(April 2026) I created a skill within Claud that anyone can use. When compared with the Chat GPT I built a year ago, it’s faster and completes thematic analysis just as intelligently.
The part of UX research that takes the longest isn’t the interviews – it’s what comes after. If you’re running a typical UX research project with six to ten transcripts, each an hour long, you’re looking at a significant chunk of time before you’ve even started to identify themes.
I’ve been using AI to support analysis for a while now and the results are genuinely impressive when you get the setup right. The problem is that general-purpose AI tools don’t behave the way a researcher needs them to.
To overcome this I built a GPT tool in 2025 which has been utilised over 100+ times and is currently rated 3.8/5. You can find it here.
Whilst the feedback has been predominantly positive, I checked the negative feedback to understand how to improve: The analysis is lengthy (a few hours), the reporting format isn’t to everyone’s taste, and work is required by the researcher to dive into the detail to extract value.
On the plus side, when the researcher does dive into the detail, it excels in supporting/disproving hypotheses and extracting supporting quotes.
So I built something more specific using Anthropic’s Claud AI tool. The general sentiment is that Claud has pulled away from Chat GPT over the past 12 months. The key drivers of this improvement include the release of the Opus 4.7, Sonnet 4.6, and Haiku 4.5 models, a 1-million-token context window, and vastly improved adaptive thinking to dynamically decide whether to apply deep reasoning to a task to speed things up.
What it does
The ‘Clicked UX Research Analysis Assistant’ is a Claude skill; A set of instructions that shapes how Claude behaves, designed specifically for multiple-transcript qualitative analysis. It works across discovery research, usability testing, and concept development.
Like the GPT tool, it takes you through a structured process: You provide context – the research topic, objectives, participant profiles and transcripts, before Claud generates the analysis & outputs.
It identifies recurring themes, acknowledges outliers, uses direct and paraphrased quotes with participant attribution, and cross-references findings against established UX heuristics from the Nielsen Norman Group when relevant.
By using the same transcripts on 3 different projects I was able to compare the GPT 5.4 and Claud Sonnet 4.6 models:
- In comparison with Chat GPT, Claud is faster; It averaged around 15-20 minutes for the thematic analysis
- Claud’s narrative is very coherent: The style of output is easy to read, more like a blog than a report, which makes it a little easier to interpret than Chat GPT outputs
- Claud maps NNg heuristics to the UX issues more effectively than Chat GPT
- It responds well to deep dives, on par with Chat GPT
- Chat GPT may be more single minded in it’s opinions, but neither model hallucinates or generates strange recommendations
Overall, the Claud skill is a little more refined than the Chat GPT tool but there’s very little in it. Both tools are equally valuable for researchers who need a second pair of eyes when doing analysis.
How to use it
Disclaimer: Don’t use it without doing your own analysis! This is hugely important. Let’s keep in mind that you ran the interviews and you observed the user actions (something AI can’t do). You need to be able to have a strong sense of what the most important events were and what the headline findings are. Run the analysis assistant after you have done your own analysis to support your findings.
You’ll need a Claude Pro account. From there, it’s a one-time install: Download the skill file from the link below, go to Settings > Skills in Claude, and upload it. After that, it’s available whenever you need it.
It’s free to use. Your transcripts stay in your own account, I will never have access.
Full installation instructions and the download link are below. Enjoy!
Download the ‘Clicked UX Research Analysis Assistant’ Claude skill here.
Installation instructions here.
Read my blog piece on the Chat GPT tool.
As always, if you have feedback or questions, get in touch.