:27:11] Peter "Fish" Fishman: there's a high premium on simplicity there.
[00:27:15] Nima Gardideh: Gotcha. So it's been simplified to a larger extent by just these business models query, I guess does the same thing
[00:27:21] where they charge you by compute. And then, so we can skip a couple steps that I'd love to hear about before you started Mozart, you had a few other stints at a few of these tech companies, but then you kind of went out of the pure data. why did you decide to go beyond just data and you were still leading data teams and then eventually were at the C-suite. So I'd love to hear your journey about that too.
[00:27:43] Peter "Fish" Fishman: Sure. So I had. Uh, a few stints after zes before starting my own company, where I worked as an executive at a number of tech companies. So I worked most notably at both OpenDoor and ease.
[00:27:55] These are two marketplaces. One is a real estate marketplace. One is a cannabis marketplace. [00:28:00] So the context being very different, but some of the challenges being pretty similar and at, both benefits and OpenDoor and Eaze, there was a big career progression into a little bit less Data reporting and more sort of operations with a data focus.
[00:28:19] So I think a lot of folks in the data space ultimately get a little bit tired of being the sort of conci of the decision maker that they wanna sort of have greater agency in driving the result. So typically, as a data person, you're trying to unearth a strategy that will give you some sort of advantage in the market.
[00:28:43] And when I was at Playdom going all the way back to my roots as a data analyst, I worked with the marketing team to try to drive what was the highest ROI investments on Facebook. So you would make Facebook ads and some novel population, for us, there are always these very wealthy….
[00:28:59] Nima Gardideh: Yeah. Gaming has power laws, right? It's like a very small percentage of the whole population makes all the money
[00:29:04] Peter "Fish" Fishman: Yeah, it was generally called whale hunting, this is a freemium game. 98% of players never paid a cent and then a handful of players would really fund your entire company.
[00:29:14] Nima Gardideh: Yeah, those companies are having a lot of issues with apparatus at Facebook and iOS 14, unfortunately it is much harder to whale hunt.
[00:29:20] Peter "Fish" Fishman: I can only imagine it just kept getting harder. But the insight that I had as a data analyst was that I paired very closely with the marketing team and the marketing team, had goals of driving a certain number of players or users. And I really wanted to transition that to driving more lifetime value.
[00:29:39] So rather than thinking about players, think about, different players had very big differences in lifetime value. unsurprisingly very wealthy countries, obviously like the United States, but also say like some of the Scandinavian countries had great LTVs right. They're just wealthier. They can spend more on like silly virtual goods whereas buying eyeballs that would ultimately [00:30:00] turn into players that would ultimately turn into players in sort of lower cost countries from an advertising perspective seemed like a great strategy, but maybe wasn't a great strategy for the company.
[00:30:10] It's that ratio that really matters. So as a data analyst, you sort of unearth the components of that ratio, you solve for what the LTV is for players, from, maybe a certain geography you solve for what the C is. And then you look for the optimal LTV to C ratio. You know, at the time I would say cutting edge approaches are a pretty standard approach these days. Sure. And actually, what's kind of interesting is that ultimately CAC has actually recently been, I'd say almost not muted, but turned down a little bit given the economic situation. The challenges of the current economic environment have caused like interest rates to be effectively, much more uncertain.
[00:30:49] And as a result, people care a lot more about payback than they do about say LTV to CAC. Both are great measures of efficiency, but your payback period is much less sensitive [00:31:00] to your assumptions about say an interest rate or how much you care about today versus the future.
[00:31:05] Nima Gardideh: Yeah, you're just not willing to wait 16 months for the LTV to be realized anymore. Effectively you have to maybe care about a three month return rate or something much closer to that. tell me about Mozart. So first of all, disclosure is that we're users. We pay for the product. We're big fans. How do you distinguish. cuz to me, and maybe I can describe it to you first and then you can tell me if I'm right. And the reason we use it is effectively like a combination of DBT plus Fivetran that I don't have to split up my own DBT instance and maintain it or anything like that.
[00:31:36] I can get access to the first level Fivetran and pull my data in, do the transformations before it gets into my warehouse. And then build my visualizations or whatever I want on top of that data which we use other tools for. And I was very excited to use it. we've been using it for, I think since you guys launched out YC did I describe it right?
[00:31:55] Is that the right area of the market you've decided to go into? Is there like a [00:32:00] long term version that I don't know about? Tell me a bit about how and why you decided to get into this space.
[00:32:05] Peter "Fish" Fishman: Okay, well, I think you described it right and wrong. So first of all obviously we love our customer base, so thank you for being a customer. And you did rightly remember that you found us during Y Combinator. We were summer 20 we connected through the Y Combinator community and we love obviously supporting startups and YC startups and all the things.
[00:32:26] that detail's correct. And then you're also correct in saying that our product does a number of things in this sort of data pipelining space. And it is true that we leverage some really great technology partners. We wanna be world class and best in class. So we do partner very closely with Fivetran and Snowflake.
[00:32:46] And what you have almost out of the box is a world class data stack that you can be up and running with in under an hour. So you've sort of hit at the highest level, but of course I'm a little bit [00:33:00] biased. And I would say that we do owe so much more than simply combining just the El and the T and the data warehousing.
[00:33:08] So the highest level description is yes, that is in fact what we do, but in practice, what we see ourselves as is partners in your data journey to get you up and running as quickly as possible and have the UI
[00:33:21] be very intuitive to a practitioner. A lot of what is considered the modern data stack is not just a bunch of stitch together tools.
[00:33:28] So this is my world. I think about the modern data stack all the time. In fact the term, the modern data stack has become passe in my world, but what it was, or what it is in my mind is this giant collection of logos that great teams, like the Yammer team that I mentioned, but many others used.
[00:33:49] Build this extensive amount of tools. And there were great sort of optimizations that these teams found, because whether it was data analysts, whether it was business users, [00:34:00] whether it was data engineers, whether data scientists, whether it was maybe now analytics, engineers, or wasting a ton of time doing very rote things.
[00:34:08] And now, tools that these companies paid, millions or tens of millions, or even hundreds of millions effectively to build internally, a lot of the popular tooling came out of say Airbnb. And why the heck did Airbnb which is obviously in the travel space waste all this time, building all this data, infrastructure tooling.
[00:34:27] They felt like there were optimizations to be had for their business. And it was well worth the sort of distraction of building data tools. Now, many of those data tools are popular sort of more broadly and companies get built out of these teams that sort of are leveraging, Hey, one key problem that we solve that company X was, but what you end up with is this giant, like display of all these different logos.
[00:34:52] That's confusing, that's daunting to your average operator. So your operator that's savvy [00:35:00] knows what they want. They know that they want these tools to get this result, and they know that they need these tools, but the idea of negotiating a dozen contracts, technically spinning them up, making sure that they all like, work together is a challenge.
[00:35:13] And we saw that as a tremendous opportunity. The sort of idea that step four or five of the company is to negotiate 15 deals with data vendors is just absolutely ridiculous. And we wanted to make a world class offering and have that be available to companies. Really just incredibly quickly in a combination of time effective, cost effective, and just giving them the ability to focus on their core business.
[00:35:38] Nima Gardideh: It feels a holistic approach whereas, let's say. Technical marketer or a CTO of a company that needs this apparatus set up for their product and marketing teams. I can come in and you'll provide everything I need to then power those teams.
[00:35:53] Peter "Fish" Fishman: Yeah, absolutely. And that's not to say that you couldn't do this yourself,
[00:35:56] so there are many great products that I can't make a Coca-Cola myself, if I did, it wouldn't taste quite right. That's not to say that you couldn't do this yourself. But it's just really inefficient.
[00:36:06] For you to do this? I love gardening. I have a garden, you know, and for me to make a tomato it's like, months of work you know, all this watering that I go do, couple times a week and you have to pay a lot of attention to it.
[00:36:17] And then off the vine comes a very delicious tomato that I could have just gone to the supermarket and purchased, probably for, fraction of the cost. There's some joy in gardening for me but if you don't take joy in building data infrastructure this is largely a solved problem.
[00:36:33] And, you shouldn't be sort of working on putting together the perfect tomato. You should be doing the things that you're really great at as a business.
[00:36:40] Nima Gardideh: Yeah, this makes a lot of sense. And it immediately comes to my mind that what your target audience is going to be is just the earlier stage companies that do need this. Don't have the resources and don't want to put in the resources to build this whole stack out. And so you said series Aish. Is the ideal customer profile.
[00:36:58] Walk me through how [00:37:00] you've been thinking about growth in general so far, You're a couple years in so far, how is, how have you been doing the growth? Has it been inbound, outbound sales? What is the current structure? And do you see yourself going beyond this ICP or like early stage companies?
[00:37:17] And are you just trying to grow with them? Are you trying to go up market at some point? Yeah. So growth certainly matters. So if you're a startup your growth, one obviously is, if you're making a margin it, effectively, fuels the business from a cash flow perspective. Two, you know, it signals that there's a strong market there, so it attracts capital.
[00:37:36] Peter "Fish" Fishman: So if you want to invest in R&D your growth matters a lot.
[00:37:41] Peter "Fish" Fishman: The one thing that has changed, especially in the last six months is that growth. Isn't every, you know, it used to be the only thing. now it no longer is. So now efficient growth matters a ton. historically what we've done for growth is everything.
[00:37:56] So what everything means is, I do. [00:38:00] Podcasts and content. And I post to forums where startups live like YC Bookface and we have an outbound team and we have an ads team and a growth team, and we try to do some clever things. In addition to all of that you wanna be everywhere.
[00:38:17] There's sort of a principle that comes from CPG from potato chips, which is, you always want to be somewhere around a customer once they have a hankering for that salty taste. We really have the challenge at Mozart is we're really ideal as you start to transition to your sort of long term data strategy.
[00:38:37] So as you wanna bring in your initial Cloud data warehouse infrastructure. So as you transition off of leveraging the information inside of the SaaS tools, or as you stop querying your database, which you should pretty much do, like right from the start as in stop doing that.[Laughs so we have this sort of moment in time where we provide an incredible amount of [00:39:00] value for companies and we have to be there at that moment.
[00:39:02] Peter "Fish" Fishman: Today, I would say a couple of things. One. Yes. Ultimately the goal is to grow with companies. So an attractive company is one that's able to retain its customers because they love them.
[00:39:14] And typically they grow with them. So we've had the good fortune of having, I think half a dozen companies become unicorns since getting on the Moza platform, which I jokingly call causality. But what that means is, you know, we have two person companies that were in our YC batch all the way up to unicorns or decacorns.
[00:39:34] So we really want to be able to scale now, of course, Snowflake itself, some of the underlying technology partners skills to the biggest companies in the world. And we wanna be able to do that as well.
[00:39:44] I think the story kind of, like you said, is, for a lot of people, growth means retention.
[00:39:49] It basically means retention plus expansion. And we do that really well as a company. Data is a very addictive drug. What we find is that companies that start [00:40:00] consuming data really love it and start doing more of it. There's just this natural growth in consumption of compute.
[00:40:06] And this is sort of the underlying business of Snowflake. But what we also find is to grow, it's not the same as, say like, a lot of consumer products. The way that I found out about Uber is that one of my coworkers kept raving about how much better it was than San Francisco taxis.
[00:40:19] And then I, in turn, got it. I don't know that you're out at dinner one night and you're just bragging about your data infrastructure. So that's, you know, a challenge, which is how do you make your product viral? People think of it in terms of the growth hack in terms of the channels, but in practice, a lot of what growth is is virality in your customer base and expansion in your customer base much greater than your turn in your customer base.
[00:40:43] Peter "Fish" Fishman: So you do have to focus on the virality and the expansion, just as much as you focus on what's the latest channel that produces the magic.
[00:40:53] Nima Gardideh: what are you doing to track that in the B2B world that has just got, I assume is gonna be harder. First of all, like what are the top metrics you [00:41:00] look at on a weekly or monthly basis to care about these things.
[00:41:10] Peter "Fish" Fishman: First off, let me say that. I think we do a great job of monitoring our metrics, leveraging the platform Mozart Data.
[00:41:11] Nima Gardideh: I hope so! [Laughs]
[00:41:13] Peter "Fish" Fishman: We are one of the best customers. So I think we use it in a very sophisticated way. We have a program that we call Mozart. Mozart, which is basically how we use Mozart Data and then try to share that with the rest of the world, because we do a lot of savvy and sophisticated stuff with our data infrastructure.
[00:41:30] One, we are a B2B business and typically our sales cycles are monthly. So what that means is we pretty much look at the data on a monthly cadence. So I don't wake up every morning at 7:01am, check my dashboards to make sure that our business is going up right.
[00:41:48] That is not what it looks like. What it does look like is that we have a set of monthly metrics, the most important of which I think is NDR. NDR stands for net dollar retention. Some people call it [00:42:00] NRR net revenue retention, but ultimately it's the net of expansion and churn.
[00:42:04] So when your churn is greater than your expansion this is less than a dollar. That's a bad thing. That means that your business is inherently shrinking now most businesses inherently shrink. you use it and then you use it slightly less over time.
[00:42:18] A great business has expansion. It means that the customer comes in and then constantly uses more and more. So then you lose the customer. That's a business metric more important than that is you want the usage to be expanding as well.
[00:42:29] that's something that we are, just as focused on, are our customers getting a lot of value out of the platform, UMAS, ultimately that's something that they will pay us for. it's not just that. I want you to have a fancy data infrastructure. I want you using that compute.
[00:42:44] Nima Gardideh: It's almost like a leading indicator towards more revenue.
[00:42:47] Peter "Fish" Fishman: Exactly.
[00:42:47] Nima Gardideh: how do you look at that information? There is a composite metric: there are your customer service managers, or I assume you have some CSM team caring about these types of things and product teams. How do they look at it?
[00:43:00] Peter "Fish" Fishman: We actually don't have a traditional CSM team. We have a data analyst team that is obsessed with the customer. and what that means in practice is. The person that's in charge of understanding whether you're using your compute would know great ways for you to get more out of the platform again.
[00:43:20] One thing is to put out content on Mozart, Mozart that says, okay, here's how we use it. Here's a report that we use. Here's a report that we look at. Here's a join of two common SaaS tools that we use that we see a lot of our customers using. All of that is very helpful, but it's much more helpful to have a partner in crime to give people a push in the back.
[00:43:39] We've had a number of webinars where we've had customers come in and show us what they're doing. And it's, it's always like blowing me away. I'm like, have you thought about doing this then the next slide is, 14 things past that.
[00:43:50] I always really love it when, the image that I have and that I've seen in the data world is if you give somebody a push in the back on a swing, they start going and they're [00:44:00] able to swing, more than you ever imagined. So it's not the case that that's universally true.
[00:44:04] Sometimes you give a person a push in the back and then they just stop and never really get the motion down, but the ones that do get the motion down it's incredible. And you have to do very little for them to get value from the platform after that. There's just a, sort of a human default state, which is inertia and, doing whatever you can to break that I think ultimately leads to great outcomes for the person on the swing, as well as the pusher.
[00:44:30] Nima Gardideh: And so this, these data analysts are sometimes customer facing in that they help the customers and the users basically take advantage of the platform as much as possible effectively. When you're trying to look at usage data, though, going back to the original question: how do you look at it? Does the head of that team look at it? What is that metric?
[00:44:49] Peter "Fish" Fishman: Ultimately we're selling insights and actions, right? So we're selling companies making better decisions. Well, It's hard to like to put that in a box or even have that on a graph or a chart. Oh, made like 14 better decisions today that were 72% better than the past decisions like that doesn't really happen.
[00:45:08] The standard way of looking at usage. In the warehousing world tends to be compute and in the sort of data ingestion world tends to be rose ingested. We sort of have old school metrics in that respect. We haven't reinvented the wheel on what using data means.
[00:45:26] We look at rows and compute, but then we don't have a complicated formula, like the root of rows times the square root of compute times, like some like constant of success, it's like we don't have a formula for looking at it, but we wanna see companies using more and more compute.
[00:45:41] And that typically means that someone's consuming it. It doesn't mean that they're doing anything useful with it, but the fact that they're doing something with it tends to mean that somebody wants to do something with it. There's that desire to find that insight And that's a really good sort of starting place, a good like 10,000 foot view as to whether something useful is happening.
[00:46:00] Nima Gardideh: Yeah. It's kind of like movement towards the potential thing that you want to be looking for. It's another one is leading things. It makes a lot of sense, especially in your world. One of the things I truly believe in when it comes to growing companies is that you should not necessarily put in an organizational model together that you may have read about in a book that you think makes sense, just because that is the conventional way a company is built and this data analytics slash customer success team that you just described as the perfect example of here is a problem we're trying to solve.
[00:46:33] Here's the best way we're gonna solve it. and you've gone after and done that. What are some of the more nuanced things about your business that you think have already happened? I think this is a good example. Where do you see it going, where you have to constantly change the team structure and the organizational structure to solve?
[00:46:49] Let's say growth and retention and growth to me is retention. And I think this is something that maybe has been muddied in the past few years of cheap capital that has become [00:47:00] Growth is about this top line number that we keep talking about. It's really about money and money out.
[00:47:05] And is that in the positive direction or not? And I think we've just not done enough work as marketers and growth people to instill that within founders. And it's so lovely to see that you understand this, and I've had like maybe 10 conversations with founders over the past two weeks about their
[00:47:21] recession, and we'll get into this, people are still not woken up to what's happening in the macro markets. And I wanna address that, but before we hop into that topic how do you think about org organizational design? How did this team come to be? Do you have some form of a ritual internally to think about how to solve for these problems?
[00:47:39] Peter "Fish" Fishman: Growing up in Silicon valley over the last, decade and a half, I would say that there's been a lot of energy and investment into different organizational design and different organizational design principles and sort of how information flows and all of this kind of has been a well studied subject and something that we took really seriously.
[00:47:58] I remember most at Yammer, like it was something that we revisited all the time. We are a small team, so we're about 20 people. The organizational design is not that complex. We have basically Dan, my CTO and co-founder that runs largely the product side of the org, the product and technology side.
[00:48:16] And I run the sort of business side of the org so starting, there at the highest level there's nothing particularly special. We're selling a I like to think of it as a vanilla B2B SaaS product. And that's got some elements of product led growth, but ultimately it is a pretty vanilla sort of business and reinventing the wheel here feels and looks wrong.
[00:48:39] So we don't wanna redefine how companies structure and organize themselves. We don't wanna redefine how we get paid. We don't want to redefine how we do pricing and billing. We don't wanna redefine. Some outbound motions. We're not looking to sort of redefine the wheel there. We wanna obviously redefine the wheel about how people get access to and consume data.
[00:48:59] [00:49:00] So we're focused on that. Wow that said every company is unique. It's about people and some of it's reflected in our core values. I think we are a company that takes a very 80/20 approach and is very intentional about the parade principle. So a lot of, sort of data products, we feel like 80% of the value can be delivered with 20% of the effort.
[00:49:21] Whereas there's of course, some data products where you have to do a hundred, so you have to really make sure that you're delivering the perfect experience. There's this concept called Satya or truth.
[00:49:32] My partner is a, yoy. So she is always preaching about this. And it's infected the way that we think about the company that we want people that can be really incredibly honest with themselves and what's going on with the company that we want and then last, we have a company core value that I call Kyle Singler.
[00:49:49] So Kyle Singler was a college and professional basketball player. I was a big fan of his in college. He played on my favorite team and he played all of the different positions. So he played the one, the two, the three, the four and the five. And they asked the coach, what position does he play?
[00:50:04] And the coach said he plays the position winner. So we have basically a similar sort of principle that, when you're at an early stage, you end up having to do a lot of different roles. So these sort of core values very much dictate some of the uniqueness of the org structure that said, like diving into some of the specifics of your question.
[00:50:20] Yes. The CSM role is much more data centric and has a much higher bar for data skills and capabilities than a CSM role at basically any other company. There's a few others that I think are important. So one, we are a big believer in hiring for strength. The org structure is a little bit less central, like who reports to who is less important to me. We have a pretty simple sort of org structure. We have basically, business side and we have an R&D side. Now it is important that there is this sort of. Communication circle that happens between them so that we can build, to go all Y Combinator on you to drink the Kool-Aid a little bit, you have to make stuff that people want
[00:50:58] and that's, making sure that you get that loop working right is very important. But we're not reinventing the wheel there in terms of the names of the positions. I think the hiring approach, again, we hire for strength. I read about this in Ben Horowitz's book, “The Hard Thing About Hard Things”, a lot of companies will try to sort of make sure that they hire folks that meet the bar on many dimensions that they care about. And that's not something that we necessarily do at Mozart. There are some dimensions that are table stakes, but beyond that, we try to find what is somebody great at?
[00:51:29] And can that greatness solve a need of ours. And then, the other thing is we do like to ask concrete problems. So less open ended problems that we think really do tell us, Hey, can this person do this particular task? So these are two sort of hiring approach differences that I think, translate to the broader organization.
[00:51:50] And then beyond that, I think a lot of the sort of efficacy of org design is dictated by the core values and how well they're internalized across the company and how well you actually live them and reward them.
[00:52:04] Nima Gardideh: Yeah. How much do you think is just, you're basically self-selecting people that are already going to like those values and live them through just their sheer personalities that match them. And how much of it do you think you're nurturing in the org?
[00:52:19] Peter "Fish" Fishman: Nice. We come back to the same question, which is how much of me was me and how much was it based on the years two to 10, right? I do think that there's a massive selection into folks that wanna be at a small stage startup. So there are a number of drawbacks to being at an early stage startup.
[00:52:37] The first is honestly some element of job security, which is to say, when you work at Google, you know that Google is going to exist tomorrow. I worked at Microsoft for three years and you might say, is the sun gonna come up tomorrow?
[00:52:49] But you would never say is, Microsoft's gonna be there tomorrow. Of course, Microsoft's gonna be there. you would take somebody who's inherently risk seeking in their career. The upshot of working at a great startup. So you get obviously a huge amount of agency, a huge ability to, put your thumbprint on the team, on the work et cetera.
[00:53:07] And then I think the biggest one is, you get a lot of challenges thrown at you rapidly. So you have that ability to have a lot of rapid career progression. Obviously there's some bias here, but for the companies where it works out, it really ends up being a great and transformational experience, from an intellectual perspective, from a monetary perspective, but mostly from a professional growth perspective and confidence too.
[00:53:30] Nima Gardideh: Yeah, I mean, that, that's sort of how I got into startups too. It was just a great learning experience. Even if there was no monetary outcome after a few years of doing it talking about job security, let's talk a bit about the recession. I think it's been something that's been on my mind for the past couple months.
[00:53:46] Our business has been affected by, to some extent, our clients are now changing their way of thinking. some of them more than the others. And you were mentioning how earlier, before we started the pot, [00:54:00] That you think there are ways to think about how to approach it with data a little bit more than maybe people just are right now.
[00:54:08] Let's start with a baseline recommendation to other founders and then how you're thinking about it as an early stage startup that is recently capitalized and what you see in the next, let's say 18 to 24 months.
[00:54:21] Peter "Fish" Fishman: I don't wanna make any macro forecasts. I sort of often joke that my favorite lunch is a Costco chicken bake, and that recently had a 33% price inflation just last week. And that means that inflation's real. And that means that the chicken bake itself doesn't improve it.
[00:54:37] But I think to level set. I think all founders should be in a place where they recognize that interest rates will hike and anybody with future cash flows. startups tend to be ones where their positive cash flows are in the future, then effectively have a lower present value.
[00:54:52] So raising at sort of astronomical prices that we've seen in the last sort of two years is sort of, a nice pipe dream, but [00:55:00] it's, that window has passed. So level set, first of all, have a soft macro perspective doesn't mean, you're gonna say, okay, I'm calling the bottom is, October 4th, I don't think that really what you need to do, but I think you need to, getting back to this concept of such yeah.
[00:55:16] Which is the truth is that it might be a few years before there's a real one positive fundraising environment. There's these concepts, default alive which, is a pretty tough place to achieve as a startup, unless you go super cockroach mode to go again into the Y Combinator vernacular or being what Craft would call default investible, which means like really excellent metrics across the board, such that you could raise in a, a strong funding environment or a weak funding environment.
[00:55:45] I think, the first thing is again, level setting. The second thing is to have some sort of understanding of, where you are on the spectrum of very strong metrics with very strong opportunity and, investible in, in headwinds or tailwinds. The last is I think that generic advice really fails here in the same way that I felt like a generic org structure was gonna work for us, but generic advice about where we are is very different.
[00:56:10] So we recently were capitalized. So we're in a very different position than a company that has maybe six months of runway or even less similarly, even a company that has 18 months of runway. That's a very different position. You could have tens of runway and the optimal strategy can be to really, go all Y Combinator cockroach on it and really try to, use that money to its fullest, to maximize your R&D opportunity to find your product market fit.
[00:56:36] So it is not necessarily the case that if you have. A large amount or even infinite amount of runway that the right strategy is just to burn through it. Frivolously. I think whatever was. Sort of the norm on growth multiples, like that's dead and a past world and you shouldn't be optimizing for it.
[00:56:54] Internally we call this faster horses, which is to say, don't seek the types [00:57:00] of metrics, the types of inefficient growth that people valued a year ago. There's the famous Henry Ford quote, or attributed Henry Ford, which is “If I asked my customers what they want, they would've said faster horses.” and the same sort of applies here.
[00:57:14] Don't be sort of a horse trader in Detroit in 1909. In fact, you wanna instead really focus on what are the efficient channels for growth. One measuring efficient channels for growth involves, I think, a lot more complexity than, say, just looking at an individual kind of channel and report from that channel, et cetera.
[00:57:33] That's where sort of data pipelining fits in. But I would say most of the metrics that you have to look on for efficiency are much more nuanced. So again, going with Craft Ventures they espouse the burn multiple, which is not just understanding how much are you growing, but how much are you spending to achieve that growth?
[00:57:50] And they sort of lay out ranges that look good or great or bad, depending on where you are in your company, life cycle.
[00:57:59] Nima Gardideh: they [00:58:00] They have like a literal chart for all of that, right? At Craft where you are, like the most investible. If you fall into this part of the chart for all these metrics.
[00:58:07] Peter "Fish" Fishman: Yeah. In fact, I asked Ethan Ruby who is an analytics partner at Craft. I said why did you publish that? That's your secret sauce? And he said, this is service, it's a view into at the highest level, how they think about companies and opportunities.
[00:58:21] But that's just step one. There's a much more rigorous way of analyzing it, but at the very least you can know, okay, am I in the game with these metrics.
[00:58:31] Nima Gardideh: And we'll make sure to actually link to that blog post in the description, because I, I do think they're doing good work at telling founders that like, Hey, this, regardless of the economic macro environment. This is just what a good SaaS business looks like. And you're likely going to be able to raise money if you're in the top echelon of those charts. and I think that's important. Let's talk about your sliver of the world. Like you're capitalized, over a million ARR, are you trying to break even and being default alive mode?
[00:59:00] Are you trying to grow at a certain rate that still looks good, but don't want to reach default alive before you have to fundraise again? Like what are you thinking?'
[00:59:08] Peter "Fish" Fishman: If you read Paul Graham's essay on being default alive, the reason to be default alive is not to be default alive. it's not that, this then makes your company able to go on for years in an okay fashion. The reason for being default alive is that once you're default alive, you can then sort of reason about what investments are worth making versus not.
[00:59:31] Whereas if you are not default alive, you have to figure out everything you can do to get to that state. Yes, we are very well capitalized, right? So we have a long runway and we can do whatever we want in that respect. That's not the plan, but I think it is certainly the case that we could choose to spend all that money inefficiently, we could throw a giant party and hope that party like is a ton of fun.
[00:59:58] And, you wake up with a hangover and no money in your bank account. And then, everybody cries and goes home. That's obviously not what we want to do. And then similarly, stretching it to, just sitting in a bank account and earning a few bips in my checking account is also not what we want it to do.
[01:00:13] There's some sort of mix of growth that we had budgeted for and growth that makes. Unambiguous sense. Whenever you're growing as a company, some of that growth is beautiful and free. The viral parts, the word of mouth there is sort of earned media.
[01:00:31] There's all this free growth that happens. And then there's all this, like the whipped cream on a sunday, which is to say the growth that you really have to pay for and is, marginal. In fact, like when you're looking at ads channels, the ads channel return tends to be very concave.
[01:00:46] So really chopping out the sort of least efficient parts of your growth. So we are still looking at growth as a metric. It's not our defining metric. We raise the money because we think there's a tremendous market opportunity and we have the ambition to go for it.
[01:01:00] One of the things is that everything in a downturn actually gets easier with the exception of selling luxury goods and raising money to the extent that you're selling a frivolous thing or something, that's just like a nice to have. That means that your company's gonna face headwinds.
[01:01:19] Again, going with the YC bit, we wanna sell stuff that not just people want, but people really need. So we wanna sell data products that we think are essential and core to these business operations. And we think that will be more recession proof. And then last we wanna plan not to raise money for the next, three plus years.
[01:01:39] What that means is, again, stretching our runway and figuring out what our frivolous spending that we don't think is appropriate in this environment.
[01:01:49] Nima Gardideh: Yeah, I was gonna immediately ask you what your timeline for fundraising is gonna be in three years. Sounds very good to me. It's been something I've been thinking about and talking to some founders about as well. The thing that stood out to me in this painkiller vitamin world, I think is very obvious, to a lot of people, but the part that I think people forget about and maybe actually there's some folks that probably forget it and are becoming too frugal.
[01:02:13] And there are some folks that assume that this is going to be their reality and they think they can fundraise again in 12 months and it's gonna be all good and dandy. Is this part of what you just said is that actually everything gets easier in a recession to some extent, especially when it comes to growth, because lots of people drop their budgets, costs will go down for acquisition and things like that.
[01:02:34] And the second factor is that you may be comparing yourself to these behemoths that I have already found. Saturation in their marketing apparatus and their acquisition of the overall market. And they're going to see a downturn in that they're gonna lose revenue, but if you're a startup at almost any stage, anywhere between seed and series E I bet you that there is a part of the market that you can capture and grow in.[01:03:00]
[01:03:00] That is pretty much available even during a downturn, you just have to find it and grow in it. And that's how I think about it in our world. We obviously help companies grow and we haven't really seen them slow down other than cutting their budgets slightly, but CPMs have dropped at the same time.
[01:03:16] So we've been able to be quite efficient because of it. And this combination is quite helpful for startups. So if you can survive for long enough, you may be one of the folks that ends up being. The winner in five to 10 years from now and being one of the tech companies that everyone talks about, how they spread out of the last recession.
[01:03:33] I'm sure you've seen all those titles of all these companies that came from 2008, 2009. And there's plenty of opportunities for this cohort of companies to be those companies as
[01:03:42] Peter "Fish" Fishman: Yeah, the cliche is that the strongest steel gets forged in the hottest fire. And you mentioned, one dimension, which is growth channels become less competitive and ultimately these growth channels are a second price auction. So as second highest bidder, shrinks their willingness to pay, this gives greater opportunity to the highest
[01:03:59] Nima Gardideh: They're no longer second price auctions, but they used to be.
[01:04:02] Peter "Fish" Fishman: Sure. So there's some variant on there's some variant on something that looks like that. And then the same goes true for hiring, which is the sort of market for talent. The talent wars also get less competitive as big company equity packages look less attractive.
[01:04:18] It just sort of gives an extra economic incentive to taking that risk.
[01:04:22] Nima Gardideh: There's a lot of talent locked up at these bigger companies. As the layoffs happen, as unfortunate as they are, they do open up a talent pool for companies for early companies to work with these wonderfully talented people. Anyway, Fish, thank you so much for coming on and talking to me about your company and your past super fun.
[01:04:41] Is there any last words you want to tell founders, marketers, listening to this about what the next six to 12 months is gonna look like? Whenever I have this conversation, it feels a little scary to folks. And I'm glad we started talking about the optimistic parts, but maybe leave them with a couple of words and we can call it.
[01:04:56] Peter "Fish" Fishman: I think the sort of opportunities to build a great business and to leverage technology are sort of never ending. I hope that everybody's sort of listening takes the heat of what's going on in reality. But again, thinking of that as more of an opportunity. I did mention at the start that my career actually started in football analytics and at the time baseball analytics had taken off that all the teams were leveraging analytics in baseball.
[01:05:24] And one of the reasons that I wanted to do football analytics was because it wasn't a solved problem. So again, thinking of the opportunity of the new challenge rather than all of the things that, you know, think we started off with my PhD, but humans have this major loss aversion.
[01:05:39] They hate feeling like something that was there is now gone. That's certainly the case in the current economic environment, but focusing on the opportunity is something that we're doing at Mozart. We're obviously helping a lot of companies leverage the challenges of more nuanced metrics through our data platform.
[01:05:56] And I hope that everybody can embrace the new set of challenges from here.
[01:06:01] Nima Gardideh: Awesome. Thank you so much.
[01:06:04] Peter "Fish" Fishman: Thanks, Nima.
[01:06:05] Nima Gardideh: I hope you enjoyed this episode. Please spend a minute rating this episode on Spotify, Stitcher, apple podcasts, or anywhere else you're listening to it. and if you want to share your hyper growth story with us, email me firstname.lastname@example.org to be our guest. Email is in the show notes.
[01:06:21] I'm super excited to have Deepak Cchugani, founder of Nuvacargo on our next episode. Tap on that follow or subscribe button to get notified as soon as it's released. Thanks for listening. And until next time.