This submission is intended for the panel "Public Policy Analysis Mishaps".
In a multi-year project, we contracted an external web design company to help us design a small interactive web application for outreach. The goal was to make the results of our project on automated analysis of policy discourse around urban sustainability accessible in a playful way.
The company was expensive but came highly recommended. We spent basically all of our outreach budget on them. The company did good work - the application is well documented, beautiful and runs with little maintenance. We also followed some good practices from UX design, including creating user stories and iterative feedback rounds with policy actors.
Still, based on a soul-crushing analysis of server logs, basically no one ever looks at the resulting interactive dashboard. It's a beautiful space on the internet mostly visited by crawlers and bots.
In this contribution to the public policy analysis mishaps panel, I will explore where I think we went wrong with our approach in hindsight. I will also show how we built a second, minimalist application ourselves for a tiny fraction of the cost of the first application, which is actually visited by humans sometimes. In essence, it turns out, the true benefit of wasting our money on an expensive application no human ever visits was the API we built to power it (and how you can and should do so straight away).