How a Y Combinator Lesson Helped Our Startup Scale to a $19M Valuation [Founder Story]

Published in

Altar.io

18 min readAug 16, 2024

For the third instalment of our series of articles sitting down with successful entrepreneurs who’ve experienced world-class accelerators firsthand, I had the opportunity to sit down with Peter Fishman.

With nearly two decades of tech experience under his belt, Peter’s portfolio boasts impressive stints at Google, The Walt Disney Company, and a key role as Principal Data Science Manager at Microsoft.

Three years ago, he set out to build Mozart Data. An innovative startup that’s democratising the complex process of data management.

With Mozart Data, mid-stage startups can now access the same calibre of data platform that was previously only available to large corporations.

In this interview, he unravels some of the complexities of his journey. From the initial challenges of building a startup to his experiences and insights from his time going through Y Combinator’s accelerator program.

He also shared one of the most valuable things he learned in the program; the importance of talking to customers and doing things that don’t scale — which is something I’ve heard many times before (notably from Airbnb’s Brian Chesky and Linkedin’s Reid Hoffman), which means, as he puts it “working for customers, and working for future customers — often for free.”

This customer-centric philosophy, Peter reveals, has been instrumental in carving out Mozart Data’s identity and driving its success. As a Marketing leader, I couldn’t agree more.

It’s an interview packed with value that any entrepreneur, or anyone else for that matter, could benefit from reading — and truly highlights just how powerful simply talking to your customers can be — something most entrepreneurs forget too often.

About Peter & Mozart Data

Rui: So, Peter, you had multiple data-related jobs in your background, including Microsoft and Google.

I’m sure that you learnt a lot from those experiences. And then you set out to build your startup, Mozart Data which provides this “out of the box” modern data stack.

Can you introduce yourself and give us more information on your startup?

Peter: Sure. So, my name’s Pete Fishman. I go by Fish. I’m the co-founder and CEO of Mozart Data.

As you mentioned we’re an all-in-one data platform. But we like to describe it as “the easiest way to spin up a modern data stack”.

So take the types of tools and infrastructure that you see empowering data teams at large companies or late-stage startups and then we make that available to earlier-stage companies.

R: Ok, and it’s going well so far right? I checked you out on Crunchbase and you’ve raised $19M to date, is that still accurate?

P: I think it might be off by a million or two but sure, roughly that.

R: And what’s the plan? Are you planning on competing with Snowflake? Or are you thinking of more a niche, fully ETL approach?

P: So, Snowflake is a great partner of ours, in fact, we sit on top of Snowflake. We enable customers to use our managed offering. So we’re a big believer in a powerful central data warehouse.

The thing that we’re building is the easiest tool to execute it. So, not just your data warehouse and not just your solution, but also all the parts that come before and after.

Not only the bringing of data to that warehouse, and then also the cleaning up of that data so that you can visualize it and start making insights and build the company from there.

R: So you intend to carve your own space in this slightly different category. Do you have an exit strategy in mind?

Because I can understand how Snowflake would look at you guys and say: hey, we could integrate these guys into our offering.

So is there anything related to that in the vision or for now, you are just focused on carving your place in the industry?

P: A general best practice in my mind is to create a lot of value and then something good will come from it.

I don’t think that there’s any sort of vision of an inevitable exit to this partner or that partner or IPOing etc. Honestly, even though we’ve made a lot of progress in the last two and a half years, still feels decades, even eons away.

And I would say that we’re trying to, as you said, solve a really important problem for a subset of customers and for us, that’s the all-in-one solution.

That makes it easy to get going. And then there’s a lot of value that the data infrastructure creates in organisations, and then it’s about sharing and capturing some of that value.

Peter’s Vision for the Future of Data

R: That makes perfect sense, and I love this idea of focusing on creating value and letting the other things fall into place.

This is an industry where I think that we’re gonna see some concentration in the future, and I’m always curious when I speak with founders and people within the industry to understand what their educated guess is at the moment.

Do you think we’ll see a reduction in players?

Or is there still room for more people to come in because there are enough problems to solve?

P: My answer is both. So yes to each of those.

So it’s counterintuitive. I expect there to be a continued explosion of players and I also expect there to be some form of consolidation.

Today when somebody looks to buy the modern data stack they Google, “Tell me about data” or “Talk to me about the modern data stack” and up pops 743 logos.

Then, you need to piece your way through all of it.

First off, I’m a big believer in best in class, but also just how challenging it is to define that by company. And a lot of what we’re trying to do is simplify the choices for a given set of customers.

We think of piecing your way through the sort of modern data stack is a lot of wasted effort and energy — and often resources and engineering time, and ultimately money.

So we try to solve that. So on the one hand, I’m a believer in consolidation that may be in the form of the big players today, like Snowflake, Amazon and Google.

That might be in the form of some of the big slice players like, Fiveten or DBT expanding their footprint into more parts of the data stack.

And it might just be in the form of just more all-in-one type solutions, like Mozart Data or many of our competitors.

So there are many flavours that consolidation could take.

The next thing that you see is this crazy explosion in the last few years of things within niches of the data space. Maybe they’re just providing a solution for direct-to-consumer companies.

Or they’re not just solving observability, they’re solving it for a particular, slice of a company and a type of industry — and that trend was largely fueled by zero interest rates and there’s a lot of capital flowing into these startups.

But these are also businesses that are coming out of larger companies that had spent millions and millions of dollars building these tools internally. So actually I think that there’s still a huge amount of efficiency that’s happening at scale.

And while it’s true, there’s so many more today than there were say five years ago. I still think that trend is gonna continue, and I think that there is value being created by many of these companies. Even if some of it is redundant or competitive with one another, and also confusing to the end customer.

I think it’ll take tools like Mozart and others — and the sort of core parts of the ecosystem like Snowflake to sort through what is useful, when and where and for whom.

Interviewer’s note: There’s a reason why I asked my team to leave so much of Fish’s description of his startup in the article; For a founder to be successful, it’s paramount that they understand their industry and can properly articulate their value proposition. Fish’s answers are a great example of that, and signal an entrepreneur in a great position to bring in customers, investors and talent.

For another great example, take a look at this conversation I had with Wade Eyerly on how he pioneered subscription-based flying.

How the Idea for Mozart Data Came About

R: I can relate that with the MarTech ecosystem.

Tools like Salesforce and HubSpot created a whole ecosystem around them.

So many needs came because of the execution of those platforms and here it’s a similar story.

In terms of data, there are a lot of things that we still need to solve. And I believe we will see fewer projects maybe, but more solid business models. Or rather, more solid executions, especially under the current macro circumstances.

Let’s get back to Mozart. How did the idea come about?

P: So, my co-founder and I, Dan, have been friends for 20 years and he and I both have been working in the data space. Myself more like a data analyst and data scientist.

Dan is more of a backend engineer and data engineer.

And we ultimately wanted to provide the services that we were using at our late-stage startups to earlier-stage ventures.

Dan worked at a company called Clover Health. I had been bouncing around a lot of late-stage startups, building out data infrastructure and data teams and we wanted to make that available kind of more as a service.

So effectively overlapping our skillsets and finding our niche within the data ecosystem where we thought that we had a real perspective based on our experiences.

R: The pattern is clear. Spending some time within an industry detecting inefficiencies and setting out to solve them. It’s the most solid entrepreneurial path in my experience — in terms of setting yourself up for success.

Ok, I need to ask about the name. How did you guys come up with Mozart Data?

In your intro video on your website, there are some music references there as well. So I’m assuming there are some music-related backgrounds in there?

P: The direct version of it is that we had an idea of playing in the data orchestration space. Bringing the data all together, composing tables, etc. And then we were thinking “Ok what was resonating with us?

And we landed in a place of really wanting to do something clever that plays off this idea of data and orchestration. Of course, we landed on the greatest of all time, Mozart.

Why Peter Decided to Apply to Y Combinator

R: Brilliant. Why did you decide to apply for an accelerator?

P: So Dan had done Y Combinator 10 years earlier. He’d worked with Paul Graham and managed to get his company off the ground.

The first thing for us was that we had an idea and some experience, but I had not ever started a company from step one.

And a lot of times accelerators end up being nice cheat codes. They help you get from zero to one very quickly.

The thought behind it was exactly that. This idea that “It was gonna be a really quick path for us to get started.”

And then, found a variety of other benefits that we got from it.

The Application Process

R: How long did the application process take? So from filling out the initial application to acceptance into YC.

Also, was the process streamlined in any way because your co-founder had already attended, or was it the usual path?

P: So, a bit of oversharing, when we applied we were quickly invited to an interview. Not accepted into the program.

We did the interview and sadly got waitlisted for the program. So it’s a very, competitive program and we did not get in initially.

And Dan and I didn’t have a lot of maturity in our idea professionally, I think.

Our idea was not at a stage where it instantly clicked.

And many companies at that point don’t have an idea that will necessarily click. But they had given us some feedback and we proceeded to work pretty tirelessly over the next month before we re-interviewed.

And ultimately in that re-interview process, we had made some progress and we had talked to several potential customers and figured out a few different layers of what people wanted. I think was a compelling part of the story for them.

Our end-to-end process was almost three months in terms of application to getting accepted. And for some people, it happens like instantly. They apply, they get an interview and they get in. That was not our personal experience.

Although I will say the first time that Dan applied to Y Combinator, he got in on his first interview. But there are many ways to get in. Nobody cares afterwards how it happened, just that had happened.

We’ve had many successes and failures post that moment and I look at all of them as badges of honour in our journey.

The Y Combinator Experience

R: I’ve had a couple of these conversations already with some people that went through the same program you did. The pattern is that most of them weren’t successful the first time around.

But they persevered and some believe that the key factor for them to get in was actually because they attempted so many times that they’d shown that trait of a founder to never give up.

And then Y Combinator has its own way to help you shape your idea or even to give you a new one, right? Reddit would be a good example there.

Now let’s talk about what happened after you were accepted.

What did a typical week look like at Y Combinator?

P: So we did YC in the Summer of 2020. There are two batches every year, one in the winter, and one in the summer.

We did the summer batch which was, famously the first online batch, so it was at the peak of the COVID breakout.

Which meant YC experience was limited to this living room I’m sitting in now. I operated almost exclusively on Zoom for three months.

And that was a disappointing part of the experience for me because I’m an extrovert and I wanted to go to Mountain View and interact with all my batch mates. But of course, understandably, with COVID it just wasn’t possible.

That said, it also was incredibly efficient.

One of the challenges with Y Combinator is fitting all of the things that you have to do within three months.

There’s this arbitrary deadline, named Demo Day. And Demo Day is the day that Y Combinator attracts all of these investors. So you have to have your story in a really solid place in a relatively short amount of time.

And since Dan and I were very late to get into the program we didn’t have a running start.

We started working tremendous amounts of hours. To be blunt, we worked almost around the clock for about three months.

Part of that was, of course, building the product. But a big part of that was what I like to call “faking the product”.

Finding beta customers that believed in us for no reason and delighting them.

So doing the work, some of it fake, which is just me and Dan doing things that don’t scale and some of it being a product that we could then scale to more companies.

R: This idea of reaching out to people not to sell the product, but to get them to buy into the vision. Buying what you’re trying to build. It’s a solid method for bringing in early adopters.

What would you say was the best thing about your Y Combinator experience?

P: This may sound a little bit counterintuitive, but some of the best stuff has been after YC.

Y Combinator did a couple of things that it promised to do. So it promised to have a Demo Day. They did and lots of investors came that were interested in us.

But it also allowed us to go to other investors and say “Ok, here’s our fundraising round, it’s coming together — do you want in or not?”

And, of course, one of the things that becomes a lot easier when you go to YC or any other reputable accelerator is raising money — because they do the job of marketing to investors for you.

They go out and say “Hey investors, we have a bunch of startups that we think are worthwhile for investments — come and check them out, see what you think.”

And, for us, they delivered on that front. We raised a $4M seed round in big part due to Y Combinator.

It was also due to the attractiveness of our product and vision — I don’t want to diminish the work that Dan and our early team did. But Y Combinator certainly helped us get our product and vision in front of the right people.

The next thing Y Combinator did was help shape the idea.

There are all of these YC tropes and cliches like “Make something people want.” But as you’re actually going through the process of making something people want — and effectively failing at it — you start to get some tangible signals on where you’re going wrong.

That’s helped shape where we are today. The result is a much, much better product than where we were two years ago during Y Combinator.

But the core of what we do, our vision and the value proposition of “The easiest way to have a modern data stack.” Is the exact offering that we had at the very start of Mozart.

It’s just, at that time, that meant Dan and I were effectively being your data engineering team. It was a sprinkle of product with a lot of man-hours to make it work.

Fast forward to today and those ratios have switched. Now it’s the product doing its job, with a little bit of human interaction from time to time.

And there were more benefits aside from that. Dan and I are two technical founders, so YC set us up with help on the legal side, some of the marketing elements, and some PR thrown in.

All of our weaknesses were filled through the knowledge base of YC and some additional support from our partners.

But ultimately, one of the great things for me has been the check-in we do with our group every three months.

They interact with our latest ideas and they know that idea well because they spent three months with us grinding through it in its very infancy.

So they know some of the real weak points of the idea and they want to constantly push on that.

R: What about the worst part? Or rather the things that didn’t work as well for you at Y Combinator.

P: First of all, as I said, we happened to be in this remote batch. The most direct answer is that I think nothing beats in person.

Part of the benefit of doing an accelerator is working with a cohort. I’m a big horse racing fan and when horses work out by themselves, they’re fast. They’re very fast. But when they’re running next to another horse, they’re even faster.

And I had lots of other horses next to me. They just happened to be on Zoom. So one of the things that I think was missed was that real sort of connection where the best parts of your competitiveness that sort of bubble to the top.

The other thing is there are the lucky coincidences that occur when you’re with other people. For example, you’re having dinner and somebody’s complaining about one of their legal bills. They did this unnecessary thing, and then you overhear it all.

You’re not eavesdropping, but you’re overhearing this really important conversation about something that somebody else is doing that would be helpful.

So, for us, we missed out on a few of those things. And obviously, we can’t do anything about COVID, but that was probably the real weakness of the experience for us.

R: I was expecting remote to be the answer there.

Next, for budding entrepreneurs, what would you say are the key benefits of joining an accelerator?

P: Again, I think the networking effects are the strong ones. For us, I think eight of our first ten customers came from, that.

You mentioned the challenge of going from zero to one and then going from one to ten, and then, going from ten to a hundred, etc.

It’s not easy. And we’re at the tail end of that particular challenge, but going from zero to one is probably the hardest bit. And that’s where YC helped us massively. We graduated from YC with three customers.

And I would say that one of the things that’s beneficial is the network of people that are willing to take a little bit of a risk and a chance on your idea.

And you should take that feedback very seriously. When people are willing to try you out, that’s a strong signal.

When they’re willing to pay for something, that’s an even greater signal.

The lessons of what those early customers were telling us what they wanted, and what they would be willing to pay for, were the foundations of where we are today with the customers and revenue we have.

All of this to say, you have to ask for that customer feedback immediately and in the right way.

Don’t just solicit their opinions on things. Really try to find out what people care enough about and why they’re choosing you.

And this is one of the real benefits of doing an accelerator because you have access to this network of early adopters that you wouldn’t necessarily have.

Talking to early adopters is such a critical point in getting from zero to one, and it just made it that much easier.

R: You mentioned that you’re still in contact with your program partners. So what about the extended community, what support do you still get from the Y Combinator community?

P: We still work with our partners at YC.

More than that, we try to support the folks in the existing batch. We also have a couple of customers in the 2023 batch which is nice.

They’ve already reached out for some advice and help, which is great for us. With Mozart Data we’re trying to serve startups so it helps inform our business as we evolve.

And then for us, there’s also that extended expertise from YC that we can call on. There are data folks, PR folks, recruiting, etc.

We’ve been able to make a successful hire through YC. There’s a forum that we try to always actively participate in.

We also did what’s called the “YC Series A Program”. It’s for companies that have raised Series A.

It’s a similar program to the initial YC program, but the focus is on helping you get from Series A to evolving your visitation for raising Series B.

That was also quite helpful.

We think of ourselves as “YC for life” but we’re growing through the process. At first, we were the baby and now we’re in adolescence. The final goal is to be all grown up.

Advice for Entrepreneurs thinking of Applying to a Startup Accelerator

R: Finally, what advice do you have for an entrepreneur or a founding team who’s about to fill out the application?

P: Again, I mentioned that we got waitlisted.

I want to share all of the tips and tricks that I used in that sort of one to two months between when we got put on the waitlist and when we got in.

The number one thing, and it’s a cliche, is to follow the YC tropes.

Start by talking to customers and doing things that don’t scale, which means working for customers, and working for future customers.

So Dan and I worked as, effectively, data engineers for a small set of companies. For free pro bono and effectively built the product out of that free work.

We talked to, I think it was 56 folks in data or data adjacent spaces to talk about their:

Wants/needs
The problems they have
The things they don’t understand
What they do understand

And we dug deep into the things that they care about that are most important to their world.

And then when we came back to YC and interviewed for a second time, we talked so much about our learnings. Not in terms of our product or product vision or even market size.

Instead, we talked about what these customers or future customers were telling us. And I think that’s the number one thing to focus on. Talk.

To get into YC you need to have gained expertise in the space you’re aiming to build in. We had done that through, our decade of working in the space.

But you really also have to talk to prospective customers.

Wrapping Up

Peter’s story underscores the paramount importance of customer engagement.

By initiating conversations, understanding needs, and delivering solutions — even when they don’t immediately scale — he established a strong customer-centric foundation for Mozart Data.

This mindset is a common thread among every successful startup founder I’ve met.

As is another factor that Peter’s story underlined. The value of perseverance.

Building a startup is an exercise in resilience and adaptability, requiring you to persist through challenges and constantly evolve.

Peter and his co-founder weren’t accepted into Y Combinator at first. So they spent two months developing their vision, and it paid off with dividends.

I hope the insights help you with your entrepreneurial journey. Thanks for reading.