Skip to main content
Help & Documentation

Run a deep analysis

Go beyond quick insights: Deep Analysis works through your dataset for top findings, cohorts, prioritized recommendations, and the context it is missing.

3 min readUpdated August 27, 2026

Quick Insights gives you fast observations. Deep Analysis is the longer pass - it works through the dataset properly and comes back with ranked findings, cohorts it detected, and a prioritized list of what to do about them.

Start one

  1. Open Discover and generate Quick Insights for your dataset.
  2. Click "Go Deeper".
  3. Wait while it runs - the progress shows stages including "AI Insight Synthesis".
  4. Read the results in the "Deep Analysis" dialog.
Note: The same analysis can be opened from the analyst pane in Analyze, so you do not have to go back to Discover to ask for it.

What comes back

  • Top findings - the observations it considers most significant. If it found none, it says "No significant findings detected" rather than padding the list.
  • Cohort Analysis - groups it detected in the data by itself, marked "Auto-detected".
  • Recommendations, grouped by priority into Critical Issues, High-Impact Opportunities, and Strategic Enhancements.
  • Missing Context - what it could not answer because the data is not there.
Tip: Missing Context is the most useful section and the easiest to skip. It tells you which question your dataset cannot currently answer, which is a shopping list for the next thing you connect.

Acting on the findings

"Explore Top Findings" takes you from the write-up into charts, so a finding becomes something you can check rather than something you have to trust. A refresh control re-runs the analysis when your data has moved on.

Heads up: If the run finishes with "AI synthesis was limited - results may be incomplete", treat the findings as partial and re-run. And if it fails outright you get "Analysis Failed" with a Retry - a deep analysis is a substantial job, and it is rate limited per workspace.

Everything here is a reading of your own data, not an outside fact. Findings and recommendations are worth checking against what you know about the business before anyone acts on them.