Political parties have access to an array of data that can potentially inform how they campaign. Kate Dommett, co-author of a new book on the subject, explains what data-driving campaigning is, the different ways in which parties use it, the problems it can cause, and how it might impact the next UK general election.

Imagine yourself running a party’s election campaign. Your mission in the run up to polling day is to develop a strategy for securing electoral victory or an increased vote share. You know that according to the latest social science research that campaigns can have important, if small effects on electoral outcomes. You know you will need to work to locate existing supporters and encourage them to turn out and cast their vote. You will also likely need to persuade undecided voters of the virtues of your party. And you will need to contact activists and motivate them to get involved in the campaign by donating money, attending campaign events or sharing party messages.

The key to all these activities is being able to communicate with the right audience. You do not want to be sending a reminder to vote to supporters of your opponents, and you do not want to waste money sending advertisements intended to persuade people who already plan to vote for you. You therefore need to gather and sort information about people’s voting behaviour to develop tailored messages. You also need to work out what messaging and mediums are most effective for achieving your desired goals. What slogans have the desired effect, what appeals prompt donations, and what campaign action most effectively reminds people to vote?

This form of campaigning is nothing new, but in recent years the availability of new forms of data, the emergence of new and more sophisticated techniques for profiling voters, and new mediums for contact are seen to have heralded a form of ‘data-driven campaigning’. Rather than relying on their gut instinct, party campaigners now use data and analytics insights to construct their campaign as never before. And yet, whilst increasingly common, our understanding of what exactly is captured by the term data-driven campaigning is often opaque.

What is a data-driven campaign?

Our new book, Data-Driven Campaigning and Political Parties, sets out to understand the practice of data-driven campaigning. Using interviews with 329 campaigners in 18 parties in Australia, Canada, Germany, the United Kingdom (UK) and United States (US) we cast new light on this often-shadowy activity.

We define data-driven campaigning as a ‘mode of campaigning that seeks to use data to develop and deliver campaign interventions with the goal of producing behavioural or attitudinal change in democratic citizens.’ Identifying four components of a data-driven campaign, we argue that it is important to think about data, analytics, technology and personnel when attempting to understand this activity.

Our definition reflects the idea that the use of data in election campaigns is nothing new. In the UK, it has long been the case that political parties are given access to information about who is registered to vote at each address across the country, when someone becomes eligible to vote for the first time, and whether they did indeed cast their ballot on election day. This data is made available to parties by the state to encourage them to inform voters about the election and promote democratic engagement. In addition, we know that most UK parties have a long history of conducting doorstep and telephone canvassing, activities they use to collect information about whether someone is planning to vote and which party they are likely to support. They also use polling and focus groups, and even buy some information from companies like Experian.

We also know that to make the most of this data the majority of parties in the UK have developed national databases for storing the information they collect about individual voters. Using these databases they have been able to combine different pieces of information and build up a picture of how particular people at particular addresses say they are planning to vote. This information has long been important in working out who to contact and the kind of messages are most likely to resonate. But in recent times, advances in analytics techniques and newly available forms of data have allowed parties to develop more detailed insights into the behaviour and attitudes of citizens. Whilst this can lead to more personalised and tailored campaign messaging that is more relevant to individuals, it can also lead to more problematic practices.

The problem of data-driven campaigning

Just over five years ago, the Cambridge Analytica scandal highlighted the potential for the use and abuse of personal data for political ends. Suddenly not only journalists, but also academics, policymakers and the public were preoccupied with the potential for personal data to be used by political campaigners in democratically problematic ways. There were claims of an erosion of personal privacy, potential for manipulation, concern about fragmented public discourse, foreign interference, and much besides.

As a result, there have been calls for regulation of the use of personal data and restrictions on the type of targeted messaging that parties can perform. These calls are beginning to be reflected in new policy, but in our book, we argue that we currently lack the necessary empirical insight into the practice of data-driven campaigning to determine what form of regulation is required. Moreover, we also lack a clear conception of the precise problem that data-driven campaigning poses, and especially need more clarity about how and why new practices differ from the widely accepted use of data in the past. For these reasons, we argue there is a need to study exactly how this activity is being performed.

How does data-driven campaigning vary?

Within the book we take each of our four components of data-driven campaigning and map the precise forms of data, analytics and technology that may be available to parties, and the different types of personnel who can be involved in each component of this activity.

Whilst we often assume that parties have access to a uniform set of tools, within the book we show that practices are not uniform when it comes to data-driven campaigning. In some parties we see campaigners drawing on relatively few data points and mobilising simple analytics processes and technologies, whilst in others we see much more diversity and complexity. To capture these differences, we develop different classificatory categories that help us understand what is being done where.

Take data, for example. We might presume that parties in different countries can access the same types of data, but this is not always the case. Distinguishing between publicly available data, disclosed data, inferred data and monitoring data, and looking at the precise data that can be accessed in each country, we can see that German parties do not have access to the type of electoral roll available in the UK and US. We can also see that the specific data points contained within the electoral roll vary not only between country but sometimes also within a country (with different data available in different US states).

Drawing these distinctions and describing practices within parties, we show how practice varies not only in different countries, but also between parties in the same national context.

This finding is significant for concerns about the problematic use of personal data as it suggests that data-driven campaigning is not being done in one way. It may therefore be that only some parties are engaging in ‘problematic’ practices, whilst others are using data in entirely acceptable ways. This makes it important to think about what exactly is problematic and whether there are uniform standards that can be agreed on in different countries. We argue that such thinking has often been absent to date, suggesting the importance of further analysis of exactly what constitutes acceptable and unacceptable data-driven campaigning practices in different nations.

What explains different data-driven campaign practices?

In calling for such reflection, we also set out to consider why data-driven campaign practice varies and what kind of response could therefore be made to counter any perceived problem.

Our approach to answering this question reacts to a tendency to focus almost solely on the data and privacy regulation or party resources as influences on the practice of data-driven campaigning. These are important explanations for accounting for how data is and is not used, but we argue that there are a wider range of dynamics that can affect the way data-driven campaigning is conducted. In particular, we point to systemic, regulatory, and party-level factors that can affect the form of data-driven campaigning. To give just a few examples, whether a country has a majoritarian or proportional system will likely affect whether and how targeting is used. Similarly, at the party level, the degree of enthusiasm towards data amongst party elites can have an impact on the degree to which data is invested in and rolled out throughout a party’s structure, leading to different configurations of this activity.

Appreciating the reasons for different practices is, we argue, important for understanding the different types of intervention which could be made to promote more or less favourable forms of data-driven campaigning. To date policy makers have focused on the tightening of data and privacy regulation; we suggest that other party-level, regulatory and even systemic changes can have an impact on how data-driven campaigning appears and what form it takes. This suggests more avenues for change than often previously highlighted.

Data-driven campaigning at the next UK general election

In studying data-driven campaigns, our book provides new insight into the activities that define modern campaigns and the choices and dilemmas that campaigners face dependent on their particular context. In contrast to much prevailing commentary, we argue that the use of data is not inherently problematic but is rather something that can be done in more or less socially acceptable ways.

As campaigners ramp up their efforts for a likely UK general election this year, our research suggests that we are likely to see different parties engage in subtly different forms of data-driven campaigning. What remains unclear is whether and in what form we will see behaviours deemed to be problematic, and if parties themselves will try to counter public concerns about their campaigning practices.

Data-Driven Campaigning and Political Parties is already available to purchase. Unit readers can, for a limited time, get a 30% discount by using the code ASFLYQ6. If you liked this post, you might also wish to hear Kate speak on this subject at a free UCL Policy and Practice Seminar, Data-Driven Campaigning: How and Why do Political Parties do it? on 8 February.

About the author

Professor Kate Dommett is Professor of Digital Politics at the University of Sheffield. Her new book Data-Driven Campaigning and Political Parties is co-authored with Dr Glenn Kefford and Dr Simon Krushinski. Her wider research looks at the relationship between digital technology and democracy, particularly in regard to election campaigns and regulation.