Analyses: overview
Maps to:
analyses
An analysis is a request for one or more time series of BrandIndex scores. It is the core of the API: you describe what to measure and the service returns the calculated data.
Anatomy
An analysis contains one or more queries. Each query is an independent series — typically one line on a chart. A single analysis can hold many queries, which are calculated and returned together.
Each query combines the building blocks described in Concepts:
- an
entity— what to measure; - a
period— over which dates; metrics_score_types— which metrics and how to score them;- optional
filters,scoring,moving_average,resampleandconfidence_bound.
Working with analyses
| You want to… | Use |
|---|---|
| Run an analysis you define in the request | Execute an analysis |
| Run an analysis you saved earlier | Execute a saved analysis |
| List or retrieve saved definitions | Managing saved analyses |
| Get results as JSON or CSV | Output formats |
| Test whether a change is significant | Significance testing |
| Get real-world event annotations | Analysis events |
| Download an Excel/CSV file | Exporting |
Relationship to rankings
Rankings share the same building blocks and patterns as analyses. Learn the analysis model here and the ranking endpoints will feel familiar.
Running an analysis from an LLM client
The BrandIndex MCP wraps this endpoint in a tool an assistant can call, taking a simplified per-query input instead of a full analysis definition. Tool descriptions covers what that shape supports and where its defaults come from.