Research Planning

Plan your research

Start with the right question

Before starting any research, ask: What do we need to learn right now to make the best decision? This question guides everything else: the phase you’re in, the method you’ll use, and how you’ll plan the work.

Get feedback and support early

Don’t plan research in isolation. Use these resources:

#design-research

Share draft plans, ask questions, and get quick feedback from the research community.

Design Jam

Present your research approach live and get real-time input from peers.

From question to insight*

Our planning process has six connected steps.

1. Clarity

   Agree on the problem space and what decision needs to be made. Research is only useful when it connects to a real choice.

   Start with what you already know

   Before planning new research, look for existing insights:

   • Use Ask Dovetail to search our research repository — query past studies, findings, and recommendations related to your topic. Ask questions like:
     • “What research have we done on document uploading?”
     • “What do we know about Medicaid enrollment challenges?”
     • “What have users said about [specific feature or service]?”
   • Ask stakeholders what research they’ve already done or commissioned.
   • Check analytics, support tickets, and other existing data sources.
   • Talk to team members who have direct contact with users.

   Ask Dovetail can help you plan smarter by showing you what’s already been learned, where gaps still exist, what methods have worked well for similar questions, and how your research connects to broader themes.

   You might already have enough to move forward, or existing research might help you ask better questions.

   Match your work to the right phase:

   • Discovery — We don’t know what we don’t know yet. Learn about the problem space, users, and context.
     Example: Exploring how to make Unemployment Insurance more intuitive and equitable
   • Problem Validation — Stakeholders think they know the problem. Confirm users experience it the same way.
     Example: Does document uploading actually frustrate users as much as we think?
   • Concept Testing — Test whether ideas match user needs and expectations.
     Example: Would users welcome a digital intake form?
   • Solution Testing — Confirm the solution is easy to use and works as expected.
     Example: Can users navigate a prototype intuitively?
   • Scaling & Evaluation — After launch, uncover why unexpected behaviors or problems are happening.
     Example: Why did portal calls increase after our update?

2. Questions

   Turn your knowledge gaps into clear research questions. These aren’t interview questions—they’re what your research needs to answer.

   Be specific. Instead of “understand users,” write: “What challenges do residents face when uploading documents for Unemployment Insurance?”

3. Refinement

   Decide what kind of answer you need and what evidence will be convincing. Choose the right scope and rigor.

   Choose your approach:

   • Qualitative research — Understand the “why” behind behaviors, needs, and experiences. Provides depth and context through conversations, observations, and co-creation. Best when you need to understand motivations, uncover new problems, or explore how people make sense of something.
   • Quantitative research — Measure the “what” and “how many.” Count behaviors, compare options, and validate patterns across larger groups. Best when you need to prioritize issues, estimate demand, or prove something is happening at scale.
   • Moderated research — A facilitator guides the session, can ask follow-up questions, and observe in real-time. Use when you need to probe deeper, test complex concepts, or work with participants who need support.
   • Unmoderated research — Participants complete tasks or answer questions on their own time without a facilitator. Use when you need broader input quickly, want to see natural behavior, or are testing something straightforward.
   • Desk research — Review existing reports, policies, analytics, and research before conducting fieldwork. Always start here to understand what’s already known.
   • Field research — Observe people in their actual environment to understand how systems work in practice and what the real context looks like.

   Most research combines approaches. For example: start with desk research, conduct qualitative interviews to understand the problem, then use a quantitative survey to validate how widespread it is.

4. Methods

   Pick the method that best answers your question. The method isn’t the goal — learning is.

5. Analysis and synthesis

   Plan how you’ll make sense of results before you collect them. Create a clear path from raw data to usable insights.

   Decide:

   • How you’ll capture and organize what you learn
   • What patterns or themes you’re looking for
   • Who needs to be involved in analysis
   • How you’ll move from observations to recommendations

   For synthesis and outputs, see Share research.

6. Integration

   Think ahead: how will findings be used across the team and organization?

   Real value comes from:

   • Sharing insights in formats people can actually use (not just reports)
   • Connecting findings to decisions that need to be made
   • Building on previous research rather than starting over
   • Making research accessible in the repository for future teams

Choose the right phase

If you want a quick refresher on phase guidance, see the Research Overview page and the Research Resources page.

Discovery

Learn about the problem space, users, and context when you don’t yet know what you don’t know.

Problem Validation

Confirm whether stakeholders’ assumptions line up with real user experience.

Concept Testing

Test whether ideas match user needs and expectations.

Solution Testing

Confirm prototypes are easy to use and work as expected.

Scaling & Evaluation

Understand why unexpected behavior happens after launch.

Planning essentials

Every research plan needs:

Goals

What do we need to learn to make a decision? Share with peers in #design-research or at Design Jam to ensure they’re relevant and realistic.

Recruitment

Represent New Jersey’s diversity across geography, age, gender, culture, language, and abilities.

Participant safety

We do not research with participants currently in crisis or in crisis within the past 12 months.

Use UserInterviews

Use UserInterviews.com for all participant management. See Conduct research for recruitment, consent, and data-handling guidance.

Create a screener

Every research study needs a screener to ensure you’re talking to the right people. A good screener:

• Confirms participants have relevant experience with the service, system, or problem
• Ensures demographic diversity across your sample
• Filters out professional research participants or those who don’t meet your criteria
• Helps combat fraud (see below)

Include screening questions that:

• Are specific to your research (not just demographic checkboxes)
• Require knowledge that only genuine participants would have
• Can’t easily be answered by AI or found through a quick web search

Example: Instead of “Have you applied for unemployment benefits?” ask “What documents did you need to provide when you applied?” or “Describe a challenge you faced during the application process.”

Combating research fraud

Research fraud is real. Bots and scammers use AI to pass screeners, fake their way through moderated interviews, and dilute data through survey responses.

Protect your research

Use open-ended questions, attention checks, and real-time interactions to make research harder to fake.

Protect your research:

In screeners:

• Use open-ended questions that require specific, personal knowledge
• Ask for details that would be hard to fake (e.g., “What was confusing about the instructions on page 3?”)
• Include attention-check questions buried in the screener
• Avoid questions that can be easily answered with a Google search
• Use CAPTCHA or verification steps in UserInterviews

In moderated sessions:

• Turn on video and require participants to keep cameras on
• Ask spontaneous follow-up questions that go off-script
• Request participants to show you something physical (a document, their screen, their environment)
• Use activities that require real-time interaction (drawing, sorting cards, navigating a prototype)
• Pay attention to unnatural speech patterns, delays in response, or overly perfect answers
• Listen for generic responses that could apply to any situation

In surveys:

• Include attention-check questions (e.g., “Please select ‘strongly agree’ for this question”)
• Use open-ended questions that require context and specificity
• Check for duplicate IP addresses or suspicious patterns in response times
• Look for identical or near-identical responses across multiple participants
• Review responses

Red flags:

• Vague, generic answers that lack personal detail
• Perfect grammar and overly formal language in casual conversations
• Delayed responses in video calls with audio that doesn’t match lip movement
• Resistance to turning on video or showing identifying information
• Cookie-cutter responses that could have been copied and pasted

When in doubt, trust your instincts. If something feels off, it probably is. Document concerns and discuss with your team before marking participant as completed.

Good questions

Keep questions short, clear, and neutral. Invite open, honest responses.

Instead of	Try
What do you like and dislike about the Anchor program?	Ask separately: What do you like? Then: What don’t you like?
Do you prefer the option with the blue button?	Which option do you prefer and why?
Do you like these colors?	Why do these colors work (or not work) for you?

Equipment

We have audio recording equipment available for in-person research. RØDE Wireless GO microphone systems can be picked up and returned to our Trenton offices.

Include equipment needs in your research plan if you’re conducting in-person sessions.

---

Using AI in research

AI can be a valuable partner throughout the research process—but it works best when used thoughtfully and with clear boundaries.

Planning and preparation

NJ AI Assistant — Use it to structure research plans, draft interview guides, and refine research questions. AI can help you think through edge cases and suggest approaches you might not have considered.

Ask Dovetail — Search across our research repository to find related studies, identify what’s already known, and discover knowledge gaps. It can suggest how similar questions have been framed in past research and help connect your work to broader themes.

AI is powerful for identifying questions or opportunities you might have missed. Use these tools early in planning, but always review suggestions critically and validate them with peers.

During research sessions

AI transcription and note-taking — When you’re conducting research solo and don’t have a dedicated note-taker, AI transcription tools can capture what’s being said so you can focus on the conversation.

Important requirements:

• Always get explicit consent from participants before using AI transcription
• Explain what the tool does and how the data will be used
• Give participants the option to decline—and have a backup plan for manual notes
• Review transcripts for accuracy—AI makes mistakes, especially with accents, technical terms, or proper nouns
• Never rely solely on transcripts; supplement with your own observations and notes about body language, tone, and context

Analysis and synthesis

Ask Dovetail — Can help identify patterns across qualitative data, generate initial themes, and organize findings. It’s especially useful when working with large volumes of interview notes or survey responses.

NJ AI Assistant — Can help structure findings and draft recommendations based on your analysis.

Critical rule: AI is a partner in analysis, not a replacement for human judgment. Always:

• Review AI-generated insights against your own observations
• Validate patterns with your research team
• Check for bias or misinterpretation
• Ground recommendations in actual participant quotes and behaviors
• Be transparent about where AI was used in your process

What AI shouldn’t do

• Make final decisions about research direction or recommendations
• Replace human facilitators in research sessions
• Interpret complex emotional or cultural nuances
• Stand in for thorough, thoughtful analysis
• Be used without participant knowledge or consent

AI is a tool that makes research more efficient. It doesn’t make research better on its own—you do.

---

Research plan templates

Choose the template that matches your research approach:

Remote Research Plan For small, lightweight studies conducted remotely via Teams, Zoom, or UserInterviews.

In-Person Research Plan For in-person projects like shadowing, intercept interviews, or contextual inquiry.

60-Minute Conversation Guide A structured guide for longer, in-depth research sessions.

30-Minute Conversation Guide A focused guide for shorter, targeted research sessions.

Templates are also available in the Research Resources section.

---

Our research planning process is adapted from the Mule Design research framework.