Paola BoltniewPAOLA BOLTNIEW
Back
01The Curiosity Lab

Protection App Sentiment Map

I extracted online ratings and reviews of digital protection apps using Claude Code and used them to gather insights: what customers feel, how they rate it, and which concerns show up together.

The artifact: An interactive sentiment map built from the same review corpus. Each dot is a theme, sized by how often it's mentioned and coloured by the ratings behind it, with lines drawn between themes that appear together more often than chance.

Protection App Sentiment Map — 1,179 app store reviews · 58 themes · 2025-09 to 2026-09

Built with

Claude Code, Claude, ChatGPT, Base44

The problem

Most people would take 7,500 reviews and write a report. I'm doing that too, because a report is the right format for findings and recommendations.

But a report reads in a straight line, and what customers say doesn't work that way. Price, confusion about what's included, support frustration and quiet reassurance all show up in the same reviews and feed into each other. A ranked list of percentages shows how big each theme is and hides how they connect.

Star ratings hide even more. A five-star review written because the app asked for one and a five-star review about a scam call that got blocked look the same, even though only one tells you the product earned that trust.

I wanted a way to see all of that at once, so the complexity is easy to take in without being flattened.

Why this

This map puts the voice of the customer in front of us in a form we can actually explore. You don't need to read 7,500 reviews or a long report to see what matters. You can move around it, follow a theme and see what sits next to it.

What it shows most clearly is the role these apps play in trust. When protection works, customers talk about feeling safe and looked after, and some of that goodwill carries over to the brand that provides the app. When it doesn't, the frustration rarely stays with the app. It becomes a question about whether the provider is worth paying for.

Seeing the data this way made a few patterns obvious that the numbers alone didn't. Price complaints tend to come with confusion about what the plan includes. Notification fatigue comes up most among customers who say the product works but are starting to lose patience with it.

AI helped me find the structure in thousands of customer comments. Deciding what that structure means, and what we should do about it, stayed with me.