Creating Custom Dashboards

Dashboards let you combine multiple charts and notes into a single view you can revisit and share with your team. Use them for weekly reporting, experiment reviews, and tracking operational health.

Create a dashboard

1. Open Analytics -> Dashboards.
2. Create a new dashboard.
3. Give it a clear name and description.

Only the dashboard owner can edit a dashboard.

Dashboards can contain multiple tabs. In edit mode, add, rename, or remove tabs to keep related views together. Each chart belongs to one tab, and its layout is saved independently from the other tabs.

For a built-in comparison of conversation volume, AI resolution, escalations, satisfaction, and resolution time across customer segments, open Analytics > Customer Segments.

Add charts

1. Click Edit.
2. Click Add chart and pick a chart type.

Common chart types:

• KPI: a single number (optionally broken down by a group)
• Line / Bar / Area: trends over time
• Stacked bars + line: positive and negative flows with a stock or rate line
• Scatter: compare two values across groups
• Heatmap: patterns across two dimensions (weekday × hour by default)
• Cohort heatmap: retention by start month and months since activation, with an optional customer-segment selector
• Pie: share by category
• Funnel: step-by-step conversion (no group breakdowns)
• Table: compare This period vs Previous period
• Note: add context, definitions, and action items

Configure chart data

Open a chart menu (…) and choose Edit manually (or Edit with AI).

Choose a data source

• Conversations: built-in outcomes like resolved, escalated, and satisfaction.
• Events: tracked actions like sessions or product/customer actions.

Tracking event metrics

Event charts rely on a metric name and optional properties. If you do not see an event metric you need, record it from a workflow step where it makes sense (for example, when a workflow starts or after a message is sent).

• Use a stable metric name (for example, product.view).
• Add properties you will want to group or filter on later (for example, channel=chat, tactic=recommendation).

If you choose Events, set Base object:

• Conversation to apply conversation filters and break down events by conversation attributes.
• None to count events on their own.

Build a metric

Each chart is powered by one or more metric rows. Only the last row is shown on the chart; earlier rows can be reused by name.

• Name: a short key (for example, resolution_rate) used for the chart and to reuse earlier rows.
• Metric: what you count.
• Baseline (optional): what you divide by to show a percentage (Count ÷ Baseline).
• Distinct on … (optional): count unique items instead of total events.
• Aggregate across groups (optional): ignore the breakdown and use overall totals.
• Property filters (Events only): limit events to key=value pairs, separated by commas.

Tips:

• Use names and metric names with letters, numbers, underscores, dashes, and dots.
• Avoid spaces and quotes in names.

Groups (breakdowns)

Add one or more group options in the Groups section. In the chart menu, use Group by to switch the breakdown. Funnels do not support group breakdowns.

Heatmaps use groups as axes: with no groups the grid is weekday × hour, one group shows weekday × group, and two groups show the first group as rows and the second as columns. Adding a Baseline to the metric shades cells by percentage rate instead of count. Use Sort rows / Sort columns in the chart menu’s Chart options to reorder grouped axes by value or alphabetically; weekday and hour axes always stay in time order.

Any chart can also group by the time-of-day dimensions Weekday and Hour of Day, derived from the chart’s date column in your workspace timezone.

Cohort heatmaps expose their available customer segments directly in the chart.

Filters

Add chart-specific filters in the Filters section. These stack with the global filters in Analytics (including the date range).

Arrange, resize, and export

In edit mode:

• Drag and drop charts to reorder.
• Resize a chart by dragging its edges or corners.

From the chart menu, you can also open a chart in full screen. Table charts can be exported as CSV.

Compare experiments (optional)

From the dashboard actions menu:

• Choose an experiment to compare variants on your dashboard.
• The date range aligns to the experiment window.

Clone and visibility

From the dashboard actions menu:

• Export to download a dashboard file for backup or reuse.
• Clone to create a copy you can modify.
• Toggle Public/Private visibility (owner only).

Next steps

• Understanding Metrics