> ## Documentation Index
> Fetch the complete documentation index at: https://conveo.ai/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Question Coding

> Review question summaries, compare segments, and inspect the evidence behind charts

## What question coding does

Question coding brings together question summaries, structured answers, and AI-coded [facets](/docs/insights-and-reports/facets). Use it to review a study systematically and compare participant groups.

![The Question coding view listing coded facets for a study, including interview transcript facets such as visible emotional response pattern and expression-statement congruence alongside assigned persona and product category framing](https://mintcdn.com/conveo/J7U335wh7HY2F11u/images/generated/question-coding.png?fit=max\&auto=format\&n=J7U335wh7HY2F11u\&q=85\&s=c1f0bb315acb73b30a6ecfef3d778780)

A direct-count chart records selected answers or source data. An AI-classified chart interprets open responses. Check the source indicator before treating a category as something a participant explicitly selected.

## Open the analysis

Open **Question coding** in the study navigation. Use the overview to move between interview-wide facets, questions, screeners, and participant-data sections available in your study.

The analysis base includes completed interviews and excludes hidden, deleted, and test interviews. Question-specific results also depend on which participants reached the question. Routing and missing responses can make bases differ between questions.

## Review sample context

Start with the available screener, panel, uploaded-data, and quota facets. These show the information actually collected; a panel is not guaranteed to supply every demographic field.

Read a chart's base and missing-data information. Participant-level facets count distinct participants, while other facets can count interviews. This distinction matters for diary studies with several entries per participant. See [Facet counts and missing data](/docs/insights-and-reports/facets#counts-and-missing-data).

## Read summaries and charts

Review the available summary for each question, then inspect its charts. A question can be covered by interview-wide facets rather than having its own separate charts.

Select a chart value to inspect the corresponding detail or evidence. Open the facet name for its detail page. Quote and media evidence depend on the facet's source; demographic or uploaded values do not necessarily have a transcript quote.

## Compare segments

Apply facet filters to define a participant group. Charts update for the selected conditions. Saving a filter set also requests summaries for that segment; text generation may finish after the charts update.

Keep the filter definition with any exported conclusion. Favoriting a facet is a shortcut to that facet, not a saved snapshot of every filter or display choice. Recheck the active filters when opening a favorite.

For repeated comparisons, use a facet's crosstab view to compare it with another facet.

## Curate and export

Favorite useful charts, or use the chart menu to duplicate or hide them. **Show hidden facets** restores hidden facets to view. Hiding declutters the page; deletion removes the facet.

Use available CSV or PNG exports and check that the displayed base and filters match what you intend to share. See [Reports](/docs/insights-and-reports/reports) for presenting a narrative with supporting evidence.

## Create or correct a facet

Choose **Create facet** under the relevant question or in the interview-transcript section when an important dimension is missing. Specify what to code and whether the source should be that question or the whole conversation.

Use [Managing a facet](/docs/insights-and-reports/facets#managing-a-facet) for editing definitions, values, and re-analysis. Changing a name or coding instruction does not, by itself, establish that existing interviews have been recoded. Run re-analysis when needed and review its results.

## Interpretation and troubleshooting

If a chart is empty, check the active filters, question exposure, available source data, and extraction status. A missing value is not automatically a negative answer or zero.

AI-coded categories and generated summaries can be wrong. Review source responses, check counterexamples, and correct ambiguous coding definitions. Multiple-value categories can overlap, so their percentages need not add up to 100%.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.