From business decision to research question
6 min read
Lesson objectives
- Distinguish a business decision from a research question, construct, and indicator, then write a testable decision brief.
Please sign in to use all the tools
Once you sign in, the reading tools are yours: larger or smaller text, typeface, line spacing, focus mode, highlighting and bold, and notes saved to your account and gathered in My study.
A problem worth researching
A manager of a fictional delivery service says, 'Repeat orders fell; send a satisfaction survey.' That jumps from a symptom to a tool before identifying the decision. The apparent fall could come from a new active-customer definition, the end of a discount campaign, late deliveries, or an influx of first-time customers who rarely return. Start with the decision management faces: change delivery fees, improve arrival-time reliability, or hold action while checking the evidence. Research reduces uncertainty around that choice. It neither replaces management judgment nor turns an association into a cause. State the alternatives, timing, and decision owner, then ask which missing information could change the comparison.
Separate four layers. The business problem describes a consequence that matters: a lower share of returning customers. A research question asks about a pattern that data could address: does repeat ordering differ by observed delivery delay when customer and time-window definitions are held fixed? A construct is an idea that cannot be observed directly, such as confidence in the promised arrival time. An indicator is an operational way to observe part of it, such as the share of orders arriving within the promised window or a carefully worded response scale. A satisfaction item is not automatically a trust measure: respondents might be judging food quality or price. Every indicator needs a unit, denominator, period, and source. For example, count customers who place a second order within 30 days and divide by all customers whose first order was completed in a specified month. Decide whether cancelled orders are excluded before comparing periods.
Test whether the question can guide work. 'Why do customers dislike us?' presumes dislike and leaves both the customer group and the feeling undefined. A stronger version asks: among people completing a first order in April, how does the chance of a second order within 30 days differ between those whose first delivery met the announced window and those whose delivery missed it? This is a descriptive comparison. Even if a difference appears, neighborhood distance, product type, and order size may contribute. If the decision requires evidence that changing the delivery promise itself causes repeat ordering to change, a stronger causal design is needed, such as a suitably assigned test with execution checks. Match the strength of the claim to the available design.
A one-page decision brief stops a study from becoming a list of questions with no use. Record the choice and alternatives, the population affected, the decision horizon, the primary measure and its definition, what is known and missing, the evidence that would change the choice, and execution limits. A fictional internal rule might be: if missed arrival promises remain associated with lower repeat ordering after basic differences are examined, test a more accurate message on a limited scale. If the difference disappears after accounting for neighborhood and order type, do not present delay as an established cause. That threshold is a rule within this case, not a universal market standard. Stating it before seeing results also reduces the temptation to select a convenient explanation afterward.
Lesson summary
Distinguish a business decision from a research question, construct, and indicator, then write a testable decision brief.