Marcus Andrey Study
Ler em português Library

UX Research · Step 2 · 6 min read

Writing a research plan

A research plan is a one-page agreement, written before any work starts, about what you'll learn, from whom, by when and for which decision. What goes in it, a filled-in example, the timings that make a schedule realistic, and the pilot session that saves the first real one.

A research plan isn't paperwork. It's the moment the team agrees on what it's about to learn, before anyone can be surprised by the answer. Without one, the product manager expected numbers, the developer expected a list of bugs, and you ran six interviews about something else. One page, read and agreed by the people who'll act on the findings, prevents all three disappointments.

What you'll be able to do after reading:

  • Write a one-page research plan with every part that matters
  • Get the team to agree on it before the study starts
  • Build a realistic schedule, with time for recruiting and analysis
  • Run a pilot session so the first real session isn't the test

1. What a plan is for

A plan does three jobs:

  1. It forces the thinking. You can't fill in "Decision" or "What would change the plan" without having done the work from Step 1.
  2. It makes an agreement. Stakeholders who've read and agreed the plan are much more likely to accept findings they didn't expect. Tomer Sharon wrote a whole book about this, It's Our Research, on getting stakeholder buy-in for research projects (Elsevier, It's Our Research).
  3. It's the brief for everything else: the recruitment brief, the screener, the interview guide, the consent form.

Keep it to one page. Long plans don't get read, and a plan nobody read is not an agreement. The card Research plan one-pager has the page as a template to copy.


2. The one-page plan

Research plan · The second invoice

Ana (designer) · v2, 4 March · agreed with Rui (PM)
Background

Only 38% of new accounts send a second invoice within 60 days. Half the team thinks setup is the problem; the other half wants recurring invoices.

Decision

Onboarding fixes or recurring invoices first, for Q2. Rui decides on 28 March.

The plan starts from the decision, not from the method.
Research questions
  1. How do freelancers bill a returning client today, step by step?
  2. What do they use instead of us for the second invoice, and why?
  3. What happened in their first week that made them stay or stop?
Assumptions on the table
  • "People stop because setup is confusing."
  • "Most clients are billed every month."
Method

6 remote interviews, 45 minutes, about real past invoices. Plus the week-by-week drop-off funnel we already have in analytics.

Participants

4 who sent one invoice in the last 90 days and none since; 2 who invoice every month, for contrast. Mixed trades. Invited in the app; €40 voucher each, one spare booked.

Schedule

Week 1: screener and booking · Week 2: interviews, two a day · Week 3: analysis, readout on 26 March.

Roles

Ana moderates. Rui and one developer watch every session and take notes. Whole team at the analysis session.

Ethics and data

Information sheet and consent before each call. Recordings in one restricted folder, deleted after 12 months. No client names in notes.

What would change the plan

Most invoice monthly and copy the last invoice by hand: recurring first. Most got stuck before the first invoice was paid: onboarding first.

Written before the first interview, so nobody can move the goalposts later.
Outputs and budget

A 30-minute readout and a one-page note. €280 in vouchers, about four days of Ana's time.

A filled-in one-page research plan. It fits on one screen, starts from the decision, and says in advance what result would change the plan. The company, people and numbers are invented.Diagram: Marcus Andrey, CC BY 4.0. Objectives after GOV.UK Service Manual, Plan a round of user research.

The parts, in the order people should read them:

  • Background. Two or three sentences: what's happening, and what the team believes.
  • Decision. What will be decided, by whom, by when. This line is why the study exists.
  • Research questions. Two to five, open, about behaviour (see Research questions that lead to decisions).
  • Assumptions on the table. The beliefs you're testing, written so people can see them being tested.
  • Method. What you'll do and why that method (see Choosing a research method). Mention the existing data you'll triangulate with.
  • Participants. Who, how many, how you'll find them and the incentive. Details go in the recruitment brief.
  • Schedule. Recruiting, sessions, analysis, readout.
  • Roles. Who moderates, who observes, who takes notes.
  • Ethics and data. Consent, recording, storage, deletion (see Research ethics and consent and Research data and privacy).
  • What would change the plan. The deciding evidence, written now.
  • Outputs and budget. What people will receive, and what it costs.

GOV.UK's guide to planning a round of research covers the same ground and puts the objectives first: what you're trying to achieve, which problems you're solving, which assumptions you want to test, and what you need to know now to make an informed decision (GOV.UK, Plan a round of user research).


3. Write it with the team

GOV.UK's advice is to "work with your team to agree what you want to learn" (same source). In practice:

  1. Draft the plan yourself, quickly, from the research questions session.
  2. Share it with the decision owner and one developer. Ask: "Is this the decision? Would these findings change it?"
  3. Fix what they say, then send version 2 to everyone who'll attend the readout.
  4. Don't start recruiting until the decision owner has said yes.

Disagreement at this stage is a gift. If the product manager says "Actually, we've already decided on recurring invoices", you've just saved a study (see When not to do research).


4. A realistic schedule

Schedules break in three places: recruiting, the number of sessions per day, and analysis. GOV.UK gives concrete numbers (GOV.UK, Plan a round of user research):

  • Recruiting: start "at least a week before" the first session. Specialist or disabled participants can take longer; GOV.UK suggests allowing up to a month when recruiting disabled participants through an agency (GOV.UK, Finding participants).
  • Sessions per day: in a lab, "no more than 6 one-hour sessions a day", with at least 15 minutes between them; when visiting people at home or work, no more than 4, with at least 30 minutes plus travel.
  • Analysis: "For every 2 hours of research, you should do 1 hour of analysis", ideally on the same day.

For remote interviews, two or three sessions a day is a kinder pace. You'll listen better in the third session if you're not exhausted from the fifth.


5. Pilot before the first real session

A pilot is a rehearsal with someone close to your participants, a colleague if nothing better, a day or two before the study. NN/g: "The point of the test is to test the study itself", and one or two sessions are usually enough (Schade, Pilot Testing, NN/g, 2015). It's especially important if you're new to the method, the topic is unfamiliar, or the study is unmoderated and instructions have to "stand on their own".

The pilot checks: Does the guide fit the time? Do people understand the tasks? Does the recording work? Does the prototype break on step three?


6. Invite observers

Findings land better with people who saw a session. GOV.UK suggests thinking about who would benefit from watching: product managers, designers, content designers, stakeholders (GOV.UK, Plan a round of user research). Put names in the plan's "Roles" line and book the time now. Tell participants who's watching: it's part of consent.


7. Common mistakes

  • A plan that's really a method description, with no decision.
  • Ten research questions for one round of six sessions.
  • No time for recruiting, or recruiting before the plan is agreed.
  • Analysis "after the sessions", with no time in the calendar.
  • Skipping the pilot, then losing the first real session.

8. Vocabulary

Term Meaning
Research plan A one-page agreement on what you'll learn, from whom, by when and for which decision
Round One batch of sessions with the same guide and goal
Pilot session A rehearsal that tests the study itself, not the design
Observer A team member who watches a session, usually silently, and takes notes
Readout The session where findings are presented to the team

9. Try it (no tools needed)

  1. Take the research questions you wrote in Step 1.
  2. Fill in the plan's sections on one page, by hand. Leave a section empty if you don't know it yet.
  3. Look at the empty sections. Those are the questions to take to your product manager.

Reflection questions: Which section was hardest to fill? Who on your team should read this plan, and what would you want them to say back?


10. If you already practise

  • Keep a plan template with the sections above, and make "Decision" and "What would change the plan" mandatory fields.
  • Link the plan from the readout and the repository entry, so anyone can see what the study set out to learn.
  • Practise the full plan in the lab Write a research plan.

Checklist

Learn more

Books, courses and videos to go deeper

Book links go to Amazon. As an Amazon Associate I earn from qualifying purchases, at no extra cost to you.

Choosing a research method · Recruiting participants · Write a research plan · The interview guide · Planning a usability test · Research reports people read

Saved only on this device.

↑ ↓ to move · Enter to open · Esc to close