Intro

Question framing is a key investment at the beginning that makes all subsequent steps more efficient.
Because only when the questions are clear does it become evident:

• which data is relevant at all
• which visualizations make sense
• how possible answers can be communicated effectively

Question framing therefore always consists of two inseparably connected dimensions:

1. Who is asking the question?
   (Role, decision context, data literacy)
2. Which questions must the visualization answer?
   And derived from this: Which decisions should the target audience be able to make?

These two dimensions influence each other. The same raw data leads to completely different questions depending on whether the target audience is a CFO, a regional sales manager, or an operational planner.

Why Question Framing is Crucial

The questions that are formulated directly determine the content of a dashboard. That is precisely why they must be developed deliberately and in a targeted manner.

In practice, however, the opposite often happens. Questions are derived from the availability of data, especially when they are formulated at the level of development teams that naturally tend to think in a data-driven way. As a result, what is shown is what exists but not necessarily what is actually needed. The outcome is dashboards that may appear informative but do not support relevant decisions.

In addition, different stakeholders often have different or even conflicting expectations of a dashboard. If these are not clarified early on, visual compromise solutions emerge that only partially meet all requirements and ultimately do not truly help anyone.

Who Should Be Responsible?

Question framing is not purely a task of data modeling or development.
It is primarily a task for UX designers and user researchers.

They use methods such as:

• End-user interviews to understand real decision-making needs
• Focus groups to align different stakeholder perspectives early

On this basis, the central questions of a dashboard can be defined clearly and in a well-founded way.

What Makes a Good Question?

A good analytical question fulfills all 4 of these criteria:

• Relevant
• Specific
• Contextual
• Action-oriented

Relevant

A relevant question arises from a real need for information not from data availability. It is directly linked to a goal or responsibility of the end user.

Examples:
Not relevant: “What is our total revenue in the current fiscal year?”
Relevant: “Which business units are missing their EBIT margin and require corrective action before the end of the quarter?”

Specific

A specific question leaves no room for interpretation. It clearly defines metric, time period, level of granularity, and comparison reference.

Examples:
Not specific: “How is our sales performing?”
Specific: “Which sales regions are more than 10% below planned revenue in the current quarter?”

Contextual

A contextual question is embedded in a real situation, process, or business objective. It reflects when and why someone needs this information. Without context, the question lacks a connection to the actual usage situation.

Examples:
Not contextual: “What is the customer satisfaction?”
Contextual: “Is customer satisfaction declining after the launch of the new app version, and in which regions is the effect strongest?”

Action-Oriented

An action-oriented question has a direct path to a decision or measure. The answer triggers something.

Examples:
Not action-oriented: “How is our revenue distributed across regions?”
Action-oriented: “Which regions require additional sales support next month in order to still achieve the quarterly target?”

Example of a Well-Formulated Question

“In my role as a sales manager, I need to know which sales regions are more than 10% below plan in terms of revenue in the current quarter, how their variance has developed over the past three months, so that I can decide in the monthly steering meeting which regions will receive additional resources.”

The question is quite long and somehow complex, but that is not a flaw. A precise analytical question is allowed to sound complex. The simplification happens through translation into a visual hierarchy, and only afterwards in the dashboard design.

Visual Translation of the Question

The prefiltered widget shows only regions that are >10% below plan in terms of revenue.

• The left bar chart shows the respective country’s revenue for the selected period.
• The middle variance chart shows the relative deviation from the planned value.
  Since this is the focus, the individual values are ranked (“highest to lowest”).
• The right variance chart shows the revenue trend over the last 3 months in absolute percentage points.

Question template

Based on the above question, here is the underlying question template that can help you formulate the required questions.

"In my role as [role], I need to know for which [dimension] the [measure] is at [comparison] in [time], how [object] has developed over [time/trend], so that I can…. [intention]."