How divisio handles your data.

When you upload a CSV or Excel file, it's sent to our server for parsing only. The file is read into memory, converted to JSON, and returned to your browser — then immediately discarded. Nothing is written to disk, saved to a database, or stored anywhere on our end.

All your data lives in your browser tab. Close the tab (or refresh the page), and it's gone. We couldn't retrieve it even if we wanted to.

Two optional features use AI: persona cards send your cluster summary table (per-segment averages and segment sizes), and variable suggestions send column names with per-column statistics (counts, ranges, top category values). In both cases only aggregates are shared — never full rows and never your file.

These statistics are sent to Anthropic's Claude API solely to generate your result. Under Anthropic's commercial API terms they are not used to train their models — see the Anthropic privacy policy. Generated results may be cached briefly in our server's memory to avoid duplicate AI calls (cleared on restart).

AI features only run when you click their buttons. Every other feature works with no data shared with third parties.

We collect anonymous interaction metadata to understand how the app is used and improve it. This includes:

• Page views and button clicks
• File metadata: row count and column count
• Algorithm choices and clustering parameters
• Feature usage: exports, filter operations, elbow method

We do not track your actual data values, column names, filter contents, or any of the numbers inside your files. Analytics never includes your data. Separately, if you explicitly run an AI feature, column names and summary statistics are sent to our AI provider for that request only — see the AI features section above.

We don't use third-party cookies. The only things stored in your browser are:

• A random session ID (divisio_session_id) in localStorage — used to group analytics events within a session
• An authentication token if you sign in (managed by Supabase Auth)

You can clear these anytime via your browser's developer tools or storage settings.

Pro users can save clustering configurations — algorithm choice, number of clusters, selected columns, filter rules, and any custom segment names you've entered. These are parameter presets only and contain none of your actual dataset content.

Saved configs are stored in our database (Supabase) tied to your account, and you can delete them at any time.