Workshop

Verifying AI Analysis in User Research

Sohvi Silius

AI in UX Research Consultant

You know you should be using more AI. But you don't trust it.

Research is a task on your table, whether it's your full-time job or something you do as part of other tasks. And you're expected to start using AI on it.

You've tried Claude or another LLM for qualitative analysis, and you don't feel like you can trust it.

You've noticed it making mistakes before.

It's doing a thematic analysis from interviews with wrong conclusions but no references, so you couldn't verify where anything came from. Sometimes one usability test participant with a strong opinion got too much emphasis.

You're hesitant to share the findings, because you can't recommend something if you don't know if it's true for sure.

You haven't really explored whether there's a way to make AI more trustworthy, because you're busy. It's easier to continue what you've been doing than set up a new process on your own.

In this workshop, we learn to do it together. You run a verification check on an analysis and read what comes back: what's incorrect, how significant it is, and what to do about it.

You'll build a process for checking qualitative analysis and synthesis, including open-ended survey answers.

Instead of relying blindly on a black box, you'll have a way to trace back every claim.

What you’ll learn

Whenever you have a qualitative analysis task where verification can help, you know how to approach it, with the tooling you have available.

  • A habit to start from: write your top-of-mind themes before you read the analysis, so you can see what it's missing.

  • Understand what AI doesn't deeply know about your context: your organisation, your business goals, which findings matter most.

  • You'll start from the templates I share and you take away.

  • Understand through examples what a good verification check will look like in your unique situation

  • An exercise on an analysis I provide, or with your own data

  • A demo in Claude. If your org uses something else, you do the exercise in that.

  • Support along the way while you run it

  • Templates and slides to take away, so the next project starts from something instead of from scratch.

Workshop agenda

  • Understand what AI can't know and how to keep your judgment

    What AI can't know about your study, and why your own themes have to exist before you read what it produced. You finish knowing which judgement should stay human.

  • Understand common LLM mistakes and decide what your check looks for

    You'll learn what are common mistakes that LLMs make, and which ones matter for the analysis in front of you.

  • Write your personalized verification prompt for qualitative analysis

    You'll build your own verification prompt with AI, from the templates I share and you take away.

  • Run your first verification and learn what to do next

    Turning what it found into a corrected analysis, deciding which findings actually matter, and making this something you reuse instead of something you set up again.

  • Make this part of your teams' workflow

    Get tips on how you can make verification a habit and a shared practice.

Learn directly from Sohvi

Sohvi Silius

Sohvi Silius

Contact

Independent AI in UX Research Consultant

See all products from Sohvi

Who this workshop is for

  • UX Researchers, Designers, Insight Strategists, Product Managers, Founders, and anyone else running qualitative research in their work.

  • You're already using AI, but you've had bad experiences with using AI in qual analysis. Trusting the output matters a lot in your role.

  • You're busy and don't have the time to learn things on your own; you want guidance and a reserved time slot for solving this problem.

Prerequisites

  • Subscription to an AI platform, preferably Claude

    You run the verification check yourself during the session. Claude is recommended, since I'll use that in the demonstrations.

  • Some experience of both conducting user/customer research and using AI

    This isn't a starting-from-basics session, on research or on AI. You have some experience using AI in research. Chat-level usage is enough.

  • Optional: bring a piece of your own work

    Bring both the raw data and the AI analysis created from it. There's demo material available for those that don't bring their own project.

What's included

Lifetime access

Go back to course content and recordings whenever you need to.

Community of peers

Stay accountable and share insights with like-minded professionals.

Certificate of completion

Share your new skills with your employer or on LinkedIn.

Maven Guarantee

Your purchase is backed by the Maven Guarantee.

Frequently asked questions

Maven for Teams

Reimbursement

Get your company to pay

Everything L&D needs: email template, receipts, and certificate of completion.

Get reimbursed

Private cohort

Run a cohort for your org

A dedicated cohort with a custom schedule and curriculum, tailored to your team.

Book a private cohort