Error Title

This is a notice message, displayed at the top of the browser, informing the user of something useful.

Continue with LinkedIn
Recover my Password
Submit your Tekpon Account E-mail address and you will receive an email with instructions to reset your password.

Data platform for centralizing, organizing, and analyzing your data

Data platform for centralizing, organizing, and analyzing your data | Peter Fishman - Mozart Data

About Mozart Data

Peter: Mozart Data is the easiest way for teams to spin a modern data stack. That means that without any data engineering, teams can get started on world-class data infrastructure and focus their energies and efforts on deriving insights from their data instead of worrying about how to data. The biggest problem is that data engineers are very expensive to hire, and a lot of the work they do is solved at a grander scale. Moving data from SaaS tools or databases into a central warehouse is very much a solved problem and shouldn’t be recreated by every single time.

Mozart Data best features

Peter: Our best features are the ability to move that data reliably. And as schema changes, maybe upstream, we can sort of fix that on the fly and have a very reliable flow of data from your tools into a central warehouse. Most of the data people move from their production system, and you don’t want to be querying your production system. Instead, you want to be querying data and joining data together in a data warehouse. And most of what we do is going to sound lame, but it is moving data from one database to another database. The point is to have it in a powerful column nerd so that you can start combining data. One of the best features of Mozart Data is data analysis.

Often that data resides in different places. To do the really powerful segmentation that’s going to help you optimize your ads, that’s going to help you basically assess which features in your product are working versus not. When I worked at Microsoft, we had this term, one plus one equals three. But nowhere is it more true than in the data world, where you can get not just the value from one source but from multiple sources, and it sort of increases exponentially how much value you get out of it.

Mozart Data pricing plans

Peter: We don’t reinvent the wheel. We have a variety of packages that basically start from trials or free all the way up to folks that spend six-plus figures with us annually. It really depends on the usage. We sort of shine by really crafting packages together, bundles of both rows that you import and compute that you use to do analysis. We kind of bundle that together in a really compelling way. Also have really easy-to-understand packages for early-stage companies, and hopefully, they grow with us. But our pricing, for the most part, is usage-based. And the way that people use data is that they basically bring rows to their warehouse, and then they also sort of compute over those rows.

The story of founding the company

Peter: We started the company actually right at the start of the pandemic. Dan and I start started the company in April 2020. What’s very interesting is that Dan and I never saw each other in person for the company’s first year. Dan and I have been friends for 25 years and live seven miles apart. We are a remote-first company, but what’s crazy is that very often, when you think about kind of two founders, I guess a classic would be like Larry and Sergey in a tight garage in Palo Alto or something like that. But this wasn’t exactly us. We operated the entire time via meetings, which was a very different founding experience. It started almost three years ago, shortly after we did the summer Y Combinator program. We really took the company in a significant direction.

The first virtual YCombinator batch

Peter: We actually were the first virtual YC batch, and I had a great experience with YC. I actually still talk to my YC partners, and I think it’s one of the best shortcuts to starting a company. Dan is a multi-time founder, in fact, a multi-time YC founder, but I was a first-time founder, even though I had been in the startup ecosystem for decades. It’s one of the best cheat codes to have your company be a real, viable company.

That was a big part of YC, and I think the biggest sort of thing that I missed out on YC was the ability to interact with so many of my classmates. We’ve done some sort of reunions post, and I’ve been involved in the YC community around the Bay area of San Francisco since graduating. Now that in-person events have started to come back. It’s nice to see that kind of community.

The idea was not a eureka moment

Peter: Sadly, this was not a light bulb moment of brilliance. This was like 20 years of Dan and myself working in different parts of the data stack. Still, effectively figuring out that combined, we were like peanut butter and jelly and able to really build off of dance strengths, which tend to be backend in data engineering. And my strengths, which tend to be data analytics, and the company were effectively what we have built and would be making. It was effectively Dan and me in a box, so there was no exciting moment of aha and no sort of cliched-like eureka.

Instead, it was really, what would we be building? I think that that sort of trivializes a little bit. Many of the details are critically important. I think the way that Dan thinks about doing development, Dan, who’s our CTO. Still, the innovative product way that he thinks about doing the action is to test ideas and then get a lot of customer feedback and duration. Some of the early versions of the product weren’t really versions of the product. There would be an interface where someone would make a command, which would cue Dan to do something in the background. We were trying to learn what people wanted, like the YC cliche says, to make something people want. And we discovered what people wanted in their data infrastructure to get started.

Not raising in 2023

Peter: Last year, we raised our Series A, and it feels like a little while ago just because every startup month is like a year, but it was in 2022. There are no plans to raise in the short run, and certainly not in 2023. Hopefully, we can do much better than making our investors’ money last for  18 months.

One of the lessons of the past few years has been to be smart about your capital. Many firms compete to get into deals in a world where capital is free, and the interest rate is zero. I think you can develop a lot of bad habits. We’ve tried to steer a little clearer of the worst of those. But we are trying to build a venture-scale company, and we are trying to take take the opportunity that we see in the market. That certainly doesn’t involve raising this year, at least we hope. Obviously, we are trying to continue sort of our aggressive growth.

And what is your vision for the future?

Peter: The vision actually is not too different than the existing product. It’s only to like augment the value proposition and shrink the time to value. What we want to do is be the sort of no-brainer choice for early-stage companies that are trying to get their data infrastructure. Maybe they’ve got a greenfield situation, or they’re querying the Postgres database or their production system. And we want this to be a no-brainer. In the same way that you’d want to set up any system very early on, even before you, you might think you need it. We want this to be that option for most companies, and the vision is to really just shrink the time to value.

To get value from any data infrastructure, including Mozart Data, you must typically connect many sources, understand that data, clean it, and clean it some more, and then a little more. And then essentially display your first dashboard, time series, bar chart, whatever. There are verticalized solutions where you put in your credentials and then out pop a bunch of bar charts. Those solutions tend to have almost no time to value, right? Our goal is to evolve into something pretty similar where the time to value is excellent, but the system’s flexibility is very high. I think one of the reasons why we didn’t just replicate the verticalized solution is that so much of data analysis is very tight about cleaning your data and understanding what rows and columns need to be excluded from your analysis.

What is your story?

Peter: My story is pretty cliched for the data world. I would say a few things, and I’m a failed academic. I did a PhD in behavioral economics and empirical economics, really getting to work with giant data sets. At the time, those were millions of rows, which today is almost an afterthought or a joke of a large dataset. Mozart Data moves billions of rows or updates billions each month. I did during my dissertation at the time, I considered massive data, but in practice was very, very small. After transitioning from Academia, I realized I wanted to stay in the Bay Area, which often meant getting a job in tech. What I did was I got a job bouncing around startups where they figured out sort of a use case for data and some kind of interplay of data and economics.

For me, that started actually in video games. Dan actually also started his career as well as our first employee, John. Video games are a great environment for rapid iteration, learning, and very difficult learning because most games fail. Most of the games I worked on failed massively, and it was in that experience I learned to get a much more stark picture of what works and what doesn’t. And then, and then from there, I actually transitioned to B2B companies. And I’ve worked at mostly late-stage startups where they just didn’t have the tools to dive deep into that data and extract a lot of value from it, and I often built teams to do that effectively.

What’s your best piece of advice for founders?

Peter: I think there are different pieces of advice for different life stages, and so many of these pieces of advice are stolen from YCombinator. I’ll say some that are cliche, and some that are not so cliche. Let’s start with some of the cliche advice, talk to customers, and make things people want. These are Paul Graham’s canonical advice, and I think the biggest part is don’t lie to yourself.

I think some of the challenges for our company was fooling ourselves into trying to fit square pegs into round holes by finding people that would show a modicum of interest but would need a lot of edge case work just to make data analysis effective to them. And that was us lying to ourselves about having some sort of product market fit with that type of company. And as a result, we wasted a lot of cycles on it.

The other piece of advice typically comes around fundraising. That person is not interested when you know somebody is interested in your idea and potentially funding you, they have to be over-the-top interested to be truly interested. One of the things that YC does incredibly well is that it sets this artificial deadline called Demo Day. What that does and the power that it gives you is fantastic. I really wanted to close my round before demo day, or I really wanted to close my round a month after demo day or three months after demo day.

That’s like a fake deadline that the founders can create in their heads, but that deadline is compelling because it signals who’s in and who’s out. Because you have a limited amount of time, and the less time you can spend raising money effectively, the more time you can spend on your company. For our company, we lost a third of the team because we were focusing on fundraising. And losing a third of the team is incredibly expensive in terms of affecting the growth.

What’s your favorite software?

Peter: Of course, I will be biased and go backward in time and talk about my career. I spent a big chunk of my career at a company called Yammer. A sort of enterprise social networking software has always had a warm spot internally. Today we use Slack, and I think there are several really great stories about the growth of Slack.