AI Dashboard Prompts for Metrics and Tables
A dashboard is a decision surface, not a collection of decorative charts. Tell the builder who uses it, which decision they need to make and which data supports that decision.
UI Discovery editorial · Updated
How to use this guide
Start with a question and a time range
An operations lead may need to know which tasks are late; an analyst may need to explain a change in conversion. These require different dashboards. Define the default period, comparison period and metric units so cards and charts cannot quietly tell different stories.
Design the table as a working tool
Specify column meanings, sorting behavior and the action available on each row. Keep filters visible when there are no matches. A narrow screen may require a horizontally scrollable table, but the page heading and controls should remain within the viewport.
Make uncertainty visible
Distinguish zero from unavailable, pending and not applicable. Show when data was last refreshed and what happens after a failed refresh. Do not leave a skeleton on screen indefinitely or substitute invented values for a request failure.
Original prompt to adapt
An editorial starting brief, not a benchmarked model output. Replace fictional content and review the result before publishing.
Build an operations dashboard for a fictional studio managing client deliverables. Use the existing app shell. Show overdue tasks, due-this-week tasks and completed tasks for a clearly labelled date range. Add an accessible status filter and a table with project, owner, due date and status. Use labelled fixture data with consistent totals. Include initial loading, empty workspace, no filter matches, unavailable metrics and failed refresh states. Keep filters and existing results visible during refresh. Use text labels in addition to color for status. On mobile contain horizontal overflow inside the table, not the whole page. Include a last-updated label and a retry button that cannot issue duplicate simultaneous requests.
A focused follow-up
Replace one metric with unavailable data and set the filter to no matches. Verify that neither state is displayed as zero and that clearing the filter restores the previous results.
Review checklist
- Cards, charts and tables use the same date and filter scope.
- Unavailable values do not masquerade as zero.
- Keyboard controls, narrow layouts and retry behavior work.
Continue with real references
- Analytics dashboards →
Compare metric and time-range hierarchy.
- SaaS dashboards →
Compare workspace navigation and operational tables.
Common questions
Can I use this prompt unchanged?
Use it as a starting brief. Replace fictional content, inspect the reference links and adapt constraints to your project. These are original editorial prompts, not benchmarked model outputs.
What should I check before publishing?
Cards, charts and tables use the same date and filter scope. Unavailable values do not masquerade as zero. Keyboard controls, narrow layouts and retry behavior work.