PromptSpark.ai Docs
Sign in
Browse documentation
Docs Overview

Overview

Understand the Overview filters, KPI cards, comparison charts, rankings, sources, recent answers, and run health.

A reliable reading order

  1. Set the scope Confirm project, date range, provider, model, country, topic, and tag filters before comparing values.
  2. Check data health Review recent runs and completed answer counts. Partially failed runs can be reported, but their coverage is smaller.
  3. Read the KPI row Compare Visibility, Mention Rate, Share of Voice, Sentiment, Avg Position, Citation Coverage, and Domain Coverage.
  4. Compare brands and providers Use rankings, trends, and provider panels to find where the aggregate hides a meaningful difference.
  5. Verify the evidence Open recent answers and source records before turning movement into a business conclusion.

Core metrics

Metric Calculation Better direction
Visibility Distinct fully analyzed responses mentioning the brand divided by all fully analyzed responses × 100; rounded to one decimal. Higher
Mention Rate The same mentioned-response coverage converted from percentage to a 0–1 rate for calculation and displayed as a percentage. Higher
Share of Voice This brand's counted mentions divided by counted mentions for all brand rows × 100; rounded to one decimal. Higher
Sentiment Average mention sentiment on a −1 to +1 scale mapped linearly to 0–100; −1 becomes 0 neutral 50 and +1 becomes 100. Higher
Avg Position Arithmetic mean of recorded first-mention positions using only responses where a position exists; rounded to one decimal. Lower
Citation Coverage Explicit citation records divided by all observed source records; zero denominator produces no rate. Context dependent
Domain Coverage Distinct cited domains divided by distinct observed source domains; zero denominator produces no rate. Context dependent

Questions to ask when a metric moves

Start by asking whether the evidence changed or the measurement changed. Check active prompts, provider completion, market filters, competitor configuration, and response counts before attributing movement to brand activity. A clean comparison uses equivalent scope on both sides.

  • Did the change appear across several providers, or only one?
  • Is it concentrated in one topic, market, or small group of prompts?
  • Did a recurring source begin or stop appearing in relevant answers?
  • Did competitors gain mentions, appear earlier, or receive stronger recommendation language?
  • Does the answer evidence show a repeated narrative, or is the movement driven by one outlier?
  • Was there a known campaign, publication, website change, or reputation event near the movement date?