Charts lets you build custom analytics visualizations in Ender Turing. Each chart compares one or more datasets — defined by conversation filters — grouped by a dimension such as agent, team, topic, or time period. You can track conversation counts, unique customers, duration, and cost. Charts support multiple chart types, formula-based calculations, benchmark lines, shadow period comparisons, and data export.

Open Charts from the main navigation. The list view shows all charts you have access to, with their name, assigned team, labels, and chart type icon.

Charts can group data by:

When grouping by Agent, Team, Tag, Customer, Topic, CRM Status, or Queue Name, a Top N items control appears. It limits how many entities appear in the chart and table, using the chart's current sort order. Both 1D charts and charts with a time interval (2D) default to 20 when no limit or individual entity selection has been saved. Existing explicitly selected limits and sorting are preserved. While editing, the untouched default is shown in muted text with a Default: 20 tooltip; an explicitly entered limit uses normal text.

Enter a positive whole-number limit, or clear the field or enter 0 to display All. All removes the Top N limit for any of these groupings and remains selected when switching between 1D and 2D. Choosing All does not clear your entity selection or conversation filters, and backend safety limits still apply. If you select individual entities without setting a limit, the chart shows the full selection and the control also displays All.

Select one of four metrics for the chart:

Each dataset can override the chart-wide metric independently.

Each dataset in a bar-type chart can override its visualization type independently.

A Quadrant chart plots one point per entity (agent, team, topic, queue, …) using the first dataset as the X axis and the second dataset as the Y axis. Two dashed lines drawn at the median of X and the median of Y split the plot into four zones, making it easy to spot entities that score high or low on both axes. The median is used instead of the arithmetic mean so a few high-volume outliers do not pull the dividers and misclassify the rest of the entities.

When Quadrant is available. The chart type is selectable only when the chart has exactly 2 datasets. With fewer or more datasets, the Quadrant icon in the type selector is disabled and a tooltip explains why ("Add a dataset — Quadrant needs exactly 2." or "Remove extra rows — Quadrant uses exactly 2 datasets."). While Quadrant is selected, Add new row is capped at 2 datasets, and the Save button is blocked with a tooltip if you somehow end up with the wrong dataset count.

Group by. Quadrant compares entities, so time-based groupings (Day, Week, Month, Year, Hour) are hidden from the Group by dropdown while Quadrant is selected — they would otherwise collapse every entity onto a single bucket. If you switch to Quadrant from a time-based grouping, the chart automatically switches to Agent. All non-time groupings — Agent, Team, Tag, Topic, Customer, CRM Status, Queue Name, FCR, and CSI — remain selectable. The per-grouping Top N control is hidden in Quadrant mode. Agent and Team groupings have no backend cap, so every matching entity is plotted. For Topic, CRM Status, Queue Name, and Customer, the backend caps each dataset at 200 entities ranked by that dataset's own metric; the scatter plots the union of those per-dataset top-200 lists, so on tenants with more than 200 entities the rendered point set is bounded by — but not strictly equal to — the primary dataset's top 200.

Best zone. While editing a Quadrant chart, use the Best zone dropdown above the chart to pick which corner represents the "good" outcome — Top-right (high / high), Top-left (low / high), Bottom-right (high / low), or Bottom-left (low / low). The chosen zone is tinted green and the diagonally opposite zone is tinted red; the other two stay transparent. Selecting Neutral (no good/bad) tints the four zones with an alternating soft/strong neutral pattern (diagonally paired) so adjacent quadrants stay visually distinct.

Reading the chart. Hover any point to see a tooltip with the entity name and both axis values. By default points are unlabeled — toggle Show values on chart to display the entity name as a label chip next to every point. Click a point to drill through to the Conversations view filtered for that entity, using the row whose dataset filters define the point. If that source row is a formula dataset, the click opens the formula explanation dialog (showing the formula breakdown for that entity) instead of redirecting to Conversations.

Visible range. While editing a Quadrant chart, use the Visible range control above the chart to narrow the scatter to a numeric window. It opens a panel with four optional fields — X minimum, X maximum, Y minimum, and Y maximum. Leave a field empty to leave that side open; entered bounds are inclusive (a point sitting exactly on a bound stays visible). Only points that fall inside every set bound remain on the chart, and both median dividers are recalculated from the visible points alone — so the four zones re-split around whatever subset you are looking at, and the chart axes snap to the bounds you entered. A Showing N of M agents count (or "…items" for non-agent groupings) appears under the chart so you can see how many entities the range is hiding. Click Reset to clear all four fields and return to the full scatter. The control's summary field reads All when no range is set, or a compact summary such as X 0–100 · Y ≥5 when it is.

If you enter a minimum that is greater than its maximum, the panel shows Minimum cannot be greater than maximum. and the Save button is blocked until you correct it. A valid range is stored with the chart and restored the next time you open it.

Viewing a saved range. When you are not editing the chart — including view-only access — the editable control is replaced by a read-only Visible range limited badge whenever the chart has a saved range. Viewers can see that the scatter is filtered and what the bounds are, but cannot change them.

Features hidden in Quadrant mode. Quadrant does not use shadow periods, benchmark lines, per-dataset chart-type overrides, per-dataset visibility toggles, or the column selector. Switching to Quadrant clears any previously set shadow, hidden datasets, per-dataset chart-type overrides, benchmark lines, column-to-show selection, and Top N limit so the scatter renders both axes without chart-side culling (subject to the backend 200-entity cap noted above).

Each dataset represents a filtered subset of conversations. Configure datasets by:

In Expert mode, you can add formula-based datasets that calculate values from other datasets. Formulas support basic arithmetic (+, -, *, /) and reference other datasets by their identifier. Use formula datasets to compute percentages, ratios, or differences between datasets.

Compare current data against a previous period by enabling a shadow:

Shadow data appears as a secondary series on the chart for visual comparison.

When Group by is set to Year, only Year-over-year shadow comparison is available. Month-over-month and Previous period do not appear in the shadow selector for Year charts because calendar-year buckets make those comparisons ambiguous. If you switch a chart with shadow enabled to Year grouping, the shadow type is automatically adjusted to Year-over-year. Clicking a shadow value on a Year chart drills into the compared calendar year — for example, the shadow value shown under 2024 opens conversations from January 1 through December 31 of 2023.

Add one or more benchmark lines to the chart to indicate target values or KPIs. Benchmark lines follow the chart's value axis: they appear horizontal on bar, stacked bar, and line charts, and vertical on horizontal bar charts. Each line has a configurable value and color. Benchmark lines are not available for pie charts.

Toggle Show values on chart to display numeric values directly on chart elements. This setting can be overridden per dataset.

Click Select columns to show to choose which groups appear on the chart. You can sort the chart alphabetically by label or by a specific dataset's values in ascending or descending order.

Toggle Expert mode to access advanced features:

Three export options are available:

In both 1D and 2D, XLSX ignores the chart's Top N limit, entity picker, custom sorting, and collapsed/revealed state. Collapsed data remains included; expanding the table does not change the export. Conversation filters, dates, Only Main topic filtering, access restrictions, metrics, and supported formulas still apply. Existing report safety limits remain in force, so a large 2D export can exceed the cell limit even when the displayed Top N subset fits.

When you open a chart, a fast filter row appears at the top of the chart view. These filters scope all datasets on the chart and are independent from the per-dataset filters configured inside each dataset.

The row always includes:

Up to three optional filters can be added through the tune icon: Type, Direction, and Topics.

Date and other filters behave differently when you switch between charts:

From the list view, each chart supports these actions:

Click any chart element (bar, line point, or pie segment) to navigate to the Conversations view with the corresponding filters pre-applied, letting you drill down into the underlying conversations.

Users with view-only access can browse all charts, interact with visualizations, export data, and click through to conversations, but cannot create, edit, or delete charts.