OpenGov's primary users are the public servants who run local governments. This includes elected officials like mayors, but the more common day-to-day users are department heads, analysts, accountants, and finance directors. My work primarily focused on accounting and finance users. These individuals are risk-averse and detail-oriented, yet they take pride in their work, viewing it as their solemn mission to present financial data in a way that is understandable to their staff and the public.

A chart of accounts is a long-standing, esoteric accounting tool, essentially a series of taxonomies. An easy analogy for the average person would be the classification of animals: just as scientists categorize animals into groups, accountants do the same for transactions. However, in accounting, multiple taxonomies are used for each transaction, and the structure can vary from one organization to another or even based on the type of report. A small government might use three taxonomies, each with 2-3 levels, while more sophisticated governments could have 4-5 taxonomies with up to six levels. We often encountered charts that tracked tens of thousands of types.

Given the complexity of this tool and the data it managed, I had to grapple with the subject matter. I began diagramming concepts, initially just to help articulate ideas to myself. I shared these diagrams with the Customer Success teams for feedback, prompting great discussions. We also realized these documents could serve as training tools, as I was not the only team member struggling to learn these concepts. Ultimately, the Customer Success team began distributing these documents to customers as well. Over the course of this project, and in my later work, I became one of the company's chart of accounts experts.

Once I felt I was beginning to grasp the concept of a chart of accounts, I looked for ways to better represent it in the interface. Inspired by a popular data visualization library, I mocked up an interface that represented taxonomies as a dendrogram. It was an exciting approach, offering an unprecedented bird's-eye view of the chart of accounts. However, there were serious problems. First, the tree-like organization made it difficult to view the entire categorization for a type—something critically important in a taxonomy. Second, illustrating the connections meant that text was not packed tightly, making it hard to see thousands of nodes in a readable way.

One of my favorite, underappreciated research methods is artifact analysis. By examining the files and objects that people use to perform their work, you can gain valuable, tacit insights. I analyzed the spreadsheet files that users were uploading to OpenGov, trying to understand how information was laid out. I realized that spreadsheets did a great job of displaying taxonomies in a compact, readable, and easy-to-manipulate way. The only drawback was that the row-column grid required labels to be repeated when many categories were grouped under a parent, leading to visual noise. It occurred to me: what if we could create something dynamic?