Analytics overview
Read every visibility metric, leaderboard, and trend the 42A dashboard shows - programmatically.
The same numbers your team sees on the 42A dashboard are available over the API. This section walks through the questions you'll most often want answered, the endpoint that answers each one, and what the response looks like.
Brand stats
Aggregated visibility for a single brand over a date range, with a daily history series.
Leaderboards
Top brands, prompts, and websites - scoped to a project or prompt.
Verticals
Visibility broken down by prompt category. Find your strongest vertical.
Topics
Auto-classified subject taxonomy. Top brands and domains per topic.
General vitals
Site-wide GEO/SEO/ADO checks (robots.txt, llms.txt, sitemap, …). Worst-first.
Page vitals
Per-page feature scores. Sort by score_low to find what to fix.
What's measured
Each brand mention 42A captures has these dimensions:
visibility_score- the headline number, 0-1. Combines mention frequency, position weight, and sentiment, modified by citation strength. Higher is better.mention_count- raw count of brand mentions surfaced through your prompts.mention_coverage- fraction of queries that surfaced at least one mention (queries_with_mentions / total_queries). A coverage of 0.4 means 40% of queries mentioned the brand.avg_position- average list position when the brand is mentioned. Lower is better. Position 1 = first item in the LLM's answer.sentiment_score- −1 (negative) to +1 (positive), 0 is neutral.trend- delta vs. the immediately prior window of the same length.trend.score_change > 0means visibility went up.
These dimensions roll up four ways: per brand, per project (analysis configuration), per prompt template, and per category (awareness / consideration / comparison / technical). The endpoints in the rest of this section let you query each rollup directly.
Time range
Every analytics endpoint accepts from and to query params (ISO 8601 date or datetime, both inclusive). Defaults: from = today 00:00 UTC, to = now. Maximum range: 30 days. Date-only values are widened — from=2026-05-01 means start-of-day UTC, to=2026-05-01 means end-of-day UTC.
Daily aggregates. brandScores rolls up at end of day UTC. The from / to fields echoed back in every response tell you the exact window the numbers cover.
Pagination, sorting, truncation
- List endpoints are cursor-paginated. Pass
limit(1-100, default 25) andcursorfrom the previous response'snext_cursor.nullmeans you've reached the end. See Quickstart §4 for a worked example. - Leaderboard endpoints (
top-*) cap at a fixed scan window - 5,000 rows for top-brands / top-websites / top-prompts and 2,000 rows per template forcategories/{category}/top-websites. Iftruncated: truein the response, you've hit the cap; narrow the time range to get accurate counts. - Org-scoping is automatic. Every endpoint is implicitly filtered to your authenticated org. User keys also need
X-Org-Id; see Authentication.
In Claude
Every endpoint here has a matching MCP tool. After installing @42a/mcp, Claude can answer the questions in this section directly:
"Use 42A - how did our visibility change in the last week, and which 3 prompts drove the biggest gains?"
That single prompt exercises get_brand_stats(from: "2026-05-08", to: "2026-05-14") and list_top_prompts_in_project(from: "2026-05-08", to: "2026-05-14", limit: 3) in one go. See the MCP overview for the full prompt-to-tool map.
Last updated