How to Write a Research Question for Your Own Organisation

Short answer

A useful research question names a situation people have lived through, the people who lived it, and the decision the answer should inform. "Why is morale low?" invites opinion. "Why do orders packed after 3 pm miss the 4 pm truck at the Lyon warehouse, and what do pickers do when they do?" invites examples. Write the question in one sentence, the way you would ask a colleague. Then check it against a decision, a population and a period before anyone is invited.

Start from the decision the answer will inform

Most internal questions begin as a concern: turnover is up, a project is late, a tool is unused. A concern is a reason to ask, but it is not yet a question. Before writing anything, name the decision someone could take differently depending on the answer, and the person who can take it.

If no decision can be named, the work may still be worth doing. Call it exploration, and expect a description rather than a recommendation.

Ask about what people have done, not what they think in general

The GOV.UK Service Manual guide to in-depth interviews advises interviewers to focus on stories and real examples, and to avoid generalities and talk about how things "should" happen. That discipline starts with the question itself. A question about behaviour and events gives each person something concrete to describe. A question about attitudes invites a verdict they may never have formed until someone asked.

Compare:

• “What do people think of head office?” asks for an opinion.
• “What do store managers need from head office to run their stores, what do they receive today, and what do they ignore?” asks about work they do every week.

The second version also holds three topics a conversation can cover in turn: what managers need, what arrives, and what gets ignored.

A one-sentence test

Write the question in one sentence and read it aloud. Then check it against five fields.

The last row matters most. A question that names its own cause, such as "too slow", tells every participant what the organiser already believes, and answers tend to follow it.

Turn the question into topics

One sentence is a starting point, not the conversation. Each interview needs three to five topics that together answer the question. For example: how the process normally runs, the last time it went wrong, what people did next, and what they would change. Remove topics that are interesting but do not serve the decision. A question about missed delivery slots does not need opinions on the new uniforms.

Test the topics on yourself or a colleague before inviting anyone. Rewrite or remove any topic that cannot be answered from experience.

A fictional example: from concern to question

A fictional hotel group notices that fewer than half of last summer's seasonal staff have applied to return. The first draft of its question is "Why are seasonal workers disloyal?" It assumes a cause, and it would put people on the defensive.

The second draft is "Why don't our seasonal staff come back?" It is neutral but broad. The third draft adds the situation, the population and the decision: "Why didn't last season's housekeeping and kitchen staff at our three coastal hotels come back this year, what were their working weeks really like, and what would make them return?"

From that sentence, the operations director agrees four topics:

• how a typical week was scheduled;
• the last time a shift changed at short notice;
• accommodation and travel;
• what would have to be different next season.

The question now points to things the group can change: rotas, notice periods, housing, and the offer it makes in the autumn.

Common mistakes

1. Two questions in one. "Why do people leave and how can we improve pay?" assumes pay is the answer. Ask the first part, and let pay come up if it is relevant.
2. A question only managers can answer. If the people you invite have never seen the process, the answers will be second-hand.
3. A question about a person. "Why is Sam's team underperforming?" turns research into an investigation. Ask about the work, the conditions and the process.
4. A period nobody remembers. Accounts of events from a year ago are reconstructions. Anchor the question to recent, specific events where possible.
5. No room for "I don't know". A good question lets people say the process is not theirs, or that they have not seen the problem.

For questions about why people leave, the exit interview questions guide and the stay interview questions show how the same principles apply to retention conversations.

Where Lontra fits

Lontra starts from the sentence you write. It proposes the topics each conversation should cover. You change the wording, remove what does not matter and add what is missing, and Lontra stays within the topics you approve. You can try the conversation yourself first; your practice answers never count. Each person you invite then has a short private conversation, and Lontra follows up on what they say. How it works shows the four steps.

A good question still has to come from you. Lontra cannot tell you which decision matters most. A question that assumes its answer will produce findings that seem to confirm it.

Write it down before you invite anyone

Keep the question, the decision, the population, the period and the topics on one page. When the findings arrive, compare them against that page. It records what you set out to learn, rather than what you hoped to find.