How One Pollster Correctly Predicted Brexit

July 1, 2016

Only one polling company correctly predicted the UK's Brexit outcome. Graeme Hutton reports.

As the world reverberates from Britain’s decision to leave the Europe Union, the shockwave making the vote all the more surprising was that no major polling company correctly predicted the outcome. Yet one new organization, Qriously, did correctly foresee what would happen.

Qriously used a different technique to the established market research organizations. Qriously’s secret was mobile. They harnessed the new advertising serving technology that is revolutionizing the digital advertising world by redeploying that technology to serve survey questions instead of ads to smartphone owners.

Critically the day before the Referendum, seven out of eight research agencies wrongly predicted that “Remain” (in the EU) would account for 51%-55% of the total vote. Bucking the trend, Qriously forecast just 44%. See their analysis here.

In defense of the seven research agencies who were wide of the mark, forecasting a one-off Referendum is fraught with issues:

  • Most pollsters compare historical survey results to previous, actual voting trends. By identifying the variations, they can weight subsequent survey results accordingly. This trend information was not available for Brexit.

  • For major issues related to a country’s sovereignty, most electorates are inherently conservative and opt for the status quo. This is exactly what happened last year when Scotland voted to remain part of the United Kingdom. The UK also decided in an earlier referendum to stay the EU in 1975.

Nevertheless, when Nate Silver, the preeminent US election forecaster, reviewed the major British polls leading up to the UK’s referendum, he ran a podcast entitled “British Referendum Polls Are A Mess”.

Having personally watched Qriously’s approach in action, I think one of the most refreshing aspects of Qriously is how their survey results build in real time. Displaying a map of the US on a computer screen, Qriously can pinpoint how individual consumers, scattered across the country, rapidly respond to a survey on a second-by-second basis. It makes for great theater and is, in part, a form of validation of their approach.

On a broader basis, Qriously’s success has some key learnings for marketers running consumer research:

  • Keep surveys and survey questions short. Qriously recommends just 45 characters per question.

  • Qriously also suggests using the largest sample size possible to help reduce any wobble in the results.

  • Rigorously interrogate the data via statistical techniques to ensure any variations stand up to scrutiny.

Graeme Hutton is SVP, Group Partner, Research, UM.He is a member of the industry Council for Research Excellence’s Social Media Group and 4A’s Measurement Task Force. His clients include J&J's consumer health business.