Hey Juno

How to write a research brief for customer research

By Hey Juno · Published · Updated

A good research brief starts from the decision you need to make, not from a list of questions. It says what you must decide, who can inform it, what you already believe, what limits you work within, and what a useful answer looks like. The questions come later, and often from whoever runs the research.

Start from the decision, not the questions

The most common weak brief is a list of questions. "Ask people what they think of the new pricing page" sounds like a brief, but it hides the real problem: what will you do differently depending on the answer?

Write one sentence in this form: "We need to decide whether to _, and the answer will change _." For example (illustrative): "We need to decide whether to keep a free tier, and the answer will change our launch plan for next quarter."

If you cannot finish that sentence, you may not need research yet. You may need a conversation inside the team first. A decision gives the research a purpose, and it gives everyone a way to judge whether the study worked.

Research objectives versus a discussion guide

These are different documents, and a brief should hold the first.

  • Research objectives say what you need to understand. "Understand how people weigh cost against convenience when they pick a provider."

  • A discussion guide is a list of questions or topics in order. "Q1: How did you first hear about us?"

Objectives are about the decision. A guide is about the conversation. If you hand over a guide as the brief, you fix the method before anyone has looked at the problem, and you invite the interviewer to work through questions in order instead of following what matters to each person.

This is how Juno, the AI researcher from Hey Juno, works. A Juno study is built from research objectives, not a discussion guide. In each interview Juno works out how best to meet those objectives with that participant. If you need exact wording for some questions, you can add them as word-for-word questions and Juno keeps them verbatim.

Who do you need to hear from?

Name the people whose experience bears on the decision. Be specific about behaviour, not only demographics. "People who cancelled in the last six months" is more useful than "adults 25 to 44".

Also think about who might disagree with you. Include people likely to disprove your current theory: lapsed customers, non-users, people who chose a competitor. A sample made up only of friendly voices will confirm what you already think. For more on finding them, see participant recruitment for AI interviews.

State where they will come from: your own customer list, a panel, or both.

What do you already believe, and want tested?

Write your beliefs down. This is not a weakness in a brief. Hidden assumptions are the problem, and stating them lets someone test them.

List two or three: "We think price is the main reason people leave." "We assume most users never open the settings page." Mark which ones you hold firmly and which are guesses. Whoever designs the study, human or AI, can then look for evidence for and against each one, instead of absorbing them unseen into the questions.

Constraints

Say what limits the work:

  • The date you need an answer by, and the date the decision is made.

  • Budget and the number of people you can reach.

  • Topics that are off limits, and any sensitive subjects.

  • Languages and channels (voice, text).

  • Anything that must be asked in exact words, such as a legal or regulatory statement.

  • Who needs to approve the findings.

Constraints are not obstacles. They let the person designing the study make honest trade-offs instead of guessing.

What does a useful outcome look like?

Describe what you want to hold in your hands at the end. A ranked list? A set of reasons people hesitate, with their own words? A recommendation between two options? Say who will use it and how.

Be honest about confidence. Qualitative research can show why people think and act as they do and can surface things you had not considered. It cannot tell you how many people in a market hold a view. If you need a figure, say so, because that changes the method. See what is AI-moderated qualitative research.

An illustrative example brief

This is an invented example for illustration only.

Decision. Decide whether to move from monthly to annual-only billing for a small business accounting tool, before the next price review. What we believe. Customers who pay monthly are less committed and churn sooner (a guess). Annual billing will feel like a barrier to trying the product (a worry). Objectives. Understand how small business owners think about committing for a year. Understand what makes monthly billing valuable to them. Find out what would make an annual plan feel safe. Who. Current monthly customers, recent annual customers, and people who cancelled in the last six months. Constraints. Findings needed in three weeks. No questions about individual clients' financial details. Interviews in English. Useful outcome. A clear view of whether annual-only would lose people, and which reassurances matter most, with participants' own words as evidence.

Notice what is missing: a list of questions. The brief gives enough for the study to be designed well.

Common mistakes

  • A brief that assumes its own answer. "Find out why customers love our new feature" presupposes they do. Write "understand how customers use and judge the new feature." For more, see leading questions in research.

  • Too many objectives. Ten objectives means ten shallow answers. Aim for a few that serve one decision. Put the rest in a later study.

  • Asking participants to design the product. People are good at describing their problems, habits and reactions. They are poor at predicting what they would want or buy. Ask about experience, then do the design work yourself.

  • Skipping the decision. Without one, findings drift into interesting but unused trivia.

  • Treating the brief as fixed. A good brief changes after someone challenges it.

How Juno turns a brief into a study

You can brief Juno in plain language, without research jargon, and upload background files. Juno listens for the real problem underneath before it designs anything.

  • Objectives. Juno turns the brief into research objectives, not a question list.

  • Assumptions to confirm. Juno shows the calls it made in reading your brief, so you can confirm or correct them before anything runs.

  • Pushback. If a brief assumes its own answer or targets the wrong people, Juno says so and proposes a change. You decide whether to take it.

  • Bias check. The design is checked for loaded words, built-in assumptions and either/or framing.

  • Rehearsal. You can preview the interview yourself or run a simulated interview to catch weak wording, screening, pacing and coverage problems before real participants take part. Simulated interviews are for rehearsal only. They are never research data or findings.

  • Readiness. If the study is not ready, Juno names the unresolved risks and what should change first.

You can see more on the brief page.

Where this doesn't apply

This guide is about briefing customer research with real participants. It is not a template for a quantitative survey or for market sizing, where sample size and question wording are fixed up front. A tightly regulated study may need a more formal protocol than a brief. And a brief cannot replace judgement: it is a starting point that whoever designs the study should test and reshape.

Frequently asked questions

How long should a research brief be?

One page is usually enough. A paragraph on the decision, a few objectives, the audience, your hunches, constraints and the outcome you want. Add background files for context rather than padding the brief itself.

Should I include the questions I want asked?

Only the ones that must be asked in exact words. Everything else is better expressed as objectives. If you list all your questions, you narrow the study to what you already thought to ask.

Do I need to know research methods to brief Juno?

No. Describe the decision in plain language. Juno handles the method and explains the calls it makes so you can follow the reasoning, and it may challenge your first framing.

What if my brief is vague?

Say what you know and what you do not. A rough brief that names the decision is better than a polished one that does not. Juno, like any good researcher, will ask what you are trying to decide.