Milk labelling and consumer opinion: a data-driven analysis from e-commerce reviews

Purpose This study analyses Spanish consumer reviews of pasteurised/UHT milk from Amazon to assess the alignment between consumer opinions and producer labelling efforts. It aims to bridge the attitude–behaviour gap in traditional market research by examining real post–purchase behaviour. Design/methodology/approach A total of 2,417 consumer reviews were collected through web scraping. Sentiment analysis was performed and an automated thematic analysis was carried out quantifying the presence of keywords related to health, sustainability, welfare, quality and emotion. Correlation analyses and generalised linear models examined the relationships between product labels, consumer ratings and review content. Findings The average rating was 4.29 stars with longer and more negative reviews correlating with lower ratings. Surprisingly, sustainability and animal welfare labels did not positively influence consumer ratings suggesting possible scepticism towards these claims. Organic-labelled products received higher mentions of sustainability, welfare and quality but also more negative sentiment, indicating that organic consumers may be more critical. Emotion-related keywords were linked to higher ratings, highlighting the importance of emotional engagement in consumer perceptions. Originality/value This study offers a novel approach by using actual consumer reviews rather than self-reported preferences to analyse purchasing behaviour. The findings challenge assumptions about the effectiveness of sustainability and welfare labels hopefully providing useful insights for producers and marketers.