Widget types
Line chart
Show metric trends over time. Best for comparing multiple series such as error rate by project.
Area chart
Like a line chart but with filled areas. Good for visualizing volume over time.
Bar chart
Discrete bars per time bucket. Useful for comparing event counts in fixed windows.
Table
Tabular results with sortable columns. Best for top-N lists such as slowest transactions.
Big number
A single large metric value with optional comparison to a previous period. Good for KPIs on a status board.
Top N
A horizontal bar chart showing the top values for a dimension. Useful for “top error types” or “top affected users”.
Dashboard queries
Every widget is powered by the Discover query engine. When building a widget you specify:- Dataset — errors, transactions, spans, or session health
- Visualization — the chart type
- Y-axis — the metric to plot (for example,
count(),p95(transaction.duration),count_unique(user)) - Filter — a search query to narrow the data (for example,
transaction:/api/checkout environment:production) - Group by — an optional field to split the metric into series (for example,
project,browser) - Sort and limit (for tables) — column to sort by and maximum number of rows
Creating a dashboard
Add a widget
Click Add Widget, choose a visualization type, configure the query, and save the widget.
Dashboard templates
Sentry ships several prebuilt Insights dashboards covering common monitoring scenarios:- Frontend — Web Vitals, page load times, JS errors
- Backend — API latency, database query performance, error rates
- Mobile — app start times, screen loads, crash rates
- Caches — cache hit/miss rates and latency
- Queues — job processing time, failure rates, queue depth
- LLM Monitoring — token usage, model latency, error rates
Sharing and permissions
- Organization dashboards are visible to all members of your organization.
- My dashboards are private to your account by default.
- Any member with at least project-member permissions can create dashboards.
- Dashboards can be shared via a direct URL.