Product School

Season 7 - Episode 8

Getting Value Out of Data by Mixpanel CEO

Amir Movafaghi, CEO of Mixpanel, talks about getting value out of data once you collect it, how product analytics is becoming accessible for everyone, and making data-based decisions.

Before running Mixpanel, you have a really interesting background, especially for a tech person, because you come from finance, right?

Yes, a long, long, long time training in the finance world. In 2010 I moved to Twitter and effectively joined Twitter at a time where the company was small enough that your title didn’t really mean anything. As long as you were there, you were solving problems and you’d solve whatever came your way. 

And that was really the first time things really opened up and I was doing like 20% finance, 80%, everything else. And that kind of morphed into me getting my hands on different functions and different problem-sets across the organization. And that kind of paved the way to where I am now.

Can you tell us about the story of Mixpanel and how you came onboard?

Mixpanel is a little over 10 years old and I actually knew Mixpanel from my Twitter days. We had to do a lot of what Mixpanel does now in-house, because at the time, pretty much the different problems of tracking the data, storing the data, analyzing the data were three fundamental problems without a defined solution in place back in 2010. So everyone in the market, from Twitter and LinkedIn, Facebook, all the companies that were scaling at that time, were building all these things in-house. It was very expensive, incredibly arduous. And you were spending a lot of amazingly valuable developer time having to build things that weren’t a competitive advantage.

I really believe then ultimately the decades that are the years that ensued, you had an incredible amount of innovation and work that went into maturing, the tracking, a lot of great options for where to store your data, but this problem of self-serve analytics, the problem of now that you’ve gone through this incredible amount of data engineering lift, you’ve cleaned up, you’ve stored your data.

How do you actually get value out of it? Right? How do you access it in a way where you can compose your questions and be able to learn from your data in a way that was meant to be, and that’s been our problem at Mixpanel to solve and something that we’ve really been chipping away at. And we’ve come to a place now where I’m really proud of the percentage of questions that we can now truly put at the hands of the end-users. And it’s something that is going to be a key focus for us for years to come, and an incredibly powerful place for us to focus on.

Back in the day, analytics was for the data people or for the engineers. It was a very technical product. And now non-engineers can also use these products and focus on getting those insights.

Even for the engineers, right? Just because you know how to write SQL, or even if you’re not an engineer and, you know, SQL, it doesn’t mean you want to spend your day taking the most difficult path to getting the answers. That should be a lot easier to get. And so, regardless of whether you are, or aren’t technical, our role as a company is to just improve productivity and make it easier for anyone to be able to interact with this data and the most efficient and enjoyable path.

It should be far easier for you to get the answers you need to do your job in an effective way, and collaborate, share and make decisions on that data. Validate assumptions, come up with new hypotheses. 

Before big data became a thing right after apps and mobile and cloud computing took off, everyone was an analyst on spreadsheets, on Excel, you could pretty much do anything. And there was this sort of like relationship that when big data came into play and you have essentially a vast majority of the most complex set of data is now this app engagement data, all of the user interaction data that’s sitting on the app. I think the relationship with data changed, people moved into this sort of okay, I need any day, another person that’s responsible for understanding this data, making sense of that data. 

And then you get into this like Q&A format, right? And I, whether you have an embedded analyst or a central data team, it makes it very hard for data to become an integral part of your day to day workflow. When you are essentially dependent on this, I have to submit something, wait for an answer. Maybe it will take a week, maybe a few weeks, or even a couple of days, but you’re out of that flow. Whereas your mind is just trying to answer these sets of questions that are in sequence, right. There’s not, it’s not just one question. It’s like 20 questions that come out of that first one. So our work is essentially, how do we make that possible for you to be able to go get that type of relationship with your data that allows you to actually fold that into your day to day execution.

How are you thinking about the future of the industry? How do we make sure that teams feel comfortable working with these tools?

I think for the most part, one of the greatest things that’s happened in our ecosystem is that so many great entrepreneurs have been innovating in all different parts of the ecosystem. A lot of the problems are the new kind of set of data solutions that people are putting together, have honestly been a paradigm shift from back when you essentially had a limited options of what, what was on the table in terms of things that you could go stitch together from what was available to you in the marketplace. And you had to build a bunch of stuff in house.

Today, there’s incredible options for the way in which you can set up your tracking, right? There’s amazing seed companies that are focused on the CDP space with segments and particles. There’s up and coming, amazing companies on the infrastructure side, like Rutter Stacks, and catering more to data engineering and kind of focusing more offerings around the cloud data warehouse. And then honestly, the way Snowflake and Bigquery have evolved over the last few years. The cloud data warehouse is so much more powerful than it was just a few years ago. And you have these amazingly exciting tools, that reverse CTL tools that are making it much more seamless for pulling the data out of the data warehouse. Then there’s obviously been an insane amount of innovation just on data transformation governance within the data warehouse.

I think that this is what essentially allows everyone to really focus on what they’re good at and not have to create these incredibly complex, bundled solutions or like building horizontally. Because the temptation is Mixpanel’s focused on self-serve analytics, and if we’re dependent on A,B,C and D the fastest way is to go build it ourselves, right.

But it takes a long time, it’s distracting and it’s not our core value. What we’ve really enjoyed has been the amount of innovation and the way that our ecosystem has evolved. Over the last few years as a company, we’ve just been laser focused just on self-serve analytics, because now it’s so much easier to have well-governed quality data coming into Mixpanel with the work that teams have already done. That is in terms of cleaning up, putting data into the warehouse. And now just looking for self-serve UI to be able to draw those insights. So I think that, in my view, this best of breed ecosystem that’s going to build around the cloud data warehouse is just beginning. We have an incredible decade of innovation that’s going to come in with all of the things that are now possible to do with a clean, well-governed and easy to extract data from Bigquery and Snowflake. 

The best strategy you can pick is what is the ideal configuration that you would recommend to a customer. And right now, I think what you can do with this modern data stack and these players that are just some of them, I just mentioned is incredible. For new startups, you look at who’s the fastest, biggest adoption of these tools. Because they don’t have any baggage, a history of accumulated debt. To them it’s a no brainer. I’m going to go implement AB and C and I’m done.

But for everyone else, it takes a little bit of kind of like, okay, we gotta go through some change management and we got to rethink the architecture and it’s a heavier lift. So naturally, there is a market for the bundled approach. And even though it’s not ideal, there’ll always be someone that just wants to say, how do I just go from a bundle to a bundle solution? 

Tell us more about the status of the company when you joined the company today and how you’ve changed as a leader?

For the most part, as a CEO of the company, you’re setting a vision for the company in terms of where it’s headed. And then your day to day is how right. What do you prioritize? How do you sequence the different problems that you want to take on now versus, you know, what you want to solve later? 

The second part of a house and the culture of the company, right? How do people work together? What are the things you value? And I think the combination of those two things helps you essentially keep marching toward, um, the ultimate vision of the company, and surrounding all of that is your, like your purpose, you know, like, why are we doing this?Why are we headed in that direction? And why are we, you know, prioritizing these things? 

I think being very deliberate and communicating the ‘why’ is really the job. I guess the biggest change for me has been just the amount of appreciation for transparency, and effective communication. You’re constantly making decisions and you’re going through changes, and there’s this incredible amount of value in just doing transparently and communicating effectively through all that. That’s been probably the biggest area for how I’ve been able to observe, take feedback and just evolve as a leader.  

How do you structure your calendar? 

A good portion of my time is just meeting with my team. The ability to connect and make sure that as an executive team, we’re on the same page, we know what problems we’re trying to solve, what we’re trying to prioritize, that is a huge part of how I spend my time.

The second is just having enough space to be able to actually objectively look at inputs, that are coming in either through, observing data, understanding data points, both quantitative, through Mixpanel, but also qualitative, through customer calls, NPS qualitative feedback. I try to really study as much objective data as possible. So the inputs I have into my own decision making is very much qualitative, as well as the quantitative bits of what’s happening in the business. I look for validation, right? We have a bunch of hypotheses that we’re taking into consideration as we evolve the product as we evolve to go to market execution. 

The thing that I want to create essentially, is this loop on, being able to understand whether we’re doing the right things and be able to get validation on that as soon as possible, so that we can course correct or continue to do more of what we’re doing.

We intentionally want to feel the joys and the frustrations that our customers face. We want to create this authentic inside, out brand that essentially most companies want to have an extremely positive view of themselves to the rest of the world. For us authenticity matters more, and so being able to actually really reflect who we are I think shines through and the way that we’re able to enable these things for our customers, because we’re living them right. And we’re seeing those challenges and we want to expose and open those up. Some of the things that we’re actually starting to think about is how do we actually take a lot of these learnings and open it up as content to our audience. That will be another kind of sprinkle of just openness and sharing our own challenges of overcoming these problems over time.

I’m also very proud of our latest collaboration, the product analytics certification. 

Our intent is to put the power of self-serve analytics for everyone and we want to earn the ability to have many of them over time become customers and help us grow. In the end, the substance driven part of this is putting the power of the knowledge in the hands of users, what we’re capable of, what things we can solve together. And the collaboration has been amazing because it’s not just the, what the technology opens up, but the way to think about where do you start asking these questions? We’ve really enjoyed working with you and your team. Carlos, it’s been amazing.

As a lifelong learner, what are you excited about learning these days and how do you have time for it?

When I’m not busy at Mixpanel, I’ve got three kids. So I’ve gotten quite passionate about education for them and just kind of spent a bit more time getting myself familiar with our public school district. One of the things that I’ve really believed in is both kind of like a sense of responsibility to get more involved in our community and just participating proactively. And then also just like I said, a huge area of passion for me in terms of what we could do to participate in, and help more kids be independent thinkers, critical thinkers. I view that as such an incredibly powerful impact on society. 

What do you think people could do to get started in product?

I think in the grand scheme of things, it’s always helpful to start with the why. What are you trying to do and why? And what is the ultimate effect you’re trying to have? The starting point is then you need to kind of form some idea, either validated through your observation or your intuition that you think is going to create a better experience. Then the way to leverage data is either to craft those hypotheses or validate them. 

Right. But if data is, is in neither part, then you have no objectivity of whether you’re making a difference, that you’re actually improving the user experience. You may own a big part of the journey, but whatever it is just be very intentional about what is the goal or outcome that you’re trying to drive toward.

And then B, use data as a way to objectively measure whether you really are making a difference. The more simple, the better right people get stuck in creating these complex models trying to be very, very sophisticated. And sometimes the actual truth, the core foundational piece gets lost. So the more you can simplify, the easier it is to explain to someone what you’re trying to do and why, and how you’re measuring that’s always a good measure, right? If you can’t explain it, or if the people don’t understand it, then go back to the drawing board and give it one more shot. 

What does the future of Mixpanel look like?

One of the biggest things that we’ve focused on as a company has been how we make the most complex dataset incredibly easy for anyone to navigate and learn from it. And, weave it into their day-to-day workflow. The thing that we’re now really focused on is how do we continue on that journey and make it from where it is now from 10 X or a hundred X better, right. On the part we’re really taking advantage of the incredible innovation around us. We’re going to be making it much easier for you to essentially augment your product engagement data with your business data.

So you’re trying to understand your business’s impact on monetization, the relationship between different parts of the data that’s getting captured in a business application and stored in your data warehouse. How do you much more seamlessly layer that into Mixpanel? So you can unlock a whole new wave of questions. You can ask with ease with data that’s already been governed, cleaned up, and now you can essentially join and, and be able to really unlock the value of the product, not just for engagement data, but also the business data that you’re capturing other SaaS applications. We’re super excited that we’ve already started on that journey, but you’ll see a lot of exciting stuff coming out in the next few months that will further deliver on that vision.

Listen to our episodes on your favorite platform

Stay tuned for new episodes

By sharing your email, you agree to our Privacy Policy and Terms of Service

Person talking on a microphone