Regional sentiment breakdowns

A single national number for sentiment is rarely useful. A poll on energy policy looks one way in Texas and the opposite way in Massachusetts; a poll on EU defense reads differently in Warsaw than it does in Lisbon. The point of Saut has always been to surface the actual shape of opinion, and a flat percentage gets in the way of that.

Regional Breakdowns are now available on every poll in the workspace dashboard. The map renders in real time as new voices come in, and every cell on the map is linked back to the underlying voices so a user can drill from a colored region down to the specific posts that drove the reading.

How we resolve location

Location is inferred from three sources, in order of confidence: explicit profile location, geotag metadata when present, and a lightweight inference pass on the post text itself. The text inference is conservative — if Saut cannot place a post in a country with high confidence, the post is excluded from the regional view and only counted in the global tally.

We never store inferred locations against individual voices. The map is rebuilt on every read from the raw post-level metadata, which means improvements to the inference pipeline propagate to all historical data automatically.

Granularity

For US polls, the map resolves to state. For European polls, it resolves to country. Sub-national granularity for the EU is on the roadmap and will follow once we have a reliable cell-level inference baseline.