Explore data with analysis

Use analysis when you want to investigate a question, explore a change, or work out which metrics belong in a report.

Open AdHoc, then use the + button (New conversation) to start. Type your question in Ask a question about your data… and select the arrow button (Send message).

Start a conversation in AdHoc and enter your question.

Give the question enough context

Include the data source, business metric, period, timezone, and exclusions that matter. Define ambiguous terms such as “active customer” or “net revenue.” Mention a specific connection when several sources contain similar data.

Compare completed orders this month with the same period last month. Use the store’s reporting timezone and separate new and returning customers.

Choose the analysis depth

Below the message box, select Auto (or the current effort level) to open Analysis depth. The options are Auto, Light, Medium, High, and Extra High. Auto lets FastBI choose. A direct lookup needs less investigation than explaining a change across several dimensions; more extensive analysis can take longer and use more analysis capacity.

Choose how deeply FastBI should analyze your question.

Iterate in the conversation

Ask follow-up questions to change the grouping, inspect an outlier, or test a possible explanation. Answer clarification requests instead of letting an ambiguous definition carry forward.

Verify before sharing

Try a reproducible example

Download the fictional orders CSV, open the paperclip (Upload files), and choose Add file on the Private tab. Wait for ready, then mention the file in your question:

Using only this fictional CSV, plot weekly net revenue for the four Monday-starting weeks from October 5 through November 1, 2026, including zero for weeks with no orders. Query exactly two columns: week_start and net_revenue_usd, with exactly four rows. Do not include an overall-total row or column in the query result. Show a bar chart.

The six input rows total $1,400: $300 for October 5, $450 for October 12, and $650 for October 19, 2026. The week starting October 26 has $0, because the sample contains no orders for that week.

Expand Thought for …, then Query Upload Sql, to inspect the query result. Select Bar chart if the table view is active. Use Table view or Open table viewer to check the values, and Show SQL to inspect the query. The chart below uses the four weekly results; verify your answer against the file.

Weekly net revenue from the sample orders CSV, including the week with no orders.

Check your own results

• Check that the source and date range match the question.
• Inspect SQL and filters when provided, especially joins and currency conversions.
• Distinguish observed differences from possible explanations.
• Check whether the underlying data has finished syncing.

For direct SQL exploration, use the SQL workbench. For recurring views, use Reports.