Three Key Principles for Building a Data Team

In my work as Talent Lead at FirstMark, I am fortunate to have the opportunity to truly be “in the trenches’’ with our founders as they build their teams. Our founders come to us with strategic questions; often, we have firsthand knowledge of answers to those questions, given our experience working across 100+ startups. And in other cases, we have the ability to tap into our network–which we’ve grown through initiatives like Guilds–to connect our founders directly to leaders who have been there before.

One team-building question that has been asked frequently over the past year is, “how do I get started building a Data team?”

When we see the same question asked by multiple founders, it’s like solving a puzzle. We marshal our network, make one-to-one connections between founders and subject matter experts, and pull together notes around the most common lessons and themes. After spending some time working on “how to build a Data team”, I wanted to share a handful of the most common pieces of advice, as I’m sure our founders aren’t the only ones asking this question…

1. Before you do anything… take stock of your data ecosystem

Before you start building out your data team, take a step back and look at your current ecosystem of data. Clean and well-modeled data is the foundation for all data products that come after it. Depending on the state of your data, you may either need to invest time in getting your house in order before building a data team; or, you may in fact need to make establishing a robust data ecosystem a core part of the first hire’s scope.

On the data science side, you can’t develop effective algorithms on an unclean or unstable data set. On the analytics side, if your data is not modeled well, you will likely be swamped with too many ad hoc requests instead of building a long-term, general structure that serves many purposes.

Your first goal should be to get a data infrastructure in a place that makes data accessible, available, timely, and accurate.

2. Who should be your first data hire?

The most common advice for making a company’s first data hire: find someone who can build and lead a team. While a lot of teams do start with a junior data hire, a more senior hire can bring a broader understanding of the business’s needs while also doing a lot of hands-on work to extract value from your data.

A strong data leader has good business acumen. Understanding how finance, marketing, product might need or use the data in ways that will help the business is a critical capability of an early leader. This helps them make sound decisions aligned to business needs.

In terms of technical competency, it’s important that the first hire also be hands-on. They should be able to build pipelines and reliable data infrastructure, which is the foundation for the successful analytics and data science strategy.

Finally, the third characteristic to hire against is to find a leader who both knows the data space broadly, and has the ability to vet talent and hire a great team.

One other question that comes up is how to sequence specialization as you build a team. If you’re building a small team of one or two, you will of course need to hire generalists who can wear multiple hats. If you are building a larger org with specialization, the first thing to prioritize is data infrastructure and engineering background — since you need to make sure you’re building on a sturdy foundation. Then, you can then hire specialists focused on, for example, Machine Learning or Data Science to build on top of that strong foundation.

For early data hires, it’s very important that you bring in folks with “0 to 1” experience. You need team members who understand the challenges of building data pipelines and infrastructure from scratch. Data leaders at larger companies may have excellent skills, but in the early days, it’s essential to prioritize those who have “0 to 1” experience to set yourself up for success.

3. Should I build a dedicated Data team? Or bring Data into every team across my org?

Another common question when setting up a data team is “should we have embedded data analysts working on individual business teams, or should there be one centralized data team?”

Leaders in our network recommend that a startup begin with a central analytics team, rather than a federated analytics organization. As a company scales, you will inevitably hire dedicated Data team members for each function (Marketing, Finance, etc.), most often reporting directly into those teams. Even at scale, though, it is important to maintain a centralized data team; this team focuses on core infrastructure work, standardizing systems and processes across the org, and more.

The ultimate goal you are building toward is self service. A democratized data culture is more sustainable for the longer term. To set you up for long term success, focus from day one on empowering teams to access their own data. Doing so also has the side benefit of freeing up time for data teams to do strategic, long-term work.

Feel free to reach out with any questions or comments. And finally, I’d like to give a huge shoutout to three incredible leaders in our network for sharing their perspective on these questions: Ihsan Kurt, Adam Denenberg, and Niels Joaquin. Thank you very much!