If it's not observable, is it really accessible?

Accessibility testing is often split between automated tools (like Lighthouse and axe-core) and manual checks (such as keyboard navigation and focus behavior). The results usually live in static reports, making it hard to understand trends, spot regressions, or measure long-term impact.

Out of curiosity, and after a lot of experimentation, UX designer Victoria Nduka started building a system that treats accessibility as something you can observe over time. It runs automated accessibility audits, captures interaction-related signals that reflect real user behavior (keyboard navigation, focus patterns, assistive tech usage), sends these data to Prometheus and Loki, and visualizes everything in Grafana.

In this talk, Victoria shares lessons learned so far, including:

• Capturing real user interaction signals and automated audit data
• Structuring the results as time-series logs for Grafana
• Building dashboards that show trends, regressions, and improvements

Instead of treating accessibility as a one-off compliance task, this approach treats it like performance or reliability: something you continuously observe. The talk shows how Grafana can be used beyond traditional metrics and logs, and how open source observability tooling can support inclusive design by making accessibility measurable, visible, and actionable.