Research shows that brief, factual SMS messages from fictional influencers can raise plant-based lunch choices among teens from 44% to 59%, with styling having little impact on effectiveness.
A short daily text message may be enough to nudge teenagers towards choosing more plant-based school lunches, according to research published in Appetite by Erasmus University Rotterdam. In a randomised controlled trial involving adolescents in the UK and the Republic of Ireland, plant-based lunch choices rose from 44% to 59% over two weeks among those who received messages, while the group that got no texts stayed at about 45%.
The study asked 587 adolescents aged 13 to 16 to join one of four groups: no messages, standard messages, linguistically tailored messages, or a crossover design that exposed participants to both versions. Each weekday morning, the messaging groups received a note from a fictional influencer called Sasha. The texts were deliberately brief and factual, with one evidence-based point about health, the environment or animal welfare, and they avoided moral pressure or judgement. The researchers counted any vegetarian or vegan lunch as plant-based, and the lunch choices were self-reported.
The most striking finding was what did not happen. Rewriting the same messages with a large language model to match each teenager’s own texting style did not improve the outcome. According to the paper and the summary released by Vegconomist, the tailored and standard messages produced almost identical results, suggesting that the content of the message mattered far more than the style in which it was written.
The authors said the result may reflect the low-friction nature of the intervention: short, factual messages may provoke less resistance from teenagers, leaving little room for stylistic personalisation to make a difference. Exploratory analysis suggested that by the second week, participants who received tailored texts were more likely to describe them as relevant and personalised, but the researchers described those findings as tentative. For caterers, brands and public health teams testing AI-driven personalisation, the study suggests that simple, evidence-based messaging can still be effective without algorithmic polish.
Disclaimer: This content is for informational purposes only and is not intended to be a substitute for professional medical judgment, advice, diagnosis, or treatment.
