Tsukuba, Japan—While research on social media advertising has largely focused on images, videos, and platform algorithms, the role of ad copy has received relatively little attention. This gap is especially evident in non-English contexts, such as Japan. In addition, the same word or expression may be interpreted differently depending on the advertised product.
To address this gap, the researchers analyzed a large dataset of Instagram advertisements in Japan between July 2021 and June 2023. The dataset included 12,206 supplement advertisements and 9,486 cosmetic advertisements. The text was analyzed using J-LIWC2015, a psycholinguistic dictionary designed for Japanese that categorizes words according to their psychological and semantic features to quantify linguistic patterns and examine their statistical relationship with click-through rate (CTR).
The analysis revealed that in supplement advertisements, words associated with risk or discrepancies were positively related to CTR. Meanwhile, in cosmetic advertisements, words related to visual perception or appearance, positive emotions, and motion were positively related to CTR. Notably, some linguistic features had differing effects depending on the product category. Motion-related language, for example, was associated with lower CTR in supplement advertisements but higher CTR in cosmetic advertisements. These results suggest that ad copy strategies should be tailored to the product category.
The findings provide large-scale empirical evidence that the relationship between the linguistic features of ad copy and CTR differs across product categories. The study thus offers practical insights to help marketers design more effective Instagram ad copy and contributes to research on how linguistic features are associated with user engagement in digital advertising environments, especially in non-English settings.
