Detailed analysis for individual research questions. Organise responses into categories and view AI-generated insights.
Navigate to Results & Analysis → Questions tab (second tab).
Use the dropdown menu at the top to select which question to analyse. Questions display as "Question 1: [question text]", "Question 2: [question text]", etc.
Each question is a single page. There is no Insights/Answers split — analysis and responses sit together, in reading order:
Summary — a paragraph synthesising the responses to this question.
Categories — the category breakdown (chart), followed by a Distinctive concepts cloud.
Respondent answers — a preview of individual responses, filterable inline by category, with links to full conversations.
Evidence — Key Themes, Opportunities/Divergent Views, an In Their Own Words voice-quote section, and Visual Evidence.
Scroll the page top-to-bottom, or use the section navigation to jump to a section.
Categories organise responses into labelled groups, such as "Information seeking" or "Shipping Concerns" for a question about why participants searched for a product.
Single Mode:
Each response assigned to exactly one category
Categories are mutually exclusive
Total percentages equal 100%
Use for: Demographics, yes/no questions, classifications
Multiple Mode:
Each response can belong to multiple categories
Categories can overlap
Total percentages exceed 100%
Use for: Open-ended questions where participants mention multiple themes
Categories display as a horizontal bar chart showing category names and response counts per category.
Chart Type:
Bar chart: Available in all modes.
Pie chart: Available in Single mode only (where every respondent falls in exactly one category, so the slices sum to the whole audience). Use the chart-type button in the top-right corner to switch. Multiple mode stays bar-only.
Display Options:
Tally view: Shows numbers (e.g., "15 responses")
Percentage view: Shows proportions (e.g., "45%")
Toggle between views using the button in top-right corner. (Tally/percentage applies to the bar chart.)
Sorting Options:
Default order: Creation order
Ascending: Lowest to highest count
Descending: Highest to lowest count
Download:
Use the "..." menu in the top-right corner to download the chart as a PNG. Long category names are wrapped so they stay readable in the exported image.
Click "AI Suggested Categories" button
Wait for analysis
Review suggested categories
Edit as needed
Click "Save"
AI generates:
Category labels
Category descriptions
Logical groupings based on responses
Click "Create My Own Categories"
Add categories one by one
Define name and description for each
Click "Save"
Add Category:
Click "Add Category"
Enter label and description
Tellet assigns a colour automatically
Edit Category:
Click in text fields to modify
Update label (display name)
Update description (defines what belongs in category)
Remove Category:
Click the delete icon (trash) next to the category
Minimum 2 categories required
Deletion is immediate: the category and its grouping of responses disappear from the chart and filters as soon as you confirm, and there is no undo or trash to recover it from
Reorder Categories:
Click and hold drag handle (six dots)
Drag to new position
Drop to save order
There is no built-in restore for a removed category. To bring one back:
Click "Add Category" in the Category Editor.
Re-enter the same label and description the category had before (for example, label "Information seeking" with the description you used to define it).
Click "Save".
Saving triggers re-categorisation, so responses that match the recreated category's description are grouped into it again. If you're not sure of the exact wording you used, or want Tellet to regroup responses from scratch, click "AI Suggested Categories" instead and review what it proposes.
The description field defines what responses belong in the category. Include:
Specific criteria
Examples of what fits
Boundaries of what doesn't fit
Example:
Label: "Shipping Concerns"
Description: "Any mention of shipping costs, delivery speed, shipping options, or delivery-related problems"
Minimum: 2 categories
Maximum: 30 categories
All categories must have both label and description
Labels must be unique
Filtering happens inline on the question page — there is no separate tab to switch to.
Click a category in the chart, or use the category chips above the Respondent answers list, to filter the responses below to just those respondents.
Select more than one category to widen the filter.
Click "Clear" to remove the filter and show all responses again.
On smaller screens or split windows, the right-hand question sidebar is hidden and a compact stats bar appears above the main content, showing the same essential metrics:
Responses — total participants who answered
Patterns found — number of recurring themes detected
Divergent views — number of minority or outlier positions
Voice quotes — number of verbatim quotes surfaced
Next to the chips, a View responses button opens the full Responses panel (same as "View all responses" in the wide sidebar). Categories are not duplicated here — they already appear in the Categories section of the main content. On larger screens the full sidebar returns automatically.
Two synthesis fields are generated per question:
Summary — a full paragraph synthesising participant responses, including common themes, patterns, notable differences, and direct answers to the research question
Short Summary — a condensed one-to-two sentence version used in overview displays and by the Research Assistant
Numbered list of recurring patterns with supporting context.
This section highlights outlier perspectives and actionable recommendations:
Outlier perspectives
Unique viewpoints
Actionable recommendations
A selection of respondent quotes chosen to illustrate the question's key findings. Each quote appears in the project's analysis language, with the respondent's original words preserved alongside when they spoke a different language. Click a quote to open that respondent's full conversation.
When respondents share images while answering a question, you see a short written summary of the patterns and notable details across those images, alongside individual standout images with captions and attribution.
If insights haven't been generated, click "Run" in the Overview tab and review the results when processing completes.
There is no minimum number of conversations required — analysis runs on however many completed conversations you have. With only a handful of responses, findings are directional, and the AI keeps its language tentative rather than claiming firm patterns.
When categories are edited and saved:
AI re-reads all responses
Responses re-assigned to categories based on new descriptions
Category counts update
The respondent answers reflect the new groupings
Changes take effect immediately after saving.