research-analysis

⚠️ Always ask for explicit confirmation before publishing or posting anything to Confluence, Slack, or any other external tool — even if the person has already approved the content.

Research Analysis

This skill covers how to go from raw research data to meaningful insights — ready to be reported and shared.

---

Step 1: Data preparation

Before you start analyzing, get your data in order.

Organize and clean your data Gather all session notes, recordings, and transcripts in one place. Remove irrelevant content and make sure the data is readable and searchable.

Add a research context statement At the top of your analysis document, briefly describe:

• What you were trying to learn (research questions)
• What method you used
• Who you spoke to (number and type of participants, without PII)
• When the research took place

Create a mini-glossary (if needed) If your research touches on domain-specific terms or internal product names, add a short glossary so anyone reading the analysis understands the context.

---

GDPR — mandatory before using AI tools

Always anonymize your data before using any AI tool for analysis.

This is not optional. Replace all personally identifiable information before pasting anything into an AI tool:

• Use participant codes: P1, P2, P3, etc. — never real names or identifying details
• Remove company names, locations, job titles, or anything that could identify a person
• This applies to direct quotes, notes, and transcripts

If you're unsure whether something is PII, remove it.

---

Step 2: Coding and tagging

Participant-level analysis

For each participant, create a summary of what you heard. Capture:

• Key themes or topics they brought up
• Direct quotes (labeled with participant code, e.g. P1)
• Observations about behavior, emotions, or decision-making
• Anything surprising or contradicting your expectations

Theme identification across participants

Once you have participant-level notes, look across them for patterns:

• What did multiple participants say or experience?
• Where do their experiences differ, and why might that be?
• What topics came up most frequently?
• What are the strongest, most consistent signals?

Group your notes by theme, not by participant. This is where cross-participant patterns emerge.

---

Step 3: From findings to insights

There's a critical difference between a finding and an insight:

Finding	Insight
What you observed or heard	What it means
"5 out of 7 participants couldn't find the export button"	"The export function is invisible to users because it's buried in a secondary menu, which breaks a key workflow"
Descriptive	Interpretive

Synthesize findings into insights Don't just report what happened — explain what it means for the product or user experience.

Identify opportunities Based on your insights, what could be improved, explored, or addressed? Frame these as opportunities, not solutions.

Keep layers visible in your documentation:

• What participants said or did (direct data)
• Your interpretation of what it means (analysis)
• Recommendations or next steps (your professional judgment)

---

Using AI tools for analysis (CRAFTe framework)

When using AI to help with analysis, structure your prompt using the CRAFTe framework:

Element	What it means
C — Context	What is the research about? What are you trying to learn?
R — Role	What role should the AI take? (e.g., "act as a UX researcher")
A — Action	What specifically do you want the AI to do?
F — Format	How should the output be structured?
T — Template	Is there a specific template or structure to follow?
e — Examples	Can you provide an example of the expected output?

This framework helps you get more consistent, useful output — especially when working with large amounts of qualitative data.

Always review AI-generated analysis critically. AI can help identify patterns and draft summaries, but the interpretation and judgment are yours.

Read references/example-ai-analysis-prompt.md for a real example of a CRAFTe-structured prompt used at Jimdo.

---

Reference examples

When helping someone with analysis, read the relevant reference file and use it as a format template for your output:

What you're helping with	Reference file
Writing a participant summary	references/example-participant-summary.md
Doing cross-interview thematic analysis	references/example-cross-interview-analysis.md
Writing recommendations	references/example-recommendations.md
Building an AI analysis prompt	references/example-ai-analysis-prompt.md

These are real Jimdo examples — follow their structure, level of detail, and way of presenting quotes and themes closely.