The Analyze feature in HeyMarvin is a hands-on, detailed analysis tool that gives you complete control over your qualitative data.
Analyze lets you manually explore, group, visualize, and code your notes and themes. While it offers powerful AI capabilities, it can also be used entirely manually without any AI features.
Learn more:
Follow the steps below to run your Marvin analysis:
Step 1: Add notes to your files
Analyze runs entirely on note data, so before you start, your files must contain notes or annotations.
Choose how to create notes
You can choose whichever fits your workflow, or mix and match across files.
Manual annotation: Highlight transcripts or text to add context, create an auto-note, summarize text, or read and summarize graphs.
Auto-notes: Let Marvin's AI automatically generate notes and annotations across your files.
Survey imports: Importing a CSV or Excel survey converts open-ended responses into individual notes.
Learn more
Step 2: Open Analyze
Once your files have notes, you're ready to dive in. You can analyze them at the project level or across your entire repository. Marvin starts you off with every note at your fingertips, and the next steps make it easy to shape the data into exactly the view you need.
Choose your scope
Project-specific analysis: Open your project, go to the left-hand menu, and click Analysis > Analyze notes. This considers only note data within that project.
Cross-project analysis: Click Cross Project Analysis in the More Research Tools menu on the main navigation to look across multiple projects, narrowing by project groups, file tags, labels, and timeframes.
Tips for success
Use project groups as quick navigation or bookmarks to organize similar projects (e.g., by product team or methodology).
Keep coding consistent across the team using predefined templates and taxonomy to enable reliable Cross Project Analysis.
Learn more
Step 3: Group your data
Use the grouping options to organize your notes into a structure that fits your research, making it easy to focus on exactly what matters.
Pick a grouping
By notes: Displays all notes in an unstructured format.
By questions: Group notes by the questions in your discussion guides so you can compare all responses side-by-side—shown in a Kanban-style grid that's ideal for usability testing. If questions show zero notes, rerun your auto-notes.
By files: Groups notes by specific files—ideal when you want to focus on a subset of interviews or survey open-ends.
Learn more:
Step 4: Filter your data
Use the filters at the top of the page to narrow your dataset down to exactly the notes you want to work with. This step is optional—but it makes every analysis sharper.
Narrow your dataset
Labels and file tags: Filter by any predefined label or file tag.
Metadata: Filter by creation date, date ranges, participant demographics, or survey metadata.
Sentiment and keywords: Filter by sentiment, specific keywords, or notes marked as “important.”
Tips for Success
Differentiate file tags vs. note labels: use file tags for whole-file metadata (persona, methodology, product line) and note labels for point-level coding (pain points, feature requests).
Avoid overlapping tags and labels unless necessary (e.g., for personas) to keep your taxonomy clean.
Learn more
Step 5: Choose your visualization
On the right-hand side of the screen, toggle how your notes are presented. Each view is suited to a different kind of work.
Pick a view
Grid view (card view): The default card-based layout, best for quick scanning and manual interaction.
Table view (database view): A spreadsheet-like layout showing content, transcripts, labels, questions, links, authors, and sources, best for cross-referencing and editing details without extra clicks.
Graph view: Pie, bar, or chord charts generated from your labels, best for quantifying themes and visualizing relationships. Click any element to drill into the associated notes.
Tips for success:
Try the Chord Chart to visualize how labels interact, relate, and change over time.
Love spreadsheets? Use Table View—it acts like a database for cross-referencing transcripts, questions, and labels.
Step 6: Run AI and manual actions
Analyze puts you in the driver's seat: you direct the AI to exactly which notes to examine and what to do with them. With your notes selected, run any of the advanced AI actions below, or take a manual action.
Pro Tip: If no notes are selected, actions apply to your full filtered view.
Quick Analysis:
Quick Summary — a fast overview of your note data or a subset (e.g., survey questions).
Quick Thematic Analysis — surfaces themes with written counts. Unlike the advanced version, it skips charts/graphics and click-through links to video snippets, so sources must be located manually. Included in the Essentials plan and above.
Quick Trend Analysis — a faster, text-based look at trends in your note data.
Advanced Analysis:
Advanced Thematic Analysis — groups notes into themes and sub-themes with reasoning, adds visualizations (typically a pie chart), and provides clickable links to source notes, transcripts, and video snippets for easy verification. You can label notes directly and re-run the analysis on a filtered subset (e.g., "Challenges") to dig deeper.
Emotion Analysis — prioritizes the transcript over notes, analyzing the specific words used to group emotions and run sentiment analysis. You can click emotions to apply labels.
AI Note Sorting:
Find Notes for Your Labels (reverse thematic analysis): The flip side of thematic analysis—instead of the AI proposing new themes, you supply existing labels and the AI finds matching notes across your project, including newly added ones. It recommends matches against your labeling structure so you can bulk-apply labels in a few clicks.
Take a manual action
Bulk-label notes, stitch notes into a shareable video playlist, add findings to an Insight report, or export to tools like Jira. Playlists can be downloaded with or without subtitles, and Insights can be private, shared, or published to Discover Research.
Tips for success:
Choose Advanced over Quick thematic analysis whenever possible—it gives deeper insight, visual charts, and clickable links back to the source video and transcript.
Use “Find notes for your labels” for retroactive coding when you discover a new theme late—scan older files and apply the label across them.
Combine auto-notes with AI affinity mapping to cluster notes into thematic groups on a canvas; drag them into buckets or let AI suggest groupings.
Edit and rerun your analysis as many times as you like to guide the AI and refine results.
Learn more