Split-sample (monadic) concept testing shows each respondent only a subset of the concepts you're testing, then compares results across groups. This guide covers how to set that up in Tellet, since a single Tellet project cannot show different respondents different subsets of concepts on its own.
Within a single Tellet project, stimulus randomisation shuffles the order concepts appear in for each respondent, but every respondent in that project is still eligible to see every concept attached to the question group. See Adding Media to Interview Questions for that randomised, single-project approach.
There is no respondent-level logic that holds specific concepts back from specific respondents inside one project. If your design requires that Respondent Group A only ever sees Concepts 1 and 2, and Respondent Group B only ever sees Concepts 3 and 4, you need a separate Tellet project for each group.
A project supports one screening quota configuration. Plan your concept groups and target sample sizes before you start building, since quotas are the mechanism that keeps each group's sample on target.
Confirm your final sample size and split before building anything. Decide how many completes you need per concept group, since this determines the Max Participants value and screening quota for each project.
Create one project per concept group. For a 4-concept test split into two pairs, create two projects: one showing Concepts 1 and 2, another showing Concepts 3 and 4.
Build the discussion guide separately in each project, attaching only the concepts that group should see. Keep question wording identical across projects so responses are comparable once merged.
Add identical screening questions to every project. Use the same qualification criteria across all projects so groups stay comparable, and so ineligible respondents are screened out before they reach the interview. Screening responses are saved even for disqualified participants, so you keep a record of who was excluded and why. See Screening Questions.
Set Max Participants on each project to the completes you need for that group. See Max Participants for how the limit and its 10% buffer work.
Enable a screening quota on each project if you need finer control within a group, for example balancing age or gender within the Concepts 1/2 sample. Only one Single-Select screening question per project can carry quotas.
Publish each project and allocate respondents evenly between them, whether you're recruiting through a panel or your own list, so each concept group fills at roughly the same rate.
Export and merge results after data collection. Each project produces its own export; combine the "Answers in columns" exports from all projects in a spreadsheet or analysis tool for cross-concept comparison. See Exporting Data.
Design | Projects needed | Trade-off |
|---|---|---|
Split into non-overlapping pairs (e.g. Concepts 1+2 vs. 3+4) | One per pair | Simplest setup, easiest targeting |
Every possible concept pairing (e.g. 1+2, 1+3, 1+4, 2+3, 2+4, 3+4) | One per pairing | Full pairwise coverage, but more projects to build, quota, and manage |
If you need all possible pairings, expect the extra setup and ongoing management to scale with the number of pairings, not the number of concepts.
No native way to show a subset of concepts to a subset of respondents inside a single project. Use separate projects instead.
Only one screening quota configuration is supported per project, so a project can't run independent quotas for two different concept groups at once.
Merging results across projects is a manual export-and-combine step; Tellet doesn't automatically pool completes from multiple projects into one dataset.