Landmark political events were seen on either side of the Atlantic last year, with the EU referendum and the US presidential election. Results for both came as something of a surprise, not least to the pollsters who had – in both cases – predicted the opposite outcomes. The political polls failed to predict the right results due to one thing: poor data, writes Vanessa Tadier, general manager of Europe, Visual IQ. As Tadier explains, without accurate data, the pollsters never had a chance of understanding how the votes would stack up.
But political pollsters aren’t the only group who rely heavily on data to understand how the public will behave. Marketers also increasingly depend on data to understand where to invest their budgets and the likely impact of their marketing efforts. So, what can marketers learn from the political surprises of 2016 to ensure they don’t make the same mistakes?
Samples are not entirely representative
The fundamental issue with political polls is that instead of analysing the viewpoints of all registered voters, they rely on random samples, which can never be truly representative of the population. Even if the sample used is very large, and includes all segments of the population, there may be sub-segments or small pockets that are not represented, which can easily sway the results.
For marketers, survey data can be just as flawed as polling samples. Even when surveys are handled by highly skilled and experienced research agencies, it is difficult to create a sample that is truly representative of a brand’s target universe of consumers. Instead, advanced marketing measurement should be used to create a full data set and understand the entire target audience, not just a small sample.
Data is needed across convertors and non-convertors
In largely unprecedented circumstances, turnout for the EU referendum was difficult to predict. This made it virtually impossible for any sample to accurately represent the portion of the population who would actually cast their vote, as opposed to the entire population of registered voters. In the US, it is thought the exceptional turnout of white working-class voters in key swing states made a huge contribution to the Trump victory, which was not reflected in the polls.
When applying learning from this situation, marketers must make use of data from both converters and non-converters to gain a complete picture of their target audience. Just as excluding people who don’t usually vote from election forecasts resulted in inaccurate predictions for the pollsters, excluding consumers who were exposed to marketing messages, but did not convert from measurement practises, provides brands with an incomplete picture of marketing performance.
Real-time data is vital
Another issue with political polls is a lack of real-time data. Relying on historical information is problematic, as voters can easily be swayed by political pledges, events such as TV debates, or breaking news stories, and an answer given on one day can be meaningless days, or even hours, later.
In a similar vein, marketing surveys can take a long time to complete and analyse, making insights gained an historical snapshot in time that may already be outdated before it is used to inform marketing strategy. Marketers must adopt an advanced measurement approach that delivers real-time insights on a continuous basis, allowing campaigns to be constantly optimised using granular and in-the-moment data.
Survey responses can be skewed
Finally, there are multiple factors that can sway the way voters respond to polls, such as the way questions are phrased – which can impact a voter’s perspective – and the methods used to ask those questions. Analysis of Brexit polls revealed people who took part in phone polls were more likely to say they were voting to remain in the EU than online polls. This may be due to voters giving politically correct answers when speaking person-to-person, rather than being honest about how they are intending to vote in the perceived anonymity of an online survey.
Just like the polls, consumer perceptions and responses to marketing surveys can be altered by survey techniques and the wording of the questions. All too often the margin for error when using survey data results in wasted marketing spend.
Because marketers work in a similar way to pollsters – relying on survey data to understand consumer needs, measure marketing performance, and inform future campaigns – they must take heed of the events of 2016. To address the inaccuracies of survey data, marketers need to adopt an advanced mathematical measurement approach, which makes use of all available data, in real-time, without the potential for distortion, and is fully representative of their target audience. By using these advanced measurement techniques, marketers can be sure they are investing in the right tactics, audiences, and channels to achieve their desired outcomes.
If only driving the results of a political campaign were so simple.