Influence Government

By Dr Ben Collier, University of Edinburgh, Dr Gemma Flynn, Strathclyde University, Dr James Stewart, University of Edinburgh, Dr Daniel Thomas, Strathclyde University

Influence Government: The fusion of behaviour change and targeted advertising in UK public bodies and law enforcement

The practices of private sector advertising and marketing have long existed in a mutual relationship with the practices of government – from wartime propaganda to public health messaging. Contemporary forms of marketing now draw on the vast targeting and profiling capacities of digital platforms, and critical scholarship has explored the capacity of these to be used for misinformation or in political campaigning. However, our research has identified something new: the use of digital targeted advertising by government departments and law enforcement for ‘influence campaigns’ aimed at directly shaping behaviour.

Our research on this began in a criminal justice context – the Cambridge Cybercrime Centre had been evaluating different kinds of law enforcement actions against cybercrime and found that one of the most effective interventions they studied appeared to be a six-month campaign of online adverts by the National Crime Agency targeted at the UK. Investigating this further, we found that this was part of a sophisticated, multi-site ‘influence operation’ involving surveillance, direct intervention, focus groups, and iterative development of messaging – the techniques of a modern marketing consultancy, fused with the operational data and capacities of a public agency. This is part of a wider phenomenon – of approaches originally developed in the Prevent counter-radicalisation programme beginning to spread to other areas of interest to law enforcement, including cybercrime and violent crime.

We found that these tactics are in fact being experimented with across government and in policing; they are a core component of the competency framework used by the Government Communication Service. This is part of a wider rise to prominence of behavioural science expertise in government – often associated with the Behavioural Insights Team – which is increasingly using the technologies and influence practices of the platform economy to drive behavioural change for social policy. Our further research has suggested that this may encompass a very wide range of targeting approaches indeed – for example, a GCS training podcast which claims that the Home Office have been using the purchasing data of people who had recently bought candles to target them through their smart speakers with fire safety adverts. While centralised campaigns have released less public detail on targeting, at the more local level there is evidence of further local targeting, such as identifying community leaders and influencers at the hyperlocal level and encouraging them to take part in adverts themselves. In criminal justice, there is evidence of these approaches being used by police, the National Crime Agency, and Violence Reduction Units under a larger expansion both of Prevent tactics and of wider ‘public health’ approaches to policing.

Our initial research suggests that targeted adverts based on online behavioural profiles, the use of influencers and ‘influence operations’, and advanced marketing strategies are now being used as part of frontline public service and law enforcement, demonstrating a complex and intertwined relationship between traditional operational duties and data. This serves three main functions – first, allowing public bodies to reach increasingly specific groups and subgroups and tailor messaging accordingly, secondly, ‘in the moment’ shaping of behaviour in particular digital contexts (such as when people search for particular topics, use particular language on social media, or view particular content), and thirdly, the broader shaping of culture – in this case, the cultures of groups deemed ‘at risk’ of engaging in particular kinds of sanctioned or harmful behaviour.

Although there are clear benefits (particularly in the context of the pandemic) to being able to target government communications, there are also some serious potential pitfalls which we believe have been insufficiently explored. There is the potential for serious unforeseen consequences – stigmatising groups who already face structural oppression through targeting and surveillance, causing potentially serious anxiety or harm. In some cases these practices could potentially have the opposite effect from that intended, with the targeting serving to spread the very unwanted narratives and behaviours they are aiming to counter. While this practice appears to be employed for the purpose of crime reduction overall, it has the potential if misused to impose deviance narratives on previously uninvolved groups, further increasing the likelihood of offending. More generally, there is the question of the relationship of these practices to democratic values – whether they are enacted ’top-down’ with messages and priorities set by centralised authorities or are actually developed and engaged with communities on the ground.

These advanced marketing approaches are more than just ‘communications’ and go far beyond media management – our research suggests that they are frontline policy interventions and need to be seen as such, and subjected to the same public debate, scrutiny and accountability as other such policies. These interventions have the dual effect of opening up the intimate spaces of citizen lives to state control on one hand and expanding the sources of data used by the government to target policy on the other. There are legal and ethical questions to answer – around the selection of particular groups and characteristics, the use of operational data to inform these campaigns, privacy and data rights concerns, and the algorithmic aspects of the targeting itself and the data which this generates and relies on. Conversely, there are potentially areas where it could be argued that the government has a duty to run some kind of responsive targeted campaign – where the targeted advertising infrastructure is already being used to maliciously target vulnerable groups (such as the advertisement of harmful and illegal services, the spread of misinformation, radicalisation by far-right groups, the spread of misogyny and racism, or targeting online scams at the vulnerable). In this case, even where there may be clear moral imperatives for the deployment of targeted interventions, careful and transparent handling of citizen data should still be prioritised.

Our early research on these methods is presented in the briefing paper which accompanies this blog. Although there is substantial academic work being done in researching digital influence and behaviour change, we argue that the role of the academy should not only be in developing these techniques, but in analysing, critiquing, and exploring how they are being deployed in practice by government.

All images c/o Canva

You can also watch this SCCJR Seminar with Dr Ben Collier and Dr Daniel Thomas which took place in May.