AI for UX Research Course: Using AI Responsibly for Faster Qualitative Analysis

Learning Objective: Understand the fundamental principles of rigorous qualitative analysis and explore how and where to gain efficiency without sacrificing rigor.

Part 1: Foundations of Rapid Analysis (45 min)

The Tension Between Rigor and Speed

UX researchers face constant pressure to deliver insights quickly while maintaining research quality. This session explores when rapid analysis is appropriate versus when deeper analysis is needed, introducing the concept of "good enough" research as described by Courage & Baxter (2005). We'll examine industry pressures versus academic standards and establish a framework for making informed decisions about analytical depth.

Theoretical Grounding

Rapid qualitative analysis isn't about cutting corners—it's about strategic efficiency grounded in established methodologies. We'll explore the pragmatic approach to qualitative research (Patton, 2015), iterative data analysis frameworks (Miles, Huberman & Saldaña, 2014), and the rapid assessment process (RAP) which originated in anthropology (Beebe, 2001). These foundations provide legitimacy and structure to faster analysis methods.

Quality vs. Speed Trade-offs

Understanding trade-offs is essential for rapid analysis. This section covers establishing clear analytical boundaries, employing sampling strategies appropriate for time-constrained research, and maintaining transparency about methodological limitations. We'll discuss how to communicate these decisions to stakeholders while maintaining credibility.

Part 2: Preparation is Speed (1 hour)

Research Design for Efficient Analysis

The most effective way to speed up analysis is to design for it from the beginning. As Maxwell (2013) notes, "analysis begins at design." This session teaches participants how to structure discussion guides that naturally facilitate coding, create analysis-friendly templates, and embed analytical thinking into research planning rather than treating it as a separate phase.

Real-time Data Collection Techniques

Active listening and simultaneous analysis during research sessions can dramatically reduce analysis time later. We'll explore Krueger & Casey's (2015) approaches to focus group analysis adapted for one-on-one interviews, note-taking frameworks like the Cornell method adapted for UX research, and the power of debriefing immediately post-session while observations are fresh.

Team-Based Analysis Setup

This section covers strategic role assignment, parallel processing techniques, and how to use observers and notetakers to maximize efficiency. We'll discuss how to set up analysis teams for success from the project's inception.

Practical Exercise (15 min): Participants will redesign a traditional discussion guide to optimize it for rapid analysis, applying the principles covered in this session.

Week 1 Homework Assignment:

Apply preparation principles to your own upcoming research project

Create an analysis-ready discussion guide or refine an existing one

Identify 2-3 AI tools to explore before Class 2

Learning Objective: Apply streamlined coding methods and collaborative analysis techniques using both traditional and AI-supported workflows.

Part 1: AI-Augmented Qualitative Analysis (1 hour)

The AI-Human Partnership in Analysis

AI tools are rapidly transforming qualitative analysis, but they work best as collaborators rather than replacements for human judgment. Drawing on Amershi et al.'s (2019) human-AI collaboration framework, we'll examine where AI excels (pattern detection, initial categorization, processing volume) versus where human judgment remains essential (contextual interpretation, nuanced understanding, strategic framing). Christin's (2020) critical perspective on algorithmic analysis helps us understand the limitations and potential pitfalls of over-reliance on AI.

AI Tools Landscape & Quick Wins

The AI landscape for qualitative research includes automated transcription tools like Otter.ai, Descript, and Whisper, which can reduce transcription time by 80-90%. For coding and analysis, researchers can choose between general large language models (Chat-GPT, Claude) and specialized UX research platforms (Dovetail AI, Notably, UserTesting AI Insights). We'll discuss the advantages of each approach and where researchers see immediate time savings.

Practical AI Workflows

The most effective approach is a hybrid model where AI handles speed-dependent tasks while humans provide depth and interpretation. AI excels at initial categorization, pattern detection across large datasets, and summarization. Humans add irreplaceable value through context interpretation, understanding nuance, and strategic framing of insights. We'll walk through a complete workflow: AI transcription → AI-assisted initial coding → human theme validation → strategic insight development. Prompt engineering for qualitative coding will be covered with concrete examples.

Maintaining Quality with AI

The "human-in-the-loop" principle (Monarch, 2021) is essential for maintaining quality when using AI. This section covers validation strategies for AI outputs, bias detection and mitigation drawing on Noble's (2018) work on algorithmic bias, and transparency requirements when reporting AI-assisted analysis. Participants will learn how to verify AI-generated codes and themes systematically.

Live Demo (15 min): The instructor will walk through an AI-assisted coding session using real interview data, demonstrating prompt engineering, output validation, and the human refinement process.

Part 2: Rapid Coding Techniques (1 hour)

Streamlined Coding Approaches

Traditional grounded theory coding can be time-prohibitive for fast-paced UX work. This session introduces template analysis (King, 2004), which uses pre-defined codes as a starting framework, and provisional coding (Saldaña, 2021), which begins with a priori codes that can be refined. The Framework Method's matrix analysis approach (Gale et al., 2013) provides structure for organizing and comparing data efficiently across cases.

When and How to Skip Transcription

Full transcription isn't always necessary. We'll explore working strategically from notes and recordings, smart transcription approaches that capture only key quotes, and using audio/video timestamps instead of full transcripts. This section also covers when to use AI transcription versus when manual approaches provide better accuracy, particularly for specialized terminology or diverse accents.

Applied Thematic Analysis

Braun & Clarke's (2006) thematic analysis, adapted for speed, remains highly effective for UX research. This section teaches prioritizing pattern identification over exhaustive coding and applying the 80/20 rule to insight generation—finding the 20% of data that yields 80% of actionable insights. We'll compare template analysis and provisional coding approaches and discuss when each is most appropriate.

Technology and Tools

Beyond AI, various collaborative tools support rapid analysis. Miro, Dovetail, and Airtable each offer different strengths for team-based analysis. We'll also cover simple spreadsheet frameworks that can be surprisingly powerful. The key is combining AI and manual coding strategically rather than viewing them as either/or choices.

Practical Exercise (20 min): Participants will code sample interview data using template analysis combined with AI assistance, experiencing the hybrid approach firsthand.

Week 2 Homework Assignment:

• Practice AI-assisted coding on the provided dataset
• Try at least one new coding technique from class on your own data
• Prepare specific questions about your own analysis challenges for Class 3

Learning Objective: Produce credible insights rapidly and communicate them effectively while clearly articulating the role AI played in the process.

Part 1: Team-Based Rapid Analysis (1 hour)

Collaborative Analysis Sessions

Rainbow analysis and collaborative data analysis sessions bring multiple perspectives to bear on data simultaneously rather than sequentially. Analysis workshops using real-time affinity diagramming can accomplish in hours what might take individuals days. The modified KJ Method (Kawakita, 1991) adapted for team synthesis provides structure for collaborative insight generation. AI-generated summaries can serve as effective discussion starters, allowing the team to focus on interpretation rather than data familiarization.

Structured Team Techniques

The "top-of-mind protocol" involves immediate post-research debriefs where the team captures fresh observations before they fade. Research sharebacks—where analysis happens through presentation preparation—force synthesizing and prioritization. Dot voting and other prioritization techniques help teams quickly identify high-impact insights. AI tools can capture and organize team discussions, creating living documents that evolve through the analysis process.

Managing Team Dynamics

Effective collaborative analysis requires managing group dynamics to reduce bias and facilitate productive disagreement. This section covers facilitation techniques for ensuring all voices are heard, strategies for handling conflicting interpretations constructively, and clear documentation responsibilities so insights don't get lost in discussion.

Part 2: Building Your Personal Toolkit (1 hour)

Each researcher's context is unique, requiring a personalized approach to rapid analysis. Participants will create their own rapid analysis workflow, develop criteria for choosing techniques for different contexts, and build decision frameworks for determining when to use which approach. The focus is integration with existing workflows rather than wholesale replacement.