Skip to content

Your First Query

You've connected a datasource and have some metrics. Here's how to start getting answers.

The easiest way to query is through the Dashboard chat.

Step 1: Open the Dashboard

Click Dashboard in the left sidebar. If you have metrics defined, you'll see smart cards showing key numbers. If not, you'll see an empty state with a chat input.

Step 2: Ask a question

Type a question in the chat bar at the bottom. Examples:

What you wantWhat to type
A single number"What is our total revenue?"
A trend over time"How has revenue changed month over month?"
A breakdown"Show me revenue by country"
A ranked list"What are our top 10 products by revenue?"
Data quality check"Is our data reliable right now?"

Step 3: Read the response

The agent responds with:

  • Narrative (left pane) — written insights with bold key numbers
  • Visualizations (right pane) — metric cards, charts, and tables

The right pane auto-opens when the response includes visual data. You can drag the divider to resize.

Step 4: Follow up

The chat keeps context. After seeing revenue by country, you can ask:

  • "Why is Germany so low?"
  • "What would happen without France?"
  • "How does this compare to last quarter?"

Using the Metrics page

For quick one-off queries without the AI:

  1. Go to Metrics
  2. Find the metric you want
  3. Click Test Query → on its card
  4. See the raw result immediately

Using time periods

You can ask about specific time periods naturally:

You sayThe agent uses
"Revenue last month"Previous calendar month
"Orders this quarter"Quarter-to-date
"Trends over the last 90 days"last_90d lookback
"Compare this year to last year"YTD vs. previous year

OnlyMetrix supports fiscal calendars — if your fiscal year starts in April, "this year" means April 1 onward. See Date Periods for the full list.

Understanding visualizations

The agent picks the chart type automatically based on your question:

Data shapeVisualization
Single number (e.g., total revenue)Metric card with big number and accent bar
Time series (e.g., monthly revenue)Area/line chart
Categorical breakdown (e.g., by country)Horizontal rank bars
Multi-column dataFormatted table

What if something looks wrong?

The agent checks data reliability before every query. If there's an issue, you'll see a warning like:

"Note: revenue data may be incomplete — the invoices table hasn't refreshed in 6 hours (SLA: 2 hours)."

This means the data exists but might not be up-to-date. If the data is critically broken, the query is blocked entirely and the agent explains why.

Next steps

MIT Licensed (SDK) | Proprietary (Server)