Design a real-time analytics dashboard
mediumComplete guide with examples and interview questions · 45 min
Say it out loud in an interview — clear, simple, no textbook wording.
If someone asks me to design a real-time dashboard, I would say it is a page of independent widgets that share one time range and update on their own schedules. The page is not one fetch with many charts. It is many small data lifecycles that happen to sit side by side.
Interviewers care because dashboards expose rendering discipline. A page that re-renders every widget on every tick, or that tries to plot ten thousand points as ten thousand DOM nodes, fails visibly. The question tests whether you isolate updates the way you isolate failures.
The senior nuance is coordination without coupling. Widgets share time range state but nothing else. That split is what lets one widget fail, refresh, or lag without dragging the rest of the page with it.
Key takeaways
- •A dashboard is a set of widgets with independent data lifecycles, not one page fetch
- •Streaming updates must not re-render the whole dashboard on every tick
- •Charts need downsampling: thousands of points cannot all be pixels
- •Time ranges, refresh rates, and granularity are one coordinated state
- •A failing widget must fail alone, never take the dashboard down with it
Track your mastery
Sign in to save Q&A scores, coding challenge progress, points, and streaks across devices.
Sign in to save progress