We built a tool for professional investors to manage millions of dollars in municipal bonds using predictive factors.
In late 2017, Neighborly wanted to build a B2B product that took advantage of their deep data around municipal bonds and ESG/Impact Investing. At the time, we were seeing many of our investors and other connections transferring their wealth to a new generation, one that was more conscious about their impact on the world. The goal was to create a platform that would become the de-facto analysis platform for RIAs, especially firms looking to differentiate themselves with impactful investing.
To prove our hypothesis that smart data could lead to stronger returns, we would build an internal firm that would manage its own client money. By engaging in dogfooding, we hoped to stand up an MVP while gaining an intimate understanding of the problem before scaling.
In this project, I was instrumental in defining the broader product vision then pairing it down to what our first feature set would become. Through conversations with the executive staff, I was able to develop sketches, initial interface designs, and the team's first design system. As the team grew, I became the director of product design, overseeing efforts.
Over a year, the team grew to approximately ten, including business team members, a product designer and product manager, and a couple engineers.
We started with a two-pronged approach to gain both a qualitative and quantitate understanding of the domain. For a qualitative approach, we mapped landscape by firm size and average client account size. We then identified a distribution of firms and placed them on the map. Our quantitive approach pulled data from available SEC data (from ADV forms).
We were able to find ten interview candidates. I was generally satisfied that while our sample was small, it was at least distributed among small and large firms. We conducted exploratory/generative interviews, asking questions about clients, perceptions on ESG investments, tooling, and pain points.
From transcripts, we highlighting interesting insights and organized them in Milanote. This allowed us to organize trends, note assumptions, and identify questions and opportunities. From these notes, we put together a presentation to the team, sharing what we learned.
We triaged the features that we'd need immediately:
I first defined views, then categorized them, building a more coherent information architecture. Features like a glanceable impact visualization would be challenging, showing three dimensions of impact in results. Other items like identifying what data to filter required more trial and error over time.
Below are a handful of sketches and artifacts produced in the first few weeks as I honed in on a direction.
As we grew, we were able to develop features that allowed for more powerful results. We arrived at a robust set of tooling that allowed us to manage a healthy number of clients with large accounts.
Below is various examples of the features and interfaces that we developed.
After a year, we managed over $150M in bond portfolios with a three-person firm and consistently delivered returns over the market benchmark. From these results, we were able to begin conversations with institutional partners to build white-label functionality and featured for larger teams.
To-date, many of these features are still cutting edge; I'm interested to see how the technology grows.