Five Lessons from Dominic Cummings – How did Vote Leave win the Brexit Referendum?

I recently watched „Brexit: The Uncivil War“ – a highly entertaining, well-scripted drama about how the Vote Leave campaign managed to win the 2016 British referendum on exiting the European Union (Brexit). The Vote Leave campaign director Dominic Cummings is played by the excellent Benedict Cumberbatch. Cummings is a very interesting and controversial guy – an Oxford graduate in History that became a political strategist. He is a thinker and a doer – a rare combination. Now he is shaping history again as a special advisor to Boris Johnson for Brexit. The remain campaign was ahead in the polls, it arguably had more money and power behind it, and status-quo campaigns almost always win referendums: How did Vote Leave manage to win? The answer to this question is highly relevant for all political campaigns and marketing in general – no matter what one thinks about the politics behind it. Cummings himself provides an answer:

How Vote Leave won the Referendum

Here is a summary based on the video above and some additional sources.

You can also go directly to the Five Lessons learned from this case study in campaigning.

• The first thing to note is that the result of the referendum was very close (51.89% vs. 48.11%): it’s easy to imagine circumstances where Vote Leave could have lost due to small differences. As Cummings points out, it’s always possible to construct a big story after the fact to explain the results („Brexit happened due to racism/the financial crisis/immigration etc.“), but the truth is that it could have easily gone the other way. Cummings calls this contingency of history „branching histories“.

• Vote Leave did focus groups to find out what the public really think about the EU. Cummings notes that the general public thinks unlike the people in London and the media. He found out that the public knew almost nothing about the EU. But, crucially, the public associated the EU with the waves of immigration since 2015, the financial crisis of 2008, and the Euro Crisis. All associations are negative: trust in the EU and the establishment was undermined.

• Cummings used polls to get an idea about the size of different groups and how they wanted to vote: One third of the population were staunch remainers, another third definitely wanted to leave no matter what, and a fifth wanted to leave, but was scared to do so. This clustering gave the campaign its goals: de-motivate the remainers, get the one third of the leavers to turn out and persuade the scared fifth to vote leave.

Kenneth Allen / “Vote Leave” poster, Market Street, Omagh

• Vote Leave devised four core messages:
  
  • „Take back control“ was the official campaign slogan. The idea was based on the slogan „Keep the Pound, Keep Control“ from earlier work by Cummings against the adoption of the Euro. The „back“ in „Take back control“ triggers loss aversion and ties into the narrative of taking back control from Brussels and the system (big corporations, London, EU – those responsible for the crises).
    
  • „We send 350 million pounds a week to the EU – money we should use for the NHS“. Vote Leave deliberately used the gross figure and left out that the UK receives money from the EU as well in order to provoke a reaction from the Remain campaign – setting a trap for their opponents. They succeeded and the public attention was on the costs of the EU, which counteracted and reframed the debate about the economic risk of leaving the EU started by the remainers. The public knew that previous PMs had unsuccessfully tried to reduce the budget, which kept going up. They also knew that the budget was not always used sensibly and that Britain was spending more than it received: The message played to this knowledge. Finally, the message gave people a chance to vote for the NHS instead of only against the EU.
    
  • „Turkey will join the EU and 70 Million people can come to the UK“: This tied back to the concerns that many voters have about immigration and was used to made staying in appear riskier, which is an important message in its own right.
    
  • An Anti-establishment message: Vote Leave presented Brexit as the Anti-establishment option. This tied in nicely later with Boris Johnson’s populism and the results from the focus groups that trust in the establishment (big banks, London, politicians, EU) had been undermined.

• The core messages were neither left- nor right-wing. Cummings argues that the left-right division is an empirically false model that many politicans hold about how people think. The average voter is both more left-wing (agrees with confiscation of property) and more right-wing (supports harsh anti-crime measures) than politicians imagine. The core messages were also very simple and specific, which is necessary for communicating with a public that knows almost nothing about the immensely complicated EU. Lastly, the messages were all validated by focus groups, to ensure their effectiveness.

• A key factor according to Cummings was hiring the right people: At the beginning of the campaign he spent eight weeks biking around London hiring the right people: Physicists and data scientists. They surveyed empirical studies about turnout, persuasion, etc. taking into account their statistical soundness (replication, RCTs etc.) to find out what really works. Their communications strategy was based on those insights. From this survey they also learned the demarcation between hard knowledge and speculation – an incredibly valuable information to have. The physicists constructed models to decide where resources ought to be send for the ground and digital campaign. They used data input from canvassing, polling, the website, e-mail, the electoral roll, and social media and they constructed a central integrated database. They built a digital canvassing system from scratch. One of the main goals of the models using this database was finding out where the crucial one-fifth of the population lived. By sending experimental canvassers to streets selected by the models, they were able to learn from data by recalibrating the models using feedback from the canvassers. They ran experiments on facebook etc. using variations of the core messages to find out which version worked best. They used huge data sets and machine-learning (for example on facebook): this gave them the statistical power to say something about specific demographics via their polling and experiments. They used ingenious ways of collecting data – like the 50million.com challenge, which promised 50 million pound, the amount that the UK sends to the EU daily, to the person that managed to guess correctly the outcome of all matches of the European championship. This helped them to get in touch with a hard to reach demographic. The whole team also worked incredibly hard according to Cummings. He recommends „Hacking the Electorate“ by Eitan Hersh as an introduction to evidence-based campaigning.

• Cummings warns against overestimating the effect of techniques like microtargeting. More important according to him is finding out the answers to a few deep questions (e.g. What does really work in campaigning? What does the public really think about the EU? How can we find and persuade the people that we have a chance of convincing?).

„The most important thing for campaigns (and governments) to get right is how they make decisions.“

(Dominic Cummings)

• Answering a few deep questions and acting according to the answers requires focus and priorities, which required keeping the politicians out of the management of the campaign. They also led a very lean campaign, only 6-10 people managing it all in a Start-Up spirit. This helped them to avoid problems of big and traditional organisations including seniority, group thinking etc. Critical feedback was encouraged, which saved the campaign from several bad mistakes according to Cummings. The hard-nosed data-driven attitude of the physicist was helpful here. His advice is to study the organisation of high performance teams of the past (like the APOLLO program) to find principles of effective action, like the importance of a focused vision of the project that everyone supports and understands.

• They dumped almost all of their budget in the last days before the referendum. This strategy was based on evidence from the literature survey that marketing is most effective close to the decision. They aimed this marketing effort on a selection of roughly seven million people that received over a billion digital ads in total. They also sent most of their canvassers in those last days.

• They used „The Alternative Government“ to control the news cycle: When Boris Johnson – former mayor of London – and Michael Gove – Secretary of State for Justice under Cameron – decided to go all in campaigning for Brexit several weeks before the referendum, Vote Leave managed to make the story about the referendum into a story about the „alternative government“ of Johnson and Gove. Campaign messages were spun as „what the alternative government thinks“ – this proved irrestible for a media addicted to stories. The alternative government announced more money for the NHS and plans for an australian-style points system for immigration (an old favorite of focus groups). They managed to lead the news cycle for a 10-day period in a crucial period.

• Lastly, Cummings also talks about the mistakes of his opponents: Cameron did a bad job on the renegotiation of the relationship with the EU and came back without a new deal on immigration. Cummings also thinks that they ran a bad campaign that lived inside the bubble: After the murder of Labour politician Jo Cox one week before the referendum by a man with far-right sympathies, the Remain campaign changed their whole campaign from talking about economics risks into saying, in Cummings’ words, „we’re the good people and you’re the bad people“ – misjudging completely how many people really thought about the murder.

Five Lessons from the Vote Leave campaign:

• Evidence and data matter: Focus groups, surveying the literature on campaigning, building a centralised database for canvassing, testing messages on facebook and in focus groups, building models for canvassing and re-calibrating those based on feedback: All those point to the importance of evidence- and data-driven decisions. How to best run a campaign is about answering relevant questions (see point 4 below); experiments, evidence, data and learning are the means for this.

• People matter: The data-driven and evidence-based campaign would have been impossible without the skills and hard-nosed attitude of the physicists and data-scientists behind it. The organisation of the campaign also depended crucially on a number of highly capable managerial and political talents – not least Cummings himself.

• Organisation matters: As Cummings puts it: „The most important thing for campaigns (and governments) to get right is how they make decisions.“. The lean campaign, allowing critical feedback, having a focused vision, a small group of leaders, keeping the politicians out of management and being willing to make mistakes were all factors for success.

• Find answers to deep questions relevant to your goal and act accordingly: Suppose an organisation has a focused goal. What are the questions whose answers would most help an organisation in reaching its goal? What are the answers to these questions? It really makes sense that a lot of brainpower ought to be thrown at figuring out the answer to these two questions (and that the insights ought to be applied afterwards). In the case of Vote Leave some deep questions were: What does really work in political campaigning? What do the relevant groups of the public really think? How can one best convince those groups? How can one best reach those groups?

• Reframe the debate: Vote Leave did not only collect quantitative data, they also held focus groups. They used the insight that the public had lost trust in the EU and the establishment to devise simple and tangible core messages. These messages reframed the debate from „Leaving the EU is a huge economic risk“ to the genius „Take back control“. This made the referendum about (lost) sovereignty instead of economics. The 350 Million pound message, emphasizing the risks of immigration, the alternative government scheme, and talking about the NHS also worked brilliantly in this regard.

Sources

Dominic Cummings: „How the Brexit referendum was won“

Dominic Cummings: „effective action #2“

Dominic Cummings: „Why Leave won the Referendum“

Dominic Cummings: „On the referendum #20“

Paul Stephenson: „How to win a referendum“

Open Questions / Ideas for Future Blog Posts

1. How did the data science part of the campaign work in detail? How well did the models work?
2. What other insights besides timing the marketing right before the decision did they discover in their literature review? (see Hersh’s „Hacking the Electorate“ for a first impression)
3. How do high performance teams work in detail and how can they be sustained? (see the relevant posts on Cummings’ blog)
4. How sophisticated is political campaigning in Germany?