Survey and decision gate
Does molecular compatibility track real liking?
A primary consumer survey tested whether flavour pairings the screen scores as more molecularly compatible are in fact liked more by real consumers. It is a validation of the screening method itself. Each of 34 respondents rated five concepts on a nine-point hedonic scale, spanning the full compatibility-score range.
A pre-committed decision gate was defined before data collection to prevent results from being read selectively. It required a positive rank correlation in the correct direction and a clear separation between the high tier and the low baseline. The result met the gate at the GREEN level.
Compatibility score versus mean consumer liking, one point per concept, coloured by affinity tier.
Mean hedonic liking per concept with standard-error whiskers. Dashed line at the neutral midpoint.
Per-concept results
| Concept | Compatibility | Tier | Mean liking | Purchase intent | Role |
|---|---|---|---|---|---|
| Coconut & Milk Tea | 0.79 | HIGH | 7.03 / 9 | 4.00 / 5 | Variant B |
| Mango & Jasmine | 0.73 | HIGH | 6.74 / 9 | 3.94 / 5 | Variant A |
| Pineapple | 0.61 | MID-HIGH | 5.18 / 9 | 2.82 / 5 | Variant C (out-of-sample) |
| Vanilla | 0.56 | MID-HIGH | 5.97 / 9 | 3.09 / 5 | Mid anchor |
| Strawberry | 0.26 | LOW | 4.32 / 9 | 2.41 / 5 | Discriminative baseline |