Self-updating dashboards: how to build a panel nobody has to feed

The dashboard seen most often in operations is a file. Someone exports the data in the morning, updates the spreadsheet, generates the charts and sends it around. By the afternoon, the numbers no longer match what is happening. A self-updating panel swaps that ritual for two decisions: where the data comes from and how often it is read again.

Why dashboards go stale

A panel goes stale because of the path that leads to it. Each export is a copy of the data at one moment. If the portfolio import finished after the export, the panel does not know.

The more manual steps between the data and the chart, the more the panel depends on someone remembering to run each one.

Read straight from the data

The first step is to connect the chart to the place where the data arrives. In DeepMiner, widgets read straight from the indexes. Imports from files or through the API land in those indexes, and the widgets count, sum and group the fields of each schema.

That way, a portfolio Pareto reflects the latest import.

Refresh at the pace of the decision

Not every number needs to change every instant. Set the interval by the pace of the decision the widget supports:

• Day-to-day operations, such as records imported today or the total customers in a segment: a short interval.
• Weekly tracking, such as the daily evolution of imports: a longer interval.
• Background analysis, such as the portfolio distribution by segment: no automatic refresh, reloaded when someone opens the panel.

In DeepMiner, the Refresh every (s) field sits on each widget. With it filled in, the widget fetches the data again without reloading the page, and the panel can stay open on an operations screen.

One panel, many cuts

Building one dashboard per region or segment multiplies the maintenance work. Global filters solve that: the same panel, with the filter applied to every widget at once.

In DeepMiner, the Filters panel works on fields from one or more indexes and shows the count for each value. Pick two segments and the KPI, the Pareto and the donut change together. A set of filters can be saved as the default and is already applied on the next visit.

Start with AI, adjust by hand

Building the first panel from scratch is the slowest part. In Create with AI, you describe what the dashboard needs to show: the goal, the period, the cuts and the comparisons. The assistant only uses fields that exist in the chosen indexes and builds the widgets. Ready-made templates help you start, and the request can be dictated with the microphone.

Then the editor lets you move, resize and configure each widget: chart type, metric, grouping, period and, on a KPI, icon, color, position and size. The dashboard assistant also takes change requests, such as adding a ranking, and proposes the change for you to accept or undo.

Charts born in a conversation with the Data Assistant also reach the panel through the Add to dashboard button.

A checklist for the operations panel

1. Each widget answers a question someone asks every week.
2. The data comes straight from the indexes, with no export in between.
3. The refresh interval follows the pace of the decision.
4. The most-used cuts are in the global filters, and the default is saved.
5. The main number sits in a KPI with an icon, readable from a distance.

Want to see an operations panel updating itself? Book a demo.