Add simulated rows to hard-to-fill cells in an existing study
Extend an instrument you already fielded with simulated rows matched to your audience cells. We map the codebook and return the agreed format; you evaluate the result against a human holdout.
From your instrument to a scored comparison
Bring the instrument
Share the question stems, answer options, audience cells and field dates. Keep the human holdout on your side.
Map and run the study
We map your codebook, review the items and agree on the output format with you, then run the simulated respondents against the same closed-ended questions.
Score the comparison
We agree on the metric before the run. Predictions lock at delivery, and you compare the simulated rows with the human results you withheld.
Your instrument, your comparison
The purpose is to test whether synthetic augmentation is useful for a particular study. The deliverable follows your agreed codebook, with the synthetic rows clearly labeled.
Book a demoAgree before the run
- The question wording and answer codes.
- The audience cells and row format.
- The holdout and scoring metric.
- The labels that distinguish synthetic rows from human responses.
Best for
Evaluating synthetic augmentation against a withheld human sample using an agreed instrument.
Not for
Increasing the independent human sample size or its statistical power.
Keep the human and synthetic samples clear.
Synthetic rows do not increase the independent human sample size or its statistical power. The default delivery keeps them separately labeled. Any blended file is an explicit contractual choice, with the original human count retained.
We can scope a single instrument as a custom study, or a recurring workflow as an enterprise engagement. Item selection, codebook mapping and the holdout evaluation are part of that work.
Bring the question you’re working on.
We’ll help define the audience, instrument and deliverables. Scope and pricing are agreed before the work starts.
Existing customer? Open your workspace.