A Project is a research study in Tellet where AI conducts interviews with participants.
Draft → Ready → Live → Completed
↓
Awaiting Approval → Approved → Live
The project is being configured but isn't complete. You can add and edit instructions, create discussion guide questions, set up screening criteria, configure all settings, and save work in progress — but you can't publish or collect responses until required fields are filled in. The project becomes Ready automatically when all required fields are complete.
The project is fully configured and can be published or sent for approval. Required fields include the project title, at least one discussion guide question, instructions, and a maximum number of respondents — see the complete list of 12 required fields in the First Project guide. From here, an Organisation Admin or Owner can publish directly, while an Editor requests approval.
An Editor has requested approval and an Admin or Owner must review. Admins and Owners receive a notification, the project can still be edited, and the reviewer either approves or requests changes. (Owners and Admins skip this step — they publish directly.)
An Admin or Owner has approved the publish request. Anyone with edit access to the project can now publish it.
The project is published and collecting responses; the status badge reads Live. A unique shareable link is created, participants can access the study, the AI conducts interviews, and responses are collected in real time. You can monitor incoming responses, view completed conversations, start analysing data, share the participant link, and complete the project when finished. You can't make major configuration changes — only a limited set of settings can be edited while published. Projects with the status Live can't be deleted — unpublish the project or complete it first.
Data collection has stopped: the project accepts no new participants, all data is preserved, and analysis features stay active. You can analyse all data, create and modify categories and filters, export results, and delete the project if needed.
The project is removed from the workspace and disappears from the project list. The project is hidden rather than permanently erased, and may be recoverable by support.
Instructions guide the AI's interview behaviour — research context and goals, the desired tone and style (formal, casual, empathetic), how deeply to probe, and topics to emphasise or avoid.
The questions asked to participants. Questions can be open-ended (rich insights), specific (focused information), or probing follow-ups, and you can attach images, videos, or audio to a question.
Screening ensures participants match your target audience. Questions can be single-select, multi-select, or numeric, and each answer is set to Qualify (allows participation), Disqualify (prevents it), or Irrelevant (doesn't affect qualification).
Set the maximum number of participants to control research scope, manage costs, and define when the study is complete. For guidance on choosing a number, see Maximum Participants.
The project must be Ready before publishing.
Completed conversations are full interview transcripts; incomplete ones are from participants who started but didn't finish.
Each question appears on a single page that combines an AI-generated summary (themes and patterns) at the top with the participants' actual responses below it.
Labels applied to responses for organisation and analysis. Categories can be single-select or multi-select; use them to tag responses, count themes, and bring quantitative analysis to qualitative data.
Segment data to analyse specific groups: text filters (specific words, screening answers), numeric filters (ranges, amounts), and date filters (time ranges).
Export findings for sharing. A report contains the response summary, key findings and insights, conversation transcripts, and analysis data (categories, filters).
Both actions require the Owner or Can Edit role in both the source and destination workspaces.
*Subject to Organisation role approval requirements