What you will finish
A published set of choice tasks and an analysis of preferences for the attributes and levels you tested.
Before you start
Define the decision you need to support: for example, which service features matter most or how preferences change across tested prices. Prepare realistic attributes, levels and any combinations that must be excluded. Use administrator access and review the applicable study purchase and recruitment budget.
1. Create the choice tasks
- Open Form Studio → Research tools. Under Tool, choose Conjoint choice study.
- Enter Attributes — one per line, Name: level | level. Each attribute describes one feature; its levels are the alternatives participants may compare. Use your study’s own currency and price range.
- Set Estimation tasks, Concepts per task and Holdout questions (prediction checks). The controls support 6–24 estimation tasks, 2–4 concepts per task and 0–4 holdouts.
- Choose whether to include I would choose none of these. Use Block a combination for combinations that are impossible for your study.
- Choose Check design and resolve the listed design issues. Then choose Add to form.
Keep the attributes distinct: “delivery time” and “support hours” measure different features; several vague descriptions of “better service” do not.
2. Preview and publish
Read every generated task in Preview, including the mobile view. Confirm participants can compare the alternatives and that instructions explain the decision consistently. Save draft, attach the pages to the intended survey and review Publish changes.
Record the published version. Changing attributes or levels creates a different comparison and needs a separately identifiable round.
3. Collect the choices
Share the survey with your assigned participants, or follow the panel-survey workflow to quote, attach, fund and start an audience. Keep the task length realistic when setting the recruitment estimate.
4. Review the analysis
Open Results → Insights for the selected survey and version. Review the available conjoint results with the tested levels and response count in view. Keep holdout checks separate from estimation tasks. For a price or feature scenario, stay within the tested values and the assumptions displayed by the analysis.
Export the original choices when you need to reproduce a calculation or include the data in a report. Use the original question and version identities rather than combining different designs into one table.
Check the result
Your report identifies the audience, design version, tested attributes and levels, and the preferences observed in that round. It explains the choices tested and the comparison being made.
If something needs attention
If Check design reports an invalid design, review level counts and blocked combinations before inserting tasks. If results differ between exports and Insights, align versions, filters and quality decisions.
Next steps
Configure research questions, understand panel quality checks and export results.