Prepare an answerable analytics request
At fictional Northstar Services, employees use this guide to connect an analysis question to a decision. Start with what someone needs to decide, then make the scope and intended use clear.
Choose a useful starting point
These are ways to shape your question, not separate request destinations.
Support a decision
- Example: decide whether to extend a pilot to another region.
Understand a trend
- Example: compare weekly order volume before and after a launch.
Clarify data or definitions
- Example: ask which events count as a completed onboarding.
Prepare the question
A short brief helps the analyst check scope and feasibility before selecting an approach.
-
1
State the decision and audience
Write the decision the analysis should inform and name the role or group that will use the answer. -
2
Define the population and timeframe
Say which customers, products, locations, or processes are in scope and give start and end dates or a clear comparison period. -
3
Note known data and definitions
List familiar reports, systems, or measures if known. Flag terms that may have more than one definition; do not assume access or availability. -
4
Describe delivery expectations
Explain how the answer will be used, the format that would help, and the date or meeting that makes the timing relevant.
Worked example: from broad request to decision question
Fictional Northstar Services example; business context is illustrative.
Too broad: “Analyze sales.” It does not identify the decision, the product group, the period, or what comparison would be useful.
More answerable: “For the quarterly assortment review, should Northstar keep the spring travel-accessory pilot in the West region? Compare weekly unit sales for the pilot product group across the 12 weeks before launch and the 12 weeks after launch, and show the same periods for the existing travel-accessory range as context. The merchandising lead will use the result to prepare the next review.”
Why this is clearer: It names a decision and its audience, defines the population and time windows, and gives a comparison that can help distinguish a pilot pattern from the wider category. An analyst can then check which measures are available, whether the definitions match, and what caveats should accompany the result. No result or data availability is assumed.
Helpful terms: “Population” means the records or entities included in the question. “Timeframe” means the dates or periods being examined. “Measure” means the value being counted or calculated, such as units sold; agree its definition before comparing results.