Number inspector
Open the inspector of a tile or a point on a chart to see the summary, facts, breakdown, why it changed versus the previous period and the SQL that ran.
Every number on a dashboard can be opened to see where it comes from. The explanation is computed by the metric engine from the definition, not written by AI.
Open the inspector
- Click Details on a tile, or click a bar or point on a chart.
- Click Why? on a number tile to jump straight to the Why it changed tab.
Tabs
| Tab | Shows |
|---|---|
| Summary | The value, period, filters applied and the comparison with the previous period |
| Facts | The key facts of the selected point |
| Breakdown | The number split by one dimension (channel, product…) |
| Why it changed | How much each part contributed to the change |
| SQL | The query the metric engine ran (you can Copy it) |
Why it changed
Choose Vs. previous period (why it rose or fell) or Why it's at this level. The tab explains it in several ways:
- By formula: the metric split into its components (for example profit after fees = net revenue − cost of goods − fees) and which component drove the change;
- By formula tree: the same idea, several levels deep;
- Pulling down most / Pushing up most: the values of a dimension (channel, product) that contributed most;
- Notable two-dimension combinations: for example one product on one channel;
- a table of This period, Comparison period and Change.
Follow up, pin
- Ask Si about this: opens the Ask Si panel with the selected point as context.
- Pin to page: see Dashboards and pinning to a page.
- Clear selection closes it.
Si uses the same explanation when you ask "why", whether in the app, Slack, Telegram or over MCP.
Related
Ask Si in the app
Ask Si in plain language on the Ask Si screen or in the dashboard's Ask Si panel; read the source and confirmation level, edit how the question was understood, use deep thinking.
Si writes and runs SQL/Python
For exploratory questions that go beyond the ready-made metrics (medians, percentiles, correlations, joining two tables, custom charts), Si writes SQL or Python and runs it in an isolated sandbox. Results are always Unconfirmed.