We Got the Brexit Result Right

Leave was rarely challenged as the Brexit campaign’s superior digital force.

The Predata Brexit Daily with Aaron Timms

SEISMIC. That’s the word most are using to describe what we’ve just witnessed but if you’re looking for an alternative, volcanic, shattering, profound, historic, revolutionary, unprecedented, unfathomable or Gavrilo Principesque might do the trick. As the markets digest the shock of the British public’s decision to back Brexit, a simple reality is emerging. Most polls got the result wrong. The betting markets, as they stood on the morning of the election, got the result wrong. And the broad analyst class — the political risk consultants, academics, think tank dwellers, financial analysts and well-connected media commentators who together constitute “informed consensus,” and had almost unanimously coalesced in recent days around a Remain victory — got the result wrong.

Predata got the result right. To be more accurate, Predata got the electoral picture right. We did not outright predict a victory for Leave yesterday — indeed we’ve been careful, as regular readers of our Brexit coverage will know, not to make any prediction throughout the campaign, stressing instead that our monitoring of the digital campaign offered a unique perspective on the shape of public opinion in an era of notoriously unreliable polls. But our data sent an unequivocal message as the campaign concluded: the gap between the two sides had narrowed, but Leave continued to dominate the digital campaign — as it had done for much of the previous three months. Over a week ago we argued that if the pattern in our digital data held into this week, Leave would likely triumph. It did.

The Predata Brexit digital campaign score yesterday was 55.74% for Remain vs. 59.32% for Leave. As it happens, this spread accurately reflects the gap between the two sides in the final vote: 48% for Remain vs. 52% for Leave. The polls, on aggregate, had Leave ahead for little over a week of the entire campaign; as the chart above shows, our monitoring of the digital campaign had things completely the other way round. Leave was rarely challenged as the campaign’s superior digital force. (The Predata “scores” are not percentages in a pure sense but correlations of our Britain In and Britain Out signals to the overall measure of volatility and activity in the digital conversation about Brexit; they were never designed to sum to 100%.) Digital messaging is becoming an increasingly important component of political campaigning, especially in the developed world. Every election is different, but the Brexit experience suggests that the type of digital campaign monitoring Predata performs can not only offer insight into fluctuations in public interest and attention, but help us understand the likely shape of electoral outcomes.

Brexit will have profound implications for the UK, the City of London and Europe. Scotland and Northern Ireland, which both heavily backed remaining in the EU, will push for independence from the UK. London mayor Sadiq Khan will work — indeed, is already working — to keep London within the European single market. Marine Le Pen, favored to make it to the second round of next year’s French presidential election, will push for a referendum on Frexit, amid other movements throughout the continent for secession from the EU. The European project will come under serious strain. A fuller analysis of these repercussions will follow. For now, here is a quick rundown of what we got right and wrong about Brexit.


  1. Despite a late swing in “digital momentum” back to Remain in the final week, Leave controlled the conversation online and captured more of the public’s interest than Remain — right up to polling day. Our analysis showed this clearly.
  2. As we argued consistently throughout the campaign, Nigel Farage and Vote Leave’s core campaign issue — immigration — was an exceptional messaging success for the Out camp. Leave’s Dodgy Dave strategy was also highly effective and managed to turn the campaign into a referendum on the Prime Minister — a referendum which the PM was bound to lose.
  3. Jeremy Corbyn’s lack of commitment to the campaign — and its corollary, his invisibility as a digital presence — was fatal to Remain.
  4. We consistently anticipated swings in public opinion 3–5 days before they materialized in the polls.
  5. The precedent of the 2014 Scottish independence referendum suggested the “noisier” campaign — in this case, Leave — would not necessarily win. But Brexit was different to Scotland 2014 as a digital campaign beast, for the reasons we explained here. “The two sides are more closely matched, and momentum in the digital campaign shifts frequently in a way that renders a simple superimposition of the Scottish example onto current circumstances unhelpful,” we wrote on May 29. “Brexit is not like Scotland 2014, and the evenness and flux in the digital campaign invite the conclusion that the outcome on June 23 will be far from the comfortable victory for Remain that polls predict.” In this, we were correct.


  1. We over-emphasized the importance of the swing back to Remain in the campaign’s final stretch.
  2. We also over-emphasized the importance of economic concerns as a vote-determining factor in the campaign’s final stretch.
  3. Disunity among the Out camp — the sniping between Farage and Vote Leave, the official Leave campaign, in particular — did less to harm its chances than we thought it would.
  4. As drivers of online interest, immigration and Farage faded from the campaign in the final days, leading us to believe immigration was also losing ground as a vote-determining factor. In fact, the UKIP-sponsored message on immigration (BREAKING POINT) appears to have carried the day for Leave.
  5. The final week swing back to Remain, in both our data and the polls, led us to believe Brexit was perhaps following the pattern of past UK referendums — a pro-change scare two weeks out from polling day, followed by a late return to (and eventual victory for) the status quo. In fact, it was not.

The Predata Brexit digital campaign score is generated by computing the daily, month-long correlations of Predata’s Britain In and Britain Out signals to the overall Brexit signal (from today back to the February 20 announcement of the referendum date). The score is not a prediction or a probability level of Britain voting to stay in or leave the EU; it is a measure of each campaign’s correlation to online interest in Brexit as a whole, which can be used as a rough guide to which side is dominating, from day to day, the digital campaign. As such, the scores together do not need to sum to 100%.

For more information on our Brexit signals and analysis, contact Aaron Timms at aaron@predata.com.