Purchase intent
Likely155
Open52
Unlikely21
Concept, ad, & pack testing
“Concept won overall. The priority cohort was n=38.”
Did the concept win where it had to?
A concept can win in total and still fail in the audience that drives launch. Simulacra predicts the concept's performance under different audience mixes, priority cohorts, and competitive contexts — using your existing research.
Where concept reads fail
The concept passed. Did it pass for the launch audience?
Underpowered
Soft on priority cohort
Total-sample win, n=38 in the strategic target. The signal exists; the precision doesn't.
Pack-test
Pack rotations split the sample
Within-pack rotations split sample further. Ten variants × three audiences = no cell big enough.
Mix risk
Audience-mix risk
"Will this hold if our buyer base shifts older / younger / urban / suburban?" doesn't have a clean answer.
Concept reads at decision-grade N
Predict performance in the audience that drives launch.
Cohort boost
Expand the priority cohort up to 10× — generated rows behave the way your respondent population behaves. Cell-stable reads on the question that drives the decision.
Audience-mix scenarios
Rebalance the sample to a target market mix and re-read the concept under that measured distribution. No new columns, no invented attributes.
Concept response varies by audience
Read intent, appeal, believability, and differentiation inside the priority audiences the original test could not support. Generated rows stay inside the measured concept-test schema.
Concept workflow
Pick the audience. Set the launch mix. Preview the read.
Start with a concept test you already fielded, choose the audience the launch depends on, and preview how the concept reads under the market mix you need to defend.
Concept variant
Priority audience
Boost changes read stability and row count, not the measured schema.
Audience mix
228 generated rows from 38 fielded rows in Heavy users, 25-34 urban, for a cut the fielded test could not support.
Fielded cohort38 rows
Generated rows228 rows
VariantMain claim
Audience mixLaunch target
Appeal
Strong169
Mixed50
Weak9
Believability
Believable157
Unsure52
Doubtful19
Differentiation
Distinct132
Familiar57
Generic39
Audience cut
Purchase intent share
Heavy users, 25-34 urban
68%
Category switchers
66%
Lapsed buyers
60%
Premium buyers
65%
- Readout
- Generated population, response distributions, audience cuts.
- Export
- CSV table, SPSS-ready cuts, generated-row flags, methodology appendix.
- Guardrail
- Unsupported cross-cuts are not estimated. Synthetic rows stay inside the measured concept-test schema.
Workflow preview. Pre-computed example; no Causal Engine API call.
Research, validated
The validations behind this use case.
Twin-2K-500
Response structure preserved across 717 variables
Twin-2K-500 validates Simulacra's preservation of relationships across hundreds of survey variables, the same property a concept cohort needs before expansion is credible.
Data Reduction80% reduction in sample needed
The data-reduction validation shows that response structure of the overall population can be predicted from a small, non-representative subsample.
Predict concept performance
Bring a concept test. We'll predict how it performs in the audience that drives launch.
Bring your concept-test data. Simulacra expands the priority cohort, rebalances under your target audience mixes, and predicts the concept's performance under future audience shifts and competitive scenarios.