As treatment launched, we were simultaneously expanding the testing engine to new conditions: COVID + Flu, Flu only, Women's UTI, Drug testing, Lyme disease. Data from customer service told us something was wrong with how users were reading their results. Working with our data team through Grafana session analysis, we identified four root causes: language barriers with overseas proctors (40.7%), test kit interpretation confusion (31.2%), deliberate result manipulation (18.8%), and self-consciousness around sensitive test types (9.3%).