Bars, horizontal, stacked
Chart
One bar per category running left to right, its series stacked end to end so the bar is their total — names down the left, the total at the bar’s end, longest first.
When to use
Section titled “When to use”Many named things, each split into parts of one whole — answers per brand split into recommended and not, tickets per agent by channel. Each part’s number is printed inside it where it fits. Show it as share and order it by one series to rank by that part’s share — the brands most often recommended first.
When not to use
Section titled “When not to use”The series overlap or nest (seen > answered): the stack would total to a number that means nothing — use horizontal bars side by side.
What the table needs
Section titled “What the table needs”A column whose values are the categories (a brand, a person), and one series per part, each narrowed by its own filter.
Asking the AI
Section titled “Asking the AI”Describe it in a sentence — for example “answers per brand, recommended vs not”.
From another AI
Section titled “From another AI”An assistant connected to Routario over MCP (Claude, ChatGPT) makes it with chart_create, then puts it on a board with dashboard_tile_add.
Settings
Section titled “Settings”- Show as — On a stacked chart: the amounts themselves, or each part’s share of its own bar, so every bar reaches 100%.
- Subtitle — One line under the title. What the reader should take from it.
- Along the bottom — The column whose values become the slots along the bottom — a date for a trend, a category for a comparison.
- Grouped by — Only applies when the column along the bottom is a date: which stretch of time each slot covers.
- Series — Each series is one measure, optionally narrowed to the rows that match a condition.
- Measure — What each series computes over the rows in a slot.
- Unit — For ‘time between two dates’: the unit the gap is measured in.
- Aggregate — For ‘time between two dates’: how several rows’ gaps become one number.
- Where — Narrows one series to the rows matching a condition — how several series share one axis.
- Colour — A plain colour picks a pigment and says nothing about the data. A named meaning — went well, needs attention, went wrong — tells the reader something they will believe.
- Show numbers as — How the value axis, the tooltips and the printed values spell a number.
- Show dates as — Only applies to a date axis; anything else is left exactly as stored.
- Targets — A flat line at a goal, a floor or an SLA the chart is judged against. Not something the data produced.
- Good side of the target — Which side of the target is good — it colours how far a number is from it.
- Legend — Where the series names sit, if anywhere.
- Value axis — Whether the numbers up the side are drawn. On unless switched off.
- Gridlines — Whether the faint lines across the plot at each value step are drawn. On unless switched off.
Auto-generated from the tile registry (core/custom_tables/tile_registry.py) — the same text the chart editor and the AI drafter read. Do not edit by hand.