Research and analysis

UK food system over 10 years: Delphi, 2021

Published 29 August 2024

Food Standards Agency 

What was the challenge? 

The UK Food Standards Agency (FSA) is a non-ministerial government department responsible for ensuring food safety in the UK and representing the wider consumer interest. To do this we need to understand how the food system might evolve, and how best to respond to some of the challenges and opportunities. 

The key purpose for this assessment was to identify what the key changes might be in the UK food system over the next 10 years. 

What was the approach? 

The UK food system is a large and complex system, so predicting change is challenging. It requires a broad range of expertise, ranging from toxicological knowledge through to behavioural insights and agricultural knowledge. We were also a small team, with only two staff to undertake the assessment, so our approach needed to be relatively resource light, while taking in a very broad range of different disciplines and expertise. To try and meet those requirements, we adapted a Delphi process, influenced by work undertaken by the Hunt Laboratory at Melbourne University, and previously used for multinational elicitation work. Our key steps in the process were: 

1. An initial survey was undertaken with an expert panel. This contained free text responses to open questions to try and ensure we accessed a full range of opinion, as free as possible from framing bias. 
2. Using the survey outputs, a series of composite reports was created. These were based on a thematic analysis of the free-text outputs, which were coded to allow a rigorous analysis of key issues emerging. The composite outputs highlighted areas of consensus, and areas of disagreement. This output was disseminated to the expert panel. 
3. An online discussion forum was created with several different threads based on specific issues in the composite outputs. Expert panel members were invited to take part in the forum which was kept open for two weeks. The forum allowed documents to be uploaded, notifications of responses and new threads to be sent to participants, a range of polls to be included, and emoji-style reactions to comments. This forum allowed a much richer picture to be gathered, building on the composite outputs – generating additional insights by allowing the expert panel to engage with each other’s views. 
4. The final stage collated the findings from each of these phases into a single overarching assessment. This highlighted key changes identified by the expert panel, timescales related to those changes, any specific opportunities and challenges for the FSA, and areas for further information gathering. 

Some key considerations and findings were: 

1. Methods of identifying an appropriate pool of potential expertise to be consulted. We had pre-existing stakeholder lists and invited around 300 participants from academia, government, industry, and the third sector to take part. We received responses from just under 100 individuals. Some of these were only able to take part in either the survey or discussion forum, some could take part in both. We also asked panel members to outline their main area of interest so that we were able to identify any areas where there might be over or under-representation. This stage is key to having sufficient confidence that prominent topics were highlighted because of their significance, rather than simply because there is a preponderance of experts working in that area. 

2. We were lucky to get a good return on our invites. However, we did offer informal drop-in sessions beforehand to all invitees to explain what we would be doing and why. Although not many attended those, feedback from those who did was that it was useful to get an idea for what they would be doing, and it showed an early willingness to openly engage with the invitees. 

3. We used a commercial provider for the discussion forum, Loomio. Although some people struggled to get to grips with the forum, it was generally well received, with an attractive interface, good (and easy-to-learn) administrative levels, and useful functionality for polling (e.g. we were able to use dot voting to ask panel members to prioritise key issues, and other types of polling to investigate different timescales for issues that were identified). 

4. We did not have access to bespoke software for the thematic analysis of the survey responses. This meant using Excel to code the material, which was extremely time consuming. While we could use it as a workaround, using bespoke software would have made the process much easier. This thematic analysis was by far the most resource-intensive part of the project. 

What was the impact? 

This work allowed us to properly focus our ongoing research programme for the next two years. It also highlighted in a robust way the significance of certain current issues, allowing the organisation to prioritise them – for example, the increasing prevalence of household food insecurity. 

Written in 2023.