Best way to test new flavor concepts with everyday consumers?

Honestly, do both but start with in-person taste tests — you can't replicate actual taste/texture reactions online. Recruit 30-50 consumers that match your target demo, run a blind tasting, and use a simple 9-point hedonic scale for scoring. Once you've narrowed to 2-3 winners, an online survey (Suzy or Respondent.io work great) can validate at scale before you commit to production.