Products

Builds the page around the question rather than the chart: funnels, retention grids and cohort comparisons are first-class objects, not panel types.

The densest display in the collection and the one with the most trained audience. Worth studying precisely because none of its choices would survive a usability test.

Progress and attainment read by a teacher about named students, where a red cell is about a person in the room rather than a server.

A public dashboard with no account, no filters worth the name, and an audience of journalists. Designed for people who will read one number and leave.

Consumer framing over a market display: the same order book and depth chart the professional products carry, deliberately buried a level down.

Everything on one surface, which makes it the best place to see what a page looks like when the product refuses to choose between metrics, logs, traces and cost.

The reference for the emergency department track board: every bed present whether occupied or not, read by a dozen roles who each need a different column.

Freight in transit as a map and an exception list rather than a chart, because the question is which shipment is late, not what the average is.

The contribution graph is the calendar heatmap every other product copied, and its five-step bucketing is the deliberate trade that made it legible.

The reference implementation for panel grids, template variables, and stat panels; most other tools are defined by how they differ from it.

A card grid people genuinely rearrange, on a wall tablet, which is where resizable layouts either work or quietly stop describing the house.

Query-first; heatmaps and BubbleUp replace the dashboard-of-panels model with draw-a-region cross-filtering.

Pipeline by stage with a forecast on the same page, which is a dashboard carrying a projection and an actual in the same visual language.

A public SCADA demo running live simulated plant data, with a runtime toggle between simple and photorealistic pump and pipe rendering—the ISA-101 argument as a setting.

Query-first rather than panel-first: the search bar is the primary control and the charts are downstream of it, which inverts Grafana's arrangement.

The full order book and depth ladder are public with no account, which makes it the only display here updating several times a second that anyone can go and watch.

A trace waterfall for model calls, where the interesting span is usually the one waiting and the cost sits on the same axis as the latency.

Almost no chrome and no configuration. A board where the layout is the data model, and the closest thing here to a dashboard with an editorial voice.

Board of saved reports rather than a designed page, which is what product analytics tends to become once more than one team is asking questions.

Per-second charts, hundreds per node, with a per-chart anomaly ribbon instead of a band on the series.

A queue rather than a chart. Two-pane list and detail with keyboard traversal, because the task is working through incidents rather than looking at them.

One column, top to bottom, where the metric row doubles as the chart's control. The clearest working argument that a dashboard can have exactly one interaction.

A builder aimed at people making internal tools, so it shows what a dashboard looks like when the person assembling it is the person who will use it.

Issue-centric rather than metric-centric; the two-pane list-and-detail structure with sparklines inline on every row.

Retail's weekly rhythm drawn against a comparison period by default, which is the only honest way to show a business with a Monday and a Black Friday.

The reference for a financial admin surface: dense tables with in-cell state, and a home page that is a small strip of numbers above a very long list.

Thousands of dashboards made by people who are not designers, published without a review step. The best available sample of what the pattern language looks like in the wild.

Deploy state, build logs and edge metrics in one console, with a log tail that has to stay readable while it moves.

Occupancy and leasing pipeline over a horizon measured in years, which is the far end of the cadence range this collection covers.

Headcount and attrition about identifiable people, which is the case where suppressing small numbers is a design requirement rather than a nicety.

Analytics for an audience of one, about their own work, where the comparison that matters is against their own previous upload rather than against a benchmark.

Queue depth and wait time for the people clearing the queue, with the same numbers again for the manager, on the same data and a different page.