The world’s leading publication for data science, AI, and ML professionals.

What I Learned in My First 6 Months on the Data Team at HealthJoy

Notes on transitioning from big to small data teams, working on a data team at a startup, and more

Photo by Myriam Jessier on Unsplash
Photo by Myriam Jessier on Unsplash

As the title suggests, this post is going to be another entry in the "what it’s like to work at a startup" genre of post that I’m sure most people have read at this point. I am, however, going to put my own unique spin on things and write from the perspective of someone who experienced not just the transition from Big Conglomerate™ to a mid-stage startup, but specifically what it’s been like to work on a small but growing Data team at said startup. Hopefully, this post can provide others in the data and analytics space some useful perspective on how to make a similar transition.

A little bit about me – I’m Joey, I’m a data professional living in Brooklyn, NY. I have a Master’s in Data Science from DePaul University, and I’ve worked in the data and analytics space for about 5 years after beginning my career in the operations space. I’ve worked as both an analytics manager and an individual contributor, and my current role is as a Senior Data Analyst at HealthJoy.

I’m going to split this post into a few sections, the first being around my transition into the new role at HealthJoy and my first couple of months on the job. The second section focuses on scaling up into my role, and the last hailing the virtue of our team and how much I’ve enjoyed learning about all of our tooling and our data stack.


The Transition

In my first two months or so at HealthJoy, the sheer amount of newness and adjustments I needed to make felt, at times, overwhelming. Moving from a 15 person, decentralized data team to a four person, centralized data team was a big shift in and of itself. Couple that with new tools and tech to learn (Gitlab, Snowflake, dbt, Looker), learning an entire new industry, lingo, etc, and it was easy to feel lost.

Pretty quickly though, and with the help of my manager and some more tenured folks on the team, I found ways to contribute. In a way, my lack of prior knowledge was an asset to me in those first couple months and helped our team structure documentation for our current growth period. I asked a lot of questions about our data and datasets (pipelining/architecture, column meanings, metric definitions, etc.) and quickly found out that a lot of that knowledge lived in other current and former team members’ heads. This helped jump-start our team’s efforts for putting together documentation standards for new data models and led to a lot of education and discussion about some of our important but less-than-perfectly documented data models and sources.

My big takeaway from all of this was that if I was confused about something, chances were that that confusion represented an opportunity to learn, document and contribute something for our team that hadn’t previously been formally documented. Some other areas of contribution were taking concepts from tool-specific training (Looker and dbt, specifically) and creating takeaway documents or recommendations for how to implement these new concepts. From what I can tell, getting new sets of eyes on well-established processes is one of the quickest ways to generate improvements. The team recently brought on two more analysts, and within their first 2–3 weeks, they had put together some great materials and documentation on Python & development environment setup.

Along with the learning curve that came with adapting to a new team came the realization that working at a smaller company meant that my role as an analyst would allow much more visibility than I’d previously experienced at any past role. Having joined in Q4 of 2020, I was quickly asked to help a more senior member of the data team with analyzing some of our customer retention data. After finishing up my part of the project and passing the work off to our stakeholders, I was pretty shocked and excited to see our work displayed and praised during just my second monthly all-company meeting. While I’d presented work to high-leverage stakeholders before, it was a real "ohh sh*t" moment to realize that a lot of the work I’d be tasked with doing was going to have visible, real company-wide implications.


Ramping Up

While I learned plenty in my first couple of months on the job and did my best to make myself useful, it took time to feel truly comfortable and like I knew enough about our data and data stack to really be dangerous. At the same time, some of the concepts and philosophy that my manager/our team’s director put into place around how he wants to run our team and how we should work as a unit really came into focus. Specifically, concepts like operating our data team like a product team and focusing on cross-functional analytics really came into light for me in my first full quarter with the team.

While analytics teams that I’ve been on previously did have OKRs and team-wide goals, they were also set up as a client-focused service organization. As analysts, we were focused on providing outputs to different internal and external stakeholders based on what our clients’ needs were at any given moment. This made it difficult to progress on any longer-term data team priorities, pay down any technical debt, or invest in data infrastructure – nearly all of my analyst time was dedicated to responding to our clients’ needs.

As I mentioned, at HealthJoy we’ve strived to operate our data team more like a product team, rather than a service desk. This process entailed us working collaboratively on our team’s quarterly priorities to give myself and my data teammates a clear roadmap and help us prioritize our work. This shift has allowed me to dive deep into my priorities and respond to stakeholder feedback not just from an output perspective, but from a collaborative perspective. Having the agency to sink my teeth into a couple of projects was freeing after years of working in a client-facing analytics role. The fruits of this philosophy and labor came about fairly quickly – putting together a recurring revenue model for our entire organization being the result of one such impactful project. I’ve also been able to dive deep into how our clients and end users activate across the HealthJoy product, leading to some key findings that challenged assumptions as to how our clients activate with our platform product and a lot of education for cross-functional stakeholders.

While all data and analytics teams strive to help our organizations and stakeholders make better decisions, it’s been an incredible learning experience so far to see how a top-down philosophy and vision is enacted. A little deliberation, planning and thinking can go a long way in the data space.

All that being said, I was lucky in that my first projects were super interesting, and I’ve received a lot of positive feedback and recognition from stakeholders. Data models I put together for recurring revenue are now being used by our CEO and other leaders. The company has a better idea of why our recurring revenue is changing and which clients are contributing the most to changes in recurring revenue for HealthJoy. What was most interesting is that these projects were not fancy machine learning models or intricate, multivariate tests (not that we’re not capable of those things) – they were just useful data analytics for stakeholders that haven’t seen data analyzed in this way previously.


Technical Skills & Wrapping Up

One of my favorite parts of working in data and Data Science is just the sheer amount of new things to try and learn. That was another reason that made me excited to join the data team at HealthJoy – the promise of the "modern data stack" and improving my coding skills while solving problems with data was something that I was really looking for in my next role. Compared to my previous analytics role, our data stack makes me feel like a kid in a candy store. Learning to use dbt and contributing to our team’s dbt project is one of my favorite parts of the job. Solving a business problem while writing SQL and Jinja really is cool to me, and I’m pretty sure the "a-ha moment" that comes with solving a tricky coding problem is something that I’m going to constantly be chasing, since it makes me feel like a genius (at least, temporarily).

I also really enjoy our team’s weekly show-and-tell meetings, where each of us spend about 5 minutes chatting about something we’ve been working on over the past week or so. It’s really cool to see how talented my team members are and learn about the awesome stuff they’re working on each week. Whether that be a new data model created in dbt, updates to some of our machine learning and predictive projects, or just a new tool that someone has found and wants to share with the team, the culture of constantly sharing and learning is really awesome. I frequently find myself inspired to research new concepts or areas of the data science/data engineering ecosystem after our team’s meetings.

Wrapping up, I’m really happy I joined the data team at HealthJoy. The experience has been extremely rewarding and eye-opening, and I’ve gotten exposure to tools and a working style that I’m not sure I would’ve experienced if I had stayed in my previous industry or continued working at bigger, more established organizations. I’m really excited for what I’m going to learn in the coming months, the direction our team and the company is headed.


Related Articles