Start here
Workflow for AI-moderated research
In a Chat project, participants complete the interview independently in their browser. You define the topics and questions in the interview guide; xelper moderates the conversation, asks the configured or adaptive follow-up questions, and prepares the responses for analysis.
The following steps apply exclusively to Chat projects.
Workflow at a glance
1. Prepare the study and interview guide
Prepare the interview guide by organizing the conversation into topics and writing an opening question for each topic. Choose whether xelper should ask no follow-up, a question you provide, or adaptive probes. Additional instructions control how the AI approaches an individual topic.
Variants represent different versions of the same guide, usually in different languages. For each variant, you also choose the collection experience, from a survey-style flow to conversations using text or voice messages. Segments can restrict individual questions to particular audiences. Images and videos, a glossary, and placeholders in questions and links provide additional context.
2. Test the interview and links
Complete a full test interview under Test Interview Guide using realistic answers. Check the introduction, order, follow-up behavior, media, language, placeholders, and redirects. Auto-Fill provides a quick technical run-through, but it does not replace a content review.
Test every active variant and segmented link. Then share only the intended Field Link from Chatbot Links with participants. Test interviews remain separate from the regular analysis.
3. Manage fieldwork
Distribute the field links through your panel, community, or own recruitment channels. Manage fieldwork under Transcripts, where you can see incoming interviews, open individual transcripts, and track conversation length, response volume, and other field data.
Only interviews under Valid enter the regular analysis. Resolve undecided cases under Pending, while automatically or manually excluded conversations appear under Invalid. Test runs and archived interviews remain separate.
4. Add data to existing interviews when needed
In a Self Service project, Admin and Member can import data into existing interviews. A spreadsheet or SPSS file can add metadata, segments, and open-ended responses. Import does not create new interviews and is not a required step in the regular workflow.
5. Prepare the topic analysis
For coded analysis, Admin and Member prepare each topic. They choose the analysis mode, generate suggested categories or scale values, and edit the codeplan before xelper classifies the notes and creates the results. Analysis becomes available when the planned number of valid interviews has been reached and their processing is complete.
xelper creates the codeplans and analysis results. Admin, Member, and Client can then view the results available to them, but they do not start this processing themselves.
Chat projects using qualitative analysis have topic reports instead of coded category charts. The analysis-mode and codeplan preparation step does not apply to them.
6. Analyze and explore the results
The completed topic results page brings together the summary, category chart, and underlying notes. Filters narrow the view to particular segments, while comparisons and Matrix View contrast groups. Statements and chart bars link back to their supporting evidence.
Under All Notes, you can view, search, and filter notes across the project. Use Chat with Your Data to investigate additional questions in natural language and open the sources behind an answer.
7. Create the project-wide topline report
When all relevant topics have been analyzed, Admin and Member combine the findings into a topline report under Project Overview. xelper creates a written report and a presentation.
xelper creates the project-wide report and presentation once the agreed analysis is complete. Admin, Member, and Client can view and download the completed outputs according to their project access.
Next step
Start by preparing the interview guide. There you can configure the questions, follow-ups, variants, and interview settings, then test the experience before fieldwork.