Research finds link between ‘human’ language and social brand uplift

The data analysis by Andrew Stephen, L’Oréal professor of marketing and Felipe Thomaz, associate professor of marketing at Saïd Business School, examined data from Facebook and Instagram advertising campaigns over the last two years.

The research focused on brand measures such as awareness and affinity.

The analysis of 235 global campaigns across 110 different brands used data from campaign effectiveness polls carried out by Kantar Millward Brown. The campaigns were 80% video, 20% display, and all mobile. Of the campaigns analysed, 49% ran on Facebook only, 48% were Facebook and Instagram and 3% appeared on Instagram only. Effectiveness was measured using control versus exposed methodology.

The analysis found a more significant impact on saliency (awareness) than brand association or motivation.

The researchers also applied machine learning and natural language processing to analyse the brands’ Facebook posts to determine what factors led to increased brand awareness. The study found that brands using ‘human’ or personable language, and avoiding functional words and phrases, tended to perform better on metrics such as saliency.

Saïd Business School also undertook a meta-analysis of Kantar Millward Brown’s broader digital effectiveness data, examining the impact of desktop compared to mobile advertising on brand metrics over the last seven years, finding the two are equally effective, leading to uplift of around 4% on brand metrics.

Kantar is a partner in the Future of Marketing Initiative with Saïd Business School at the University of Oxford. It was announced earlier this week that Facebook had joined the academic-industry initiative.