Much of advertising research is based on listening to what consumers say and then adapting campaigns accordingly. It seems a logical enough approach. However, it's based on the premise that what consumers say and what they do are aligned. Unfortunately there's a growing body of evidence that shows that the two things are often at odds.

Take sex. When heterosexual men and women are asked about the number of partners they have slept with the numbers vary dramatically. A 2013 Lancet study of 15,000 adults found that UK women admit to sleeping with an average of eight partners compared to twelve for men. The scale of the difference is not logically consistent. The most plausible explanation is that men feel a cultural pressure to exaggerate their exploits whilst women feel a corresponding pressure to play it down. Surveys, therefore, tell us more about what people feel they should say than the absolute truth.

To complicate matters further sometimes people simply don't know their genuine motivations. An experiment by Hargreaves at Leicester University exemplifies this. The experiment demonstrated the importance of background music on supermarket wine buying. Over a two week period they alternated between traditional German oompah music and French accordion music. When accordion music was played French wine accounted for 77% of wine sales whereas when oompah music could be heard German wine accounted for 73% of sales. When shoppers were asked about their purchase motivation only 2% spontaneously mentioned the impact of the music. Even when they were prompted 86% of people stated that the music had no impact at all. This experiment suggests that when we ask consumers to explain their behaviour they are liable to give overly rational explanations. After all, who wants to admit they're so fickle as to be influenced by background music?

Respondents can be influenced by far more than just background music. The very wording of questions can have marked effects. When we asked 300 consumers about their views on the bedroom tax five times as many thought it was a poor policy than a good policy. However, when it was re-labelled as a repeal of the spare room subsidy the number objecting to and agreeing with the policy were roughly equal. This suggests that in many circumstances consumers have "non-attitudes". Their views are not firmly fixed. However, they feel a pressure to be more definitive when questioned. Who wants to admit to being so ill-informed that they don't have an opinion? By relying on brand tracking we might believe consumer views are far more concrete than is the case.

The examples above are more than interesting anecdotes. Many brands rely heavily on claimed survey data. Recent research from ZenithOptimedia about the influence of sponsorship shows why this is problematic. Claimed data tends to show that consumers are not influenced by less rational media, such as sponsorship. Only 16% of the 1,000 consumers we talked to said sponsorship influenced their purchasing even a little. However, when we looked at their actual behaviour it was a different story.

We created a fake brand called "Black Sheep Vodka" which we told consumers was launching in the UK in 2015. We then served images and information about the product to two different groups of consumers. The only difference was that one of the groups also heard that Black Sheep Vodka was sponsoring a major TV programme. The results were stark. The group who heard about the sponsorship were 19% more likely to try the vodka and willing to pay 18% more than the control group. And the people who said that sponsorship had no influence at all? They were just as influenced by the sponsorship message.

These results show that brands need to avoid relying on claimed data alone. An overly literal reliance on these results could lead to brands making serious mistakes. The best practice approach is to corroborate claimed data with behavioural data. One simple method for doing this is the test and control methodology outlined above. Brands who adopt this approach can gain a significant advantage over their competitors.