Dashboards | ATX AnalyticsTable Handbook

DashboardsKPIs, slicers, tables, pivots and every chart type — forecasts and histograms included — on a free canvas, bound directly to the source. Dashboards are separate tabs — a free canvas of placed widgets, saved with the project. The design decision that matters is that data widgets bind to a connection, schema and table and query it directly, so a dashboard is independent of whatever happens to be pinned in a sheet. That is what lets one project carry an executive view, an operational view and a reconciliation view without any of them breaking when someone reorganises a worksheet. The Add Widget menu covers five families: text (Title), numbers (KPI, Table, Pivot), slicers (Filter, Range filter, Search box), charts (Chart, Heatmap, Waterfall, Histogram, Forecast) and plans (Timeline). Each chart answers a different question — how much (column, bar, pie), how it moved (line, area, combo), where it is concentrated (heatmap), how one figure became another (waterfall), what the values look like (histogram) and where they are heading (forecast) — and all of them share one renderer and one settings panel with the charts on the Reports tab.

What it does

Fifteen widget types on an absolutely-placed canvas: Title, KPI, Table, Pivot, Filter, Range filter, Search box, Chart, Heatmap, Waterfall, Histogram, Forecast, Clusters, Key drivers and Timeline
Comparison charts — column, bar, line, area, pie and combo (columns and lines together, each series its own type), with stacking, palettes, axis and legend options and reference lines
Heatmaps — a rows × columns grid coloured by value, with date grouping, colour scales and midpoint, conditional colour rules, row and column totals and outlier flags
Waterfall bridges — from an opening figure to a closing one, or actual against budget in variance mode, with subtotals, a closing total and inverted sentiment for cost bridges
Histograms — the distribution of one numeric column, with automatic or custom bins, normal and density curves, percentile markers and specification limits with Cp and Cpk; clicking a bar filters the board to that range (chapter 20)
Forecasts — a value over time projected ahead, with a backtested choice among eleven methods, 80 % and 95 % prediction intervals, scenarios, a management overlay and one forecast per region or product (chapter 19)
Clusters — k-means segments found in several numeric columns at once, with k chosen automatically, a profile of each segment and its quality; clicking a segment filters the board to its members (chapter 26)
Key drivers — which columns explain a target such as revenue or churn, by their share of R², in which direction, and the segments that stand out; clicking a segment filters the board to it (chapter 26)
Data story — Dashboard → Share → Data story writes a short narrative of the board in which every sentence cites its sources and every figure is checked; copy it as Markdown, download it, or add it to a PDF report (chapter 27)
Widgets that read governed metrics instead of a table, so a KPI, chart or pivot shows the certified definition of the number (chapter 27)
Project timelines — milestones, phases and a today marker on a date axis
KPI aggregation over SUM, AVG, COUNT, MIN and MAX with prefix, suffix, decimals, thousands separators and k/mln/bln scaling
Slicers that other widgets follow: a dropdown Filter (single or multi-select, with a configurable cascade scope), a Range filter with from / to sliders over a number column, and a Search box that matches a phrase across chosen columns
Table widgets with chosen columns, conditions, number formats, colour rules and an optional refresh every few seconds
Pivot widgets with row and column dimensions, subtotals, grand totals, optional period comparison, distribution bars, colour scales and highlight rules
Every chart exports as PNG, SVG or CSV — a forecast exports every period and series, a histogram one row per bin
Dashboard definitions export and import as JSON — widgets, layout, bindings and formatting, never row data

How it works

Create a dashboard tab, add widgets, and bind each data widget to its connection, schema and table.
Slicers publish a selection that the widgets in their scope follow, and they apply before anything is aggregated: a histogram counts, and a forecast learns from, only the rows the board is filtered to.
Clicking a chart cross-filters the other widgets on the board; a histogram filters to its bar's range of values and never filters itself.
A dashboard chart and a sheet chart are the same object — same configuration, renderer and settings — so any chart can be switched to another type, Heatmap, Waterfall, Histogram and Forecast included, and a type added to one surface appears on the other.
Because bindings are direct, a dashboard refreshes on its own terms and is unaffected by sheet edits.
Exporting a dashboard as JSON moves the design between projects without carrying a single row of data with it.

Practices that keep it reliable

Build one dashboard per audience rather than one dashboard with every widget on it
Choose the chart by the question: column or bar for how much, line or forecast for where it is heading, histogram for what the values look like, heatmap for where they concentrate, waterfall for how one figure became another
Place a Range filter beside a histogram so viewers can zoom into the band of values that matters
Scale large KPI figures (k / mln / bln) — a board-level number nobody can read at a glance is a number that gets re-typed into a slide
Move a proven layout between environments as JSON instead of rebuilding it

In a regulated environment

A dashboard JSON carries design only, which makes it safe to move from production into a test tenant or a partner's project.
Direct bindings mean dashboard numbers and sheet numbers come from the same source rather than from two copies that drift.
Forecast and histogram settings — method, horizon, bins, limits — are saved chart definitions in the project file, so a changed assumption is a reviewable diff in Git.

Screens

An executive dashboard — KPI tiles, cascading filters, charts and a pivot on one canvas.
Dashboard — widget palette, dashboard switching, and JSON import/export.
Dashboard → Add Widget — Chart, Heatmap, Waterfall, Histogram and Forecast side by side.
A dashboard mixing chart families — slicers on top, a revenue forecast with 80 % and 95 % bands, histograms at the foot.
Pivot configuration — dimensions, subtotals, grand totals and period comparison.

Related guides

ATX Institute on LinkedIn

This site needs JavaScript for the full experience. Every page above is reachable without it.