Understand results
Analytics and reports
Choose the right report and follow a result back to its tracked answers.
View as MarkdownEqualAI analytics describe answers collected for your tracked prompts. They are not a census of every answer an AI service gives, and do not measure website visits or crawler requests.
Choose a starting point
| Task | Open |
|---|---|
| Switch projects and review workspace usage | Dashboard |
| Review mentions, visibility, citations, and competitors | Overview |
| Manage questions and inspect prompt-level results | Prompts |
| Read the answer behind a metric | Responses |
| Investigate a cited domain or page | Citations |
| Understand tone around your brand | Sentiment |
| Compare model performance | AI Model Insights |
| Assess evidence-backed reputation dimensions | Reputation |
| Inspect captured background searches | Query Fanouts |
| Review captured ad placements | Ads |
| Review a saved weekly summary | Weekly Report |
| Investigate a URL target | URL Report |
| Open a previously saved artifact | Report History |
Compare like with like
- Select the same project.
- Check the date range and any comparison period.
- Match model, country, language, topic, and tag filters where offered.
- Read the metric definition before comparing screens. In particular, Citation rate does not mean the same thing everywhere.
- Open the underlying responses before acting on a change.
A percentage based on two answers is less informative than one supported by many repeated runs. A missing answer, unavailable provider evidence, and a recorded zero are different conditions. Clear filters and check run completion before interpreting an empty panel.
For formulas and known differences between screens, use the metrics reference. For data collection, see schedules. Website visits and automated requests are covered separately in Traffic Analytics and AI Agents.