As already noted the panelists should not be trained and the panel must be representative of the intended population (Ask children to serve for products intended for children,......).
There are certain criteria for determining the panel size for "valid" conclusions (Please refer to: Hough, G. et al. Number of consumers necessary for sensory acceptability tests, Food Quality and Preference, (2006), 17:522-526). Many papers report a sample size of around 70 in hedonic tests.