Datasets · region × topic × week

Social Discourse Index

Salience and stance of key topics across regions and time.

The base layer. Every topic in the taxonomy gets a weekly salience share and a mean stance per region, computed on transcribed speech rather than captions or hashtags. Confidence intervals ship with every point.

Coverage: US, FR, ES, GB, DE, IT, +7
Frequency: Weekly
History: 12 months
Unit of observation: region × topic × week
Labels: topic · entity · stance · sentiment · region
Delivery: API · Parquet · CSV

Schema

One record, fully specified.

The contract is stable across delivery formats. New fields are additive and versioned; existing fields never change meaning.

{
  "date": "2026-05-20",
  "region": "US-NY",
  "topic": "housing",
  "entity": "rent",
  "sentiment": -0.42,
  "stance": "negative",
  "share": 0.031,
  "views": 18432,
  "language": "en"
}

Field	Type	Description
date	date	Observation date (UTC)
region	string	Geographic region (DMA / country / city)
topic	string	L1 topic taxonomy
entity	string	Entity or concept
sentiment	float	Sentiment score (−1 to 1)
stance	enum	Stance toward the entity
share	float	Share of total discourse
views	integer	Estimated views for the record
language	string	Detected language

Request a sample dataset.

Tell us the market and the question. We return a real slice — schema, sample sizes and confidence intervals included — not a brochure.

Request access Read the methodology