Skills Under the Microscope: Prompt Engineering for UX Researchers

Pujit Siddhant

Sep 08 2026

<div class='bc_element' id='bc_element1' style='width:auto;padding:5px;max-height:100%;'><span><p class="no-margin startPlaceholder">AI is already becoming part of everyday UX research work. <a href="https://maze.co/resources/user-research-report-2025/??utm_source=WorkTote"> Maze’s 2026 research </a> found that 69% of surveyed researchers, designers, and product professionals were using AI in at least some research projects. Teams are using it to help plan studies, draft research questions, transcribe interviews, organize findings, and analyze data. That changes what prompt engineering means for a UX researcher. The useful skill is not memorizing prompt templates. It is knowing how to give an AI system enough product context, user context, research goals, and evidence to produce something that actually helps the study. Imagine you are preparing a usability study for a new onboarding flow. A prompt such as “write ten interview questions” may give you a usable starting list. The output becomes more valuable when the AI also knows who the users are, what the product team needs to decide, what previous research has already shown, and which behaviors you are trying to understand. You can then use it to review questions for bias, suggest follow-ups, challenge assumptions, or check whether the study still matches the original research objective. <b>Where AI Fits Into the Research Workflow </b> Different tools can be useful at different stages of research, so UX researchers benefit from understanding what each one is good at rather than trying to force the entire workflow into a single product. <a href="https://dovetail.com/changelog/new-chat/?utm_source=WorkTote"> Dovetail </a> is useful when a team already has a large body of customer research. Its AI can work across interviews, documents, research projects, and other customer evidence while keeping responses connected to the underlying source material. Teams can also provide ongoing context such as product terminology, strategy, and research guidance. That can make analysis more useful because the system already understands some of the language and background of the product. Maze is particularly useful during the study itself. Its AI capabilities can support study planning, question development, question refinement, AI-assisted research, transcription, and analysis of responses. Maze can also connect research data with tools such as ChatGPT and Claude, which gives researchers more flexibility to move between specialized research software and general-purpose AI tools. ChatGPT can support research projects where files, instructions, and previous conversations need to stay together. Deep research can also help with secondary research across current public sources. Claude offers project-based workspaces where documents and instructions can be maintained over time. NotebookLM is useful when the researcher already has a defined set of source material and wants answers tied back to those sources. AI can speed up parts of the process, but researchers still need evidence from real users. A summary generated from five interviews is only useful if it reflects what those participants actually said. If the AI identifies a theme, go back to the source material and check it. If it suggests that a question is leading, review the wording yourself. If it creates a neat conclusion, look for participants whose experience does not fit that conclusion. Researchers also need to be careful with the information they put into AI tools. Interview transcripts can contain names, company information, health information, contact details, or other sensitive material. Before uploading research data, check what the organization has approved, what participants consented to, and how the tool handles that information. Knowing when a prompt should not contain certain information is part of using AI responsibly. <b>Learn the Research Skill Behind the Prompt </b> Prompt engineering will keep changing as AI products become better at understanding context. Some tasks that require carefully written prompts today may become built-in features later. That makes the underlying research skill more important than any particular prompting technique. A good workflow starts with a research question you understand. Give the AI the objective, user segment, relevant product context, and the material it should use. Ask for a specific task, then review the result against the original evidence. The researcher still decides whether a pattern is meaningful, whether two users said similar things for different reasons, whether a finding matters to the product team, and whether the evidence is strong enough to support a decision. Staying current does not require following every AI launch. Follow a few sources that stay close to UX research. <a href="https://www.nngroup.com/?utm_source=WorkTote"> Nielsen Norman Group </a> regularly publishes work on AI and UX. Maze and Dovetail publish research and product updates on AI-assisted research. It is also useful to check the release notes for the AI tools you actually use. Every few months, take a research task you already know well and test whether a new feature genuinely improves the way you do it. <b>WorkTote Takeaway </b> For a UX researcher, adding “Prompt Engineering” to the skills section of a resume is only a starting point. Employers will learn more from seeing where AI was used in the research process and what the researcher did with the output. On a resume, connect AI use to actual research work. Instead of writing “Used AI for research synthesis,” explain that you used AI to organize interview material, validated the themes against the original transcripts, and used the findings to support a product decision. If AI helped improve a study plan, refine questions, speed up transcription, or compare findings across research projects, make the purpose and result clear. Your profile should also show the broader skills behind the tool. Experience with qualitative research, study design, synthesis, product thinking, data interpretation, research ethics, and stakeholder communication all make prompt engineering more valuable. Someone who understands those areas can judge whether an AI response is useful, incomplete, or simply wrong. A portfolio can show this in more detail. Explain where AI entered the workflow, what context you gave it, what it helped you do faster, what you checked yourself, and whether it changed the final research or product decision. That gives employers evidence that you can use AI as part of a professional research process rather than simply generate outputs with it. As AI tools become easier to use, the advantage will come from knowing what to ask, what evidence to trust, and what still requires human judgment. &nbsp;</p> <span></div>

Other stories

Powered by

RADAR108