How Influential Are Celebrities? (Updated)

AceMetrix-Effectiveness-TV-Ads-Featuring-Celebrities-Feb2014[This article, originally published in late 2013, has been updated with new research on the effectiveness of celebrities in advertising.] A new research paper [download page] from Ace Metrix analyzing the impact of celebrities in TV advertising has found that, well, they have little to none. Based on advertising data gathered from the beginning of January 2012 through October 2013, Ace Metrix found that in the aggregate, TV ads containing celebrities underperformed those without. There was a wide range in performance, however, leading the researchers to conclude that celebrity advertising is a “mixed bags” for the brands using them.

The dataset used consisted of more than 12,000 ads, of which more than 1,200 contained a celebrity. That was a significant upgrade in sample size from Ace Metrix’s previous study [download page] on the topic that was published in 2011 (and which came to the same conclusion regarding celebrities’ impact).

Celebrity ads underperformed in each of the measures Ace Metrix uses to construct its Ace Score, which is a measure designed to understand how a commercial performs from the twin angles of voluntary consumer consumption and the business goal of the advertiser. Celebrity ads underperformed the most in the elements of “desire” and “relevance;” indeed, the analysts note that the worst performers were those where the celebrities featured had little apparent connection to the brand (such as Jay-Z for Samsung Mobile Phones). By contrast, those that out-performed had a strong connection to the endorsed brand along with a “clever and integrated script.”

The researchers went beyond a simple comparison of Ace Scores, employing some nifty statistical techniques to isolate the celebrity impact by removing the impact of other variables such as the advertiser’s industry and the demographic campaign target. Having accounted for those, the study found that even for the demographic that gave the ad the highest score, the benefit of having a celebrity in an ad was just 3.7 points on Ace’s 950-point scale.

Which celebrity had the best performance in relation to the average? Ellen Degeneres, whose average Ace Score benefit of 59.3 was an outlier in the study. None of her ads landed in the top 10 celebrity ads by Ace Score, though: the leader on that end was Jordyn Wieber, appearing for AT&T Corporate Promotions in September 2012.

Below, some more research on celebrities’ public influence (or lack thereof):

Harris-Celebrities-Making-A-Difference-Nov201345% of US adults believe that celebrities can make either a large (11%) or some (33%) positive difference to issues they are promoting, but a greater proportion (51%) feel that they make little to no difference, per results from a study by Harris Interactive. Interestingly, respondents were more convinced of celebrities’ potential negative impact: 55% believe that celebrities’ negative publicity can have a somewhat (35%) or very (20%) damaging impact on the issue they’re promoting.

The study follows research suggesting that celebrities don’t have much influence over consumers when it comes to marketing campaigns. In a study of global consumers’ responses to advertising messages, Nielsen recently found that while humor resonates with a leading 47% of respondents, celebrity (12%) and athlete (8%) endorsements resonate with the fewest consumers. Another study – from Boston Consulting Group, reported by MediaPost – indicates that celebrity endorsements garner less trust from consumers around the world than any other type of brand promotion.

In an earlier study examining why Americans engage with online ads (clicking on, watching or paying attention to an ad), Ipsos revealed that just 1 in 10 respondents claimed to engage with ads because they like the spokesperson or celebrity, or because that person is someone they recognize. By comparison, 37% professed to have engaged with an online ad because it was for a brand that interested them. A more recent study commissioned by the Energy Saving Trust in the UK and conducted by Ipsos MORI, discovered that only 1% of respondents would be swayed to buy a product by a celebrity endorsement. By contrast, a leading 57% would have their purchase intent influenced by statistics or evidence to support the advertised claim.

Finally, while many celebrities have huge followings on social media, research from SocialToast suggests that its “Super Fans” (a collection of social media experts and professionals) are more likely to be influenced on important issues by posts from their close friends, family members and even well-known bloggers (arguably celebrities of some sort) than politicians and athletes.

Interestingly, while a fairly small 19% of respondents to the Harris Interactive survey said they have ever gotten more information or done anything to support a cause because of something they heard an actor, single or other celebrity say or do, that figure was up from 15% in 2008. Celebrity-influenced support for causes also tends to be higher among 18-36-year-olds (27%) than older Americans, with just 10% of Matures (68+) claiming to have supported a cause due to a celebrity’s actions.

Perhaps celebrities should champion their political views more? A surprising 58% of respondents believe that the support of a celebrity definitely (10%) or probably (48%) changes people’s views about which candidate to support, versus just one-quarter who believe that such support probably or definitely doesn’t have an impact.

Even with that, slightly more American adults feel that celebrities becoming more and more involved in politics is a bad (40%) rather than good (38%) thing.

About the Data: This Harris Poll was conducted online within the United States between September 18 to 24, 2013 among 2,577 adults (aged 18 and over). Figures for age, sex, race/ethnicity, education, region and household income were weighted where necessary to bring them into line with their actual proportions in the population. Propensity score weighting was also used to adjust for respondents’ propensity to be online.

7ads6x98y