Skip to main content

Tool descriptions

The BrandIndex MCP publishes ten tools. Your AI client picks which ones to call based on what you ask, so you never name a tool yourself. This page is here so you can see what the agent has access to, and understand why it sometimes asks you for a region or a metric before answering.

A typical sequence is: find the region, find the sector, find the brand, then run the analysis.

get_regions​

Returns every region available to you. Takes no parameters and cannot be filtered.

get_sectors_in_region​

Returns the sectors (also called categories) in a region. Requires a region code, for example us, uk, de.

get_custom_sectors_in_region​

Returns the custom sectors in a region. On accounts that hold a lot of custom sectors this response gets very large, so searching by name is usually faster.

get_brands_in_sector​

Returns the brands in a sector. Requires both a region and a sector ID. The agent asks you for whichever is missing rather than guessing.

get_industry_tree_for_region​

Returns the full hierarchy for a region: industries, the sectors within each, and the brands within each sector. Useful for orienting yourself in an unfamiliar market.

get_analysis_filters_in_region​

Returns the demographic filters available for analyses in a region, with their valid values. Filters vary by region, so check here before filtering.

get_metric_filters​

Returns the metric filters available for analyses. A metric filter narrows the respondents to people who gave a particular answer on a brand metric, for example current customers, people who heard positive Buzz, or people who would consider the brand. Metric filters work the same in every region and sector, so this tool takes no parameters.

  • Available metric filters: ad awareness, awareness, buzz, consideration, customer status, impression, purchase intent, quality, recommend, reputation, satisfaction, value, and word of mouth.
  • Three-point metrics filter on positive, neutral, negative or unaware. Yes/no metrics filter on yes, no or unaware. Customer status filters on current, former, never (aware) or never (unaware).
  • A metric filter applies to the brand being analysed. You cannot score one brand among another brand's customers.

Finds brands, sectors or custom sectors by name. You can search one type or all three, optionally within a single region. Matching is a string similarity search rather than a semantic one, so check the result is the brand you meant.

execute_analysis​

Runs one or more analyses and returns the data. This tool produces the numbers.

  • Each query names an entity, a date range and the metrics you want. An entity is either a region plus sector plus brand, or a custom sector.
  • Up to 10 queries per call, so you can retrieve several brands or metrics together.
  • Between 1 and 4 metrics per query - split into multiple queries to ask for more than 4 at once. If you haven't said which metric(s) you want, the agent asks rather than picking for you.
  • Available metrics: ad awareness, aided awareness, attention, buzz, consideration, current customer, former customer, impression, index, purchase intent, quality, recommend, reputation, satisfaction, value, and word of mouth. Some have a common name the agent also recognises - "Awareness" for aided awareness, "Purchase Intent" for the purchase intent metric, "WOM Exposure" for word of mouth.
  • Each metric has a default score type taken from the BrandIndex platform, so results match the platform without you specifying one. Buzz and other three-point metrics default to net score; two-point metrics default to positives. Other score types available where the metric supports them: positives, negatives, neutrals, positives and neutrals, negatives and neutrals.
  • Up to 10 filters per query, demographic and metric filters combined. Filters combine with AND across filters, and OR within a single filter's values.
  • Scores are calculated over the Total scoring population - all respondents, including non-customers. Aware-only or opinion-only populations aren't available.
  • Dates are absolute, in YYYY-MM-DD format. Your agent converts relative phrases such as "last quarter" into absolute dates before calling. The response includes the latest date for which data is available, so you can anchor date ranges correctly despite fieldwork lag.
  • Results use BrandIndex's default 28-day moving average unless you ask for something else. Two settings control this, and they're mutually exclusive:
    • Moving average - smooths a daily series over a rolling window (in days) ending on each date, without changing how many dates come back. Ask for it in whatever unit is natural ("a 4-week moving average," "smooth it over a month"); the agent translates it to days.
    • Periodicity - buckets the data into fixed periods (weekly, four-weekly, monthly, quarterly, half-yearly, yearly, or a single whole-period figure) and returns one score per period instead of a daily series. Ask for this when you want one number per period rather than a smoothed daily trend ("monthly figures," "a single number for the range").
  • Volume in the results is the respondent count (sample/base size).
  • The agent may mention that you can save the results as a chart. It won't save one unless you ask.

save_analysis_chart​

Saves a set of analysis queries as a new, permanent chart in BrandIndex and returns a link you can open to view it. Unlike execute_analysis, which only returns data for the current conversation, this persists the chart in the product.

  • Only runs when you explicitly ask to save, keep, or share an analysis as a chart - never automatically after execute_analysis.
  • Takes the same query shape as execute_analysis - one query per line/series on the chart. The agent reuses the queries you already ran rather than re-deriving them, where applicable.
  • All queries in one chart must share the same moving-average or periodicity setting - a chart has a single scoring setting covering every series.
  • Queries with a periodicity save as a bar chart, matching BrandIndex's own periodicity control; everything else saves as a line chart.
  • Current scope: static styling only.
  • This is the one MCP capability that writes something back to your BrandIndex account rather than only reading data. It creates a saved chart visible in the platform UI, the same way saving one manually would.
  • Available only through the MCP - there is no equivalent public BrandIndex API endpoint for this.