PersonaWorks

Inspiration

Product teams don’t just want “sentiment” — they want to know why customers feel a certain way, what changed after a launch, and which user groups are driving the narrative. PersonaWorks was inspired by the idea that a company’s real identity is shaped in public conversations, not just marketing decks.

What it does

PersonaWorks turns Reddit-scale product discussion into a research-backed customer satisfiability score across key dimensions like expectations, quality, and value. It also clusters users into personas, explains each metric with transparent reasoning and evidence, and tracks satisfaction shifts over time with a timeline view.

How we built it

We built a pipeline that ingests posts and comments from product-related subreddits, cleans and filters them, then extracts topics, aspects, and embeddings to power clustering. Gemini generates metric-level satisfiability scores with structured reasoning, and we store each run so new refreshes automatically produce a satisfaction timeline and “what changed” summaries.

Challenges we ran into

The hardest parts were managing noisy, unstructured text at scale and avoiding shallow “positive/negative” judgments in favor of real product drivers. We also had to design prompts and scoring outputs that are consistent, auditable, and UI-friendly, so users can trust the score and click into the exact reasoning behind it.

Accomplishments that we're proud of

We built an end-to-end system that goes from raw community discussion to an interpretable satisfiability index with persona segmentation and evidence-backed explanations. We’re especially proud of the metric expandability feature, where users can drill down from a score into the reasoning and representative quotes that justify it.

What we learned

We learned that perception is multi-dimensional — people can praise product quality while being unhappy with value, support, or expectations mismatch. We also learned that interpretability matters: teams trust dashboards much more when scores come with transparent explanations and counterpoints, not just a number.

What’s next for PersonaWorks

Next, we want to add event-aware satisfaction tracking (pricing changes, feature launches), richer persona modeling, and cross-platform sources like app reviews and YouTube comments. We also plan to improve evaluation with human-labeled benchmarks, making the satisfiability score even more reliable for real product decision-making.

Built With

• flask
• python
• react
• sqlite