Framework · Research · 4 min read
Research plan one-pager
A one-page agreement, written before any work starts, on what the study will learn, from whom, by when and for which decision, so nobody is surprised by what it delivers.
What will be decided, by whom and by when. The reason the study exists.
Read firstTwo to five open research questions about behaviour.
Not the interview questionsThe beliefs on the table, so people see them tested.
"People don't know the rules"What you'll do, why that method, and which existing data you'll check.
Interviews plus support ticketsWho, how many, how you'll find them, and the incentive.
Details go in the recruitment briefRecruiting, sessions, analysis, readout. Who moderates, observes, takes notes.
At least a week to recruitConsent, recording, storage and deletion.
Tell people who's watchingThe result that would change the plan, written now.
"If we see X, we'll Y"Use it when
- A product manager asks for "some research" and you want everyone to expect the same result.
- You're running your first study with a new team and need to agree roles, dates and what they'll receive.
- A previous study's findings were dismissed with "that's not what we needed".
How to use it
- Start from the decision. Write the background in two or three sentences, then the decision, its owner and its date. Bring the research questions from your question session. Output: the top third of the page.
- Show the assumptions and the method. List the beliefs you're testing so people can see them being tested. Name the method, why you chose it and which existing data you'll check alongside it. Output: the assumptions and method lines.
- Plan people and time. Say who you'll talk to, how many, how you'll find them and the incentive. Build the schedule backwards from the decision date: at least a week to recruit, and about one hour of analysis for every two hours of sessions. Output: participants and a schedule that fit before the decision.
- Fill in roles, ethics and outputs. Name who moderates, who observes and who takes notes. Say how you'll get consent and how you'll store and delete recordings. Write what people will receive and what it costs. Output: a complete page.
- Write what would change the plan. Add the deciding evidence: the result that would push the decision one way or the other. Output: one or two "If we see…, we'll…" lines.
- Get the yes, then pilot. Share the draft with the decision owner and one developer and ask "Is this the decision? Would these findings change it?". Fix it, send it to everyone who'll attend the readout, and book a pilot session a day or two before the first real one. Output: an agreed plan and a booked pilot.
Worked example
A team on an internal expense app wants to know why so many claims are sent back by finance.
| Part | What the plan says |
|---|---|
| Background | A third of last month's claims came back for corrections. The team believes people don't know which receipts are needed. |
| Decision | The product manager decides by the end of the month whether to redesign the receipt step or add checks before submission. |
| Research questions | How do people collect receipts during a trip? What do they do when one is missing? What happens after a claim is sent back? |
| Assumptions | People don't know the receipt rules. Corrections happen on the phone. |
| Method | Six remote interviews with a short walkthrough of a recent claim; read the return reasons finance already logs. |
| Participants | Employees who submitted at least three claims this quarter, found through finance; no incentive needed for internal staff. |
| Schedule | Recruit week 1, two or three sessions a day in week 2, analysis on the same days, readout on day 1 of week 3. |
| Roles | Designer moderates, product manager observes, a developer takes notes. |
| Ethics and data | Consent form before each session; recordings kept in the team folder and deleted after the readout. |
| What would change the plan | "If most returns come from missing receipts, we redesign the receipt step. If they come from wrong categories, we add checks." |
| Outputs and budget | A one-page readout and three short clips; no cost beyond people's time. |
Watch out for
- A method description with no decision. "We'll run six interviews" isn't a plan. Without a decision, findings have nowhere to go.
- Too many questions. Ten research questions for one round of six sessions means none of them gets a real answer. Keep two to five.
- No time for analysis. "We'll analyse after the sessions" with nothing in the calendar means the readout slips past the decision.
- Recruiting before the yes. If the decision owner hasn't agreed, you may be recruiting for the wrong study.
Template
Copy this into a document and keep it to one page. Leave a line empty if you don't know it yet: the empty lines are your questions for the product manager.
RESEARCH PLAN Version: Date:
Background (2-3 sentences):
Decision: Owner: By when:
Research questions (2-5):
1.
2.
3.
Assumptions we're testing:
Method (and why): Existing data to check:
Participants: who, how many, how we'll find them, incentive:
Schedule: recruit ........ sessions ........ analysis ........ readout ........
Roles: moderator ........ observers ........ note-taker ........
Ethics and data: consent, recording, storage, deletion:
What would change the plan:
If we see ............................, we'll ............................
Outputs: Budget:
Pilot session booked for: Agreed by (decision owner):
- It's Our ResearchTomer SharonBook · 2012 (Amazon)
- Just Enough ResearchErika HallBook · 2024 (Amazon)
- Plan a round of user researchGOV.UK (external site)
- Pilot Testing: Getting It Right (Before) the First TimeNN/g (external site)
- Plan user research for your serviceGOV.UK (external site)
"See on Amazon" links are affiliate links. As an Amazon Associate I earn from qualifying purchases, at no extra cost to you.