What the data reveals
The insights the engine surfaced
Processed across both markets, the corpus produces a clear brief. A small set of flavours dominates positive sentiment, the pain signal is complementary between markets, and the consumer segments hold across languages. These are the signals the molecular layer then acts on.
Flavour preferences and pain points
English flavour preference by net sentiment. Mango leads, followed by vanilla, caramel, coconut, and milk tea.
English pain points are led by chalky and gritty texture. Texture is the dominant complaint.
Cross-market reading
Read together, the two markets give a clearer brief than either alone. Mango, coconut, milk tea, and matcha are positive in both, which is what gives confidence that the flavour direction is not a local artefact. The pain signal is complementary: English foregrounds texture, China foregrounds aroma, so the combined corpus describes the full surface of the protein off-note problem.
Flavours positive in both markets define the MoBai flavour territory.
China competitor sentiment. Master Kong appears in the conversation with positive sentiment, relevant to the brief emphasis on Master Kong leverage.
Consumer segmentation
Segmentation runs on the joint bilingual corpus, so consumer types are defined by behaviour rather than by language. Both segments draw from both markets, confirming genuine cross-market consumer types rather than a language split.
| Share | Leading flavour | Leading pain | Occasion | Source mix (China / Twitter) |
|---|---|---|---|---|
| 81.5% | Coffee | Chalky | Morning | 20% / 80% |
| 18.5% | Coffee | Bad Smell | Morning | 45% / 55% |