In this technological era, we have seen a huge increase in the number of reviewing sites in the internet. In case of online food delivery stores, these reviews are very important as they, on the whole express public sentiment towards a particular restaurant or cuisine. In this paper, we are proposing an approach to predict which cuisines and restaurants are “trending” in a country based on the analysis of social media. We mine social media platforms like Twitter for food-related tweets and extract these tweets by using our own manually curated food lexicon. From these tweets, we use similarity matching to extract the food items that were tweeted about and run each of these items through a cuisine classifier based on logistic regression and word2vec word embeddings. This is done for all the tweets and thus, we can get which cuisines and restaurants have been popular while, which restaurants are fading. Our approach can, therefore be used by restaurants to analyze which markets they need to expand into and also where they have to revamp their business strategies. © Springer Nature Switzerland AG 2020.