> ## Documentation Index
> Fetch the complete documentation index at: https://docs.custral.com/llms.txt
> Use this file to discover all available pages before exploring further.

# BI Best Practices

> Get the most out of Custral reporting with these core principles.

Good reporting is less about building charts and more about answering questions with data you trust. In Custral, every report is a **block on a dashboard**. You pick a source object, a **group-by** dimension, and a **measure**, and the chart reads your live records. These practices help you build reports people actually act on.

## 1. Start with the question

Before you add a chart, write the business question in one sentence.

* **Vague**: "I want a chart of all sales."
* **Sharp**: "Which pipeline stage has the highest drop-off for Enterprise deals?"

A sharp question tells you the source object, the group-by, and the measure before you touch the block picker.

## 2. Match the chart to the question

Every chart is configured the same way (source → group → measure), so pick the shape that fits what you're asking. See [Reporting](/bi/reporting/overview) for the full list.

| The question is about…             | Reach for                                                              |
| ---------------------------------- | ---------------------------------------------------------------------- |
| A value compared across categories | [Bar](/bi/reporting/charts/bar)                                        |
| A trend over time                  | [Line](/bi/reporting/charts/line) or [Area](/bi/reporting/charts/area) |
| Parts of a whole                   | [Pie](/bi/reporting/charts/pie) or [Donut](/bi/reporting/charts/donut) |
| Drop-off across pipeline stages    | [Funnel](/bi/reporting/charts/funnel)                                  |
| One headline number                | [Stat card](/bi/reporting/overview)                                    |

## 3. Use drill-down for the "why"

A number is a starting point, not an answer. Click any segment (a slice, bar, point, or cell) to open the underlying records in a table. When a stage looks stuck or a count spikes, drill in and read the actual records instead of guessing.

## 4. Filter with intent

Reports have two levels of filtering. Use both deliberately:

* **Per-block filters** scope a single chart (e.g. "Enterprise only").
* **Dashboard-level filters** connect across blocks so the whole board moves together. Change the date range or owner once and every report updates.

Connect the filters that belong to a dashboard's story; leave one-off scoping on the block.

## 5. Keep dashboards focused

Resist putting every metric on one board. Build focused dashboards per team, per pipeline, or per objective.

* **Sales leader**: pipeline health, win rate, stage velocity.
* **Support lead**: ticket volume, response time, resolution rate.

Custral can also **draft a starter dashboard from a prompt** when you're beginning. A fast way to get a first layout you can then refine.

## 6. Keep the data current

Reports read your **live** records. There's no separate warehouse, so accuracy is entirely a function of the data underneath.

* Make sure teams consistently fill in the fields your reports group and measure by.
* Use required properties, [property agents](/ai/property-agents/overview) (AI auto-fill), and [automations](/automation/overview) to keep records complete and up to date.
* A funnel or a count is only as honest as the Status and date fields feeding it.

## 7. Iterate and retire

Your reporting needs change as you grow. Review dashboards regularly, retire reports nobody opens, and reshape the ones that no longer match how the team works.

## What lives elsewhere

Reporting covers charts and dashboards over your records and conversations. Conversation analytics, sentiment, and call coaching are separate surfaces. See [Conversations](/comms/conversations/overview) and Coaching, not BI.

## Related

* [Reporting overview](/bi/reporting/overview): chart types and how to build one.
* [Build your first report](/bi/getting-started)
* [Team performance](/bi/reporting/team-performance): the per-teammate aggregate table.
* [Troubleshooting](/bi/reporting/troubleshooting)


## Related topics

- [Best Practices](/ai/best-practices/overview.md)
- [Conversation Widget](/blocks/widgets/conversation.md)
- [Property Lists](/blocks/embedded/object/property-list.md)
- [Getting Started](/data/getting-started.md)