Writing a Research Question

A great research question is the single most important ingredient in any study — it determines what you'll learn, who you'll talk to, and whether your findings will actually be useful. Before you open Koji, before you choose a methodology, before you write a single interview question, you need to get this right.

Why Your Research Question Matters

Think of your research question as a compass. Every decision in your study — from the methodology you pick to the probes your interviewer uses — flows from this one statement. A vague or poorly framed question leads to vague, unusable findings. A sharp, well-scoped question leads to insights you can act on immediately.

When you create a new study, the very first thing Koji asks is what you want to learn. Your answer to that prompt shapes everything the AI Consultant suggests, so it pays to spend a few minutes thinking it through before you type.

What Makes a Good Research Question

A strong research question has five qualities:

1. It's Specific Enough to Be Answerable

Your question should point to a clear area of inquiry. You don't need to know the answer yet — that's the whole point of doing research — but you should be able to imagine what a useful answer looks like.

Too broad: "What do customers think about our product?" Better: "What frustrations do first-time users encounter during their first week with our onboarding flow?"

The first question could lead anywhere. The second tells you exactly who to talk to, what to ask about, and what kind of findings to expect.

2. It's Open-Ended

Research questions should start with "how," "why," "what," or "in what ways" — never with "do," "is," or "will." Yes/no framing leads to yes/no answers, which rarely reveal the deeper stories you need.

Closed: "Do users like the new dashboard?" Open: "How do users incorporate the new dashboard into their daily workflow?"

3. It Focuses on Behavior, Not Opinion

People are unreliable narrators of their own preferences, but they're excellent at describing what they actually do. Frame your question around actions, decisions, and experiences rather than hypothetical preferences.

Opinion-based: "Would customers pay more for a premium tier?" Behavior-based: "How do power users currently work around the limitations of the free plan?"

4. It's Scoped to a Timeframe or Context

Adding a specific context or timeframe makes your question more researchable and your findings more precise.

Unscoped: "Why do people churn?" Scoped: "What factors lead new subscribers to cancel within the first 30 days?"

5. It Doesn't Contain the Answer

Watch out for questions that smuggle in assumptions. If your question presupposes a specific conclusion, you'll unconsciously design your study to confirm it.

Leading: "Why is our complicated pricing page causing users to leave?" Neutral: "How do prospective customers evaluate and compare our pricing before making a purchase decision?"

Examples by Use Case

Here are research questions across common scenarios to spark your thinking:

From Research Question to Study Design

Once you have your research question, you're ready to start building your study in Koji. Here's how the pieces connect:

1. Enter your question when you create a new study. Be as specific as you were when you refined it above.
2. Chat with the AI Consultant — it will use your question to suggest a methodology, identify your target audience, and draft interview questions. The more precise your research question, the better these suggestions will be.
3. Review the research brief that gets generated. Your research question becomes the problem statement at the top of the brief, and every other section should trace back to it.

If at any point the brief doesn't feel right, it's often worth revisiting your research question. Sometimes the act of designing a study reveals that you were actually asking two questions at once, or that the real question is slightly different from where you started.

Common Pitfalls to Avoid

Trying to answer everything at once. If your question has the word "and" in it, you might actually have two studies. It's better to run two focused studies than one sprawling one.

Writing for stakeholders instead of for learning. Your research question isn't a slide title. It doesn't need to sound impressive — it needs to guide your inquiry. Keep it honest and practical.

Skipping this step entirely. It's tempting to jump straight into writing interview questions. Resist that urge. Five minutes spent on a clear research question saves hours of wading through unfocused data later.

Being too attached to your first draft. Your research question will often evolve as you work with the AI Consultant. That's not a sign of failure — it's a sign that your thinking is sharpening.

Quick Self-Check

Before moving on, run your research question through this checklist:

• Can I explain what a useful answer would look like?
• Is it open-ended (starts with how, why, or what)?
• Does it focus on behavior or experience, not just opinion?
• Is it scoped to a specific audience, context, or timeframe?
• Does it avoid baking in assumptions or conclusions?

If you can check all five boxes, you're in excellent shape. Head over to Creating Your First Study to put that question to work.