Neidio i’r prif gynnwys
English Cymraeg
Prosiect ymchwil

Trends in Twitter conversations about food during 2019-20

This report gives a high-level summary of social media analysis divided by key categories to build our understanding of the views and concerns of the public when it comes to food. The analysis includes deep dive analyses on key category macro trends and key findings from audience analysis.


As part of our responsibilities, we work to understand the continuing evolution of the food landscape to identify opportunities to improve standards of food safety and/or authenticity. To do this, we use science and evidence to tackle the challenges of today, to identify and address emerging risks, and to ensure the UK food safety regulation framework is modern, agile and represents consumer interests. One way we build our understanding of consumer interests and concerns is through social media analysis, which permits real time monitoring of key issues relating to food safety and other consumer concerns.    

Research approach

An initial broad pool of around four million tweets mentioning keywords such as cooking, eating, or ingredients in March-August 2019, was used to identify prominent topics that could be considered to be trends. The first stage used broad keyword language rules to capture the wider food conversation, leading to the development of bespoke keyword search syntax and keyword exclusions for every trend to ensure relevant results were captured. Trends were grouped in seven categories:

  1. Food systems
  2. Ingredients and processes
  3. Health
  4. Intolerances and Hypersensitivities
  5. Diets, Habits and Routines
  6. Food Safety and Standards
  7. Sustainability

Year on year comparisons were then made to identify changes in conversation between 2018-19 and 2019-20 (September – August). To provide context to changes in trend categories, individual trends within said categories were grouped into a smaller number of macro trends.

Digital audience segmentation was also carried out on the ‘Food Safety and Standards’ and the ‘Intolerances and Hypersensitivities’ categories, to capture who was engaging with the topics. Audience maps were created, inferring interests by analysing the relationships and interactions throughout the audience, understanding how users are connected to, or engaged with, other users and accounts in the network.

Social media users are not representative of the general population (with social media use being more widespread amongst younger age groups) and different people use these media in different ways (with most posts being made by a subset of overall users, who again are likely to suffer from problems of representativeness). Findings presented here relate to relevant trends in twitter discussions, which do not necessarily reflect FSA views or future policy direction.


Research report