02 / Case study · From data to design
I used a RAG and sentiment pipeline to turn product reviews into personas, a journey map, and design recommendations.
Role
Researcher & developer
Timeline
[Dates · duration]
Tools
RAG pipeline · sentiment analysis · [stack]
Methods
Review analysis · synthesis · journey mapping
The project in brief
Analyzed: product reviews with RAG and sentiment analysis. Made: personas, a journey map with pain points, and recommendation mockups. Evidence: [review count · source · validated findings].
01 / Problem
Finding the experience inside the data.
Product reviews hold detailed accounts of what people value and where an experience breaks down. I wanted to make that volume of feedback useful for design, not just summarize it. [Add the product context, audience, and decision this analysis needed to support.]
02 / Research
Listening at scale, then looking closer.
Methods / I analyzed product reviews with a retrieval-augmented generation (RAG) and sentiment pipeline. [Add the retrieval approach, model or stack, coding process, and quality checks.]
Data / [N reviews] from [source · product · date range]. [Add selection criteria, exclusions, and how I handled privacy or sensitive content.]
Measures / [Add sentiment definitions, theme coding, retrieval evaluation, and the checks used to assess the quality of the analysis.]
03 / Key findings
What the reviews brought into focus.
I translated patterns in the reviews into experience-level questions and design opportunities. [Add the supported insights and their evidence below.]
01 / [Add insight]
[Describe a recurring need and show the review evidence that supports it.]
02 / [Add insight]
[Connect a pain point to a journey stage, with context and supporting evidence.]
03 / [Add insight]
[Add a design opportunity and note what the review data cannot tell me.]
“[Add an anonymized, verbatim review excerpt.]”
[Review source · date · context · permission or attribution requirements]
04 / Design process
From patterns to a possible experience.
I created personas, a journey map with pain points, and design recommendations expressed as mockups. [Add the reasoning behind a recommendation, then show sketches, wireframes, and any prototype when available.]
![[Alt text: describe the personas and the review evidence behind them]](https://framerusercontent.com/images/9EhrNWQV3LjR7239EsAkTgesqg.jpg?width=2200&height=1467)
Personas / [Add the personas and their evidence]
Journey map / [Add stages, pain points, and opportunities]
01 / Sketches · [Add concepts from review insights]
02 / Wireframes · [Add the proposed flow]
03 / Mockups · [Add recommendations and prototype status]
Mockups or prototype / [Add the artifact URL]
05 / Outcome & results
A clearer route from feedback to design.
The outputs are personas, a journey map with pain points, and recommendations shown as mockups. [Add what was validated, what changed, and what remains a recommendation rather than a demonstrated outcome.]
[N]
[Number of reviews included in the analysis]
[Themes]
[Add a supported theme or sentiment finding]
[Next test]
[Add a validation result or proposed test]
06 / Reflection & next steps
A review is a clue, not the whole story.
I see review analysis as a starting point for asking better questions. My next step is to check the personas and journey map with [target users] and test [priority recommendation]. [Add limitations, including whose experiences may be missing from the data.]