Consumer taste preferences vary from one region to the next. But, for food manufacturers, there is currently no standard for sensory evaluation. So, when samples are sent to a trained sensory panel for assessment, the scores not only reflect the preferences of just one region. The Excel file, where the scores are recorded, also fails to capture the complexity of taste and smell.
What if it were possible to relate the scores from one sensory evaluation to consumer preferences around the world – and, in that way, support manufacturers when developing brands for multiple markets?
The team behind digital sensory science company Ajinomatrix believe they may have the solution. In collaboration with Eklo, they are investigating the development of open source business-to-business software that will make sensory evaluation faster, simpler and relevant across markets. The solution will make use of a machine and deep learning neural network.
Final report summary
We have proven feasibility through collecting advice from two industry experts and conducting a market study. The goal has been achieved entirely and surpassed with a pretotype. The next steps are financing a complete PoC (Validation Voucher or regional financing) and serialising production into a generic minimum viable product.