Designing Question-First Dashboards

A technically correct dashboard — clean charts, working slicers, no gridlines — can still fail if the person opening it does not immediately know what to look at, what the numbers mean, or what decisions the dashboard is meant to support.

Data-first vs. Question-first

• Data-first: start with the data you have. Build charts from available tables. Add slicers for fields that can be filtered. The result shows everything possible — but not necessarily what is needed;
• Question-first: start with what the stakeholder needs to know and what decisions the report must support. Work backwards to select only the metrics, breakdowns, and filters that answer those questions. Everything on the dashboard has a reason to be there.

Template:

Task

Step 1 — Define your questions using the template

On the My_Questions sheet, create a table with four columns: Question, Metric, Breakdown, and Filter. Fill in at least three rows: one for each chart you plan to build.

The rules are:

1. Each Question must be written as a full sentence, not a chart description.
2. 'Sales by Region' is not a question. 'Which region is generating the most revenue this quarter?' is.
3. Each Metric must be a DAX measure already in the model: Total Sales, Transaction Count, Avg Order Value, or Distinct Customers.
4. Each Breakdown must come from a dimension table: Region or Segment from Customers, Category from Products, or Month/Quarter from Dates.
5. Each Filter should only list dimensions the person reading the report will genuinely need to change.

You may use some of the same questions as the Chapter 5.2 dashboard if you choose, but at least one question must be different from the three already covered.