Turning a Business Request into an Analytical Question

Why does question framing matter?

A request often describes a concern rather than a measurable problem. For example, "Customers are unhappy" could refer to:

• Product quality
• Delivery time
• Price
• Customer support
• A mismatch between expectations and reality

A good analyst clarifies the concern before choosing data or tools. A well-framed question saves hours of unnecessary analysis.

Use the who, what, where, and when prompts

Example: from vague to specific

A good question identifies:

• An outcome (revenue, completion rate)
• A comparison (vs last year, vs another group)
• A time period (April to May)
• A scope (categories, regions, cohort)

Choose a measurable metric

Metrics translate an idea into something observable.

Before comparing results, define:

• Numerator – what is being counted
• Denominator – out of what total
• Time period – which dates
• Exclusions – e.g., test orders, internal accounts, cancelled orders

Check assumptions and limitations

• Does the available data actually measure what you care about?
• Who is missing from the data?

Example: Customer satisfaction scores may not represent every customer, because only some people respond to surveys. Very happy and very unhappy customers often respond more than everyone else.

State such limitations openly instead of presenting the result as universal truth.

Key takeaways

• A clear question guides the choice of data, method, and visualisation.
• If the question is unclear, pause and clarify before beginning the analysis.
• Every metric needs an agreed definition before results are compared.