Food marketing on social media is built around an intuitive assumption: make the meal look inviting, make the message feel warm and give people something emotionally appealing to share. A large field study of Facebook advertising suggests that this formula may be incomplete.
Across 22,475 promotional observations from quick-service restaurants and food-delivery services, posts containing stronger signals of competence consistently performed better when the outcome was sharing. Analytical language and cooler blue-green visual cues strengthened the relationship between food content and shares. Warmth cues, including emotional language and warmer colours, tended to weaken that relationship or remove it altogether.
The research, by Uchitha Wijesinghe, Frank Mathmann and Di Wang of Queensland University of Technology, appears in the International Journal of Hospitality Management. Rather than relying on a laboratory experiment or asking consumers what they might share, the researchers analysed behaviour around real promotional posts published by major food brands and delivery platforms.
More than 22,000 real-world posts and images
The project comprised two main field studies, each separating textual and visual content. The first focused on Facebook posts from the ten leading US quick-service restaurant brands. The second examined content from the three largest app-based food-delivery services.
For the quick-service restaurant analysis, the researchers collected 12,008 Facebook posts published between January 2009 and May 2024. A visual subsample included 5,000 photo posts. The delivery-service analysis contained 3,827 text posts and 1,640 photographs from the same broad period. Together, the four analyses produced a sample of 22,475 textual and visual observations.
The team did not classify posts manually according to whether they simply looked friendly or competent. Text was processed using LIWC-22, a linguistic-analysis system that identifies categories of words. Cognitive-process language associated with thinking, reasoning, causation and explanation was used as a competence signal, while affective language was used as a warmth signal. The researchers also measured the proportion of food-related words in each post.
Images were analysed computationally. Food objects were detected automatically, while dominant colours were extracted and grouped into warmer red, orange and yellow tones or cooler blue and green tones. The statistical models then tested whether competence and warmth changed the relationship between the amount of food content and the number of times a post was shared.
Analytical language strengthened the value of food cues
The first text analysis produced a clear interaction. Among the 12,008 quick-service restaurant posts, greater use of food words was positively associated with sharing. More importantly, competence-oriented language strengthened that association.
The interaction coefficient between food language and competence language was positive and statistically significant at b = 0.001891, p < 0.001. A Johnson-Neyman analysis showed that the positive relationship between food language and sharing became statistically significant once competence-related words made up more than 3.69% of the relevant language in a post.
Warmth moved in the opposite direction. Its interaction with food language was negative at b = -0.0008457, p < 0.001. Once warmth-related language exceeded 13.99%, the positive food-to-sharing relationship was no longer statistically significant. Above 30.54%, the estimated relationship became negative.
This does not mean emotional language is inherently bad marketing. The models address a more specific question: whether warmth or competence changes how strongly food cues translate into sharing. Within that setting, the competence signal was more consistently helpful.
The same pattern appeared in the images
The visual analysis of 5,000 restaurant photographs reached a similar conclusion. Blue and green colour cues associated with competence positively moderated the relationship between visible food content and shares, with an interaction coefficient of b = 0.0000762, p < 0.05.
The food effect became significant once competence-oriented colours exceeded 16.39% of the image. Warm colours showed a negative interaction at b = -0.00016, p < 0.001. A robustness test comparing the dominance of competence colours against warmth colours produced another significant positive interaction.
The importance of this result lies in convergence. Text and images were measured differently, yet both pointed in the same direction. Food content was more shareable when accompanied by cues that suggested competence, credibility or analytical processing rather than simply warmth.
Delivery brands showed the pattern again
The second study tested whether the result was peculiar to restaurant brands. It was not. Among 3,827 delivery-service posts, food language itself was positively related to sharing at b = 0.01872, p < 0.001, and competence language had a positive main effect at b = 0.017, p < 0.001.
Competence again strengthened the food-to-sharing relationship, with an interaction coefficient of b = 0.000829, p < 0.05. The warmth interaction in the text analysis was not statistically significant at p = 0.11, making this one of the places where the evidence for warmth was less clear.
The visual results nevertheless reproduced the broader pattern. In 1,640 delivery photographs, competence colours positively moderated the relationship between food imagery and sharing at b = 0.00015, p < 0.01. Warm colours had a significant negative interaction at b = -0.000099, p < 0.05. The food effect became significant when competence colours exceeded 13.24%, while high proportions of warm colour weakened it.
Why competence may matter when people decide to share
A like is a relatively private, low-effort response. Sharing a branded post is different because the user actively places that content in front of other people. The researchers argue that this makes credibility and evaluation more important. Before passing a promotional message to friends, consumers may look for cues that the content is worth endorsing.
That interpretation helps explain why competence cues can complement appetising food imagery. Analytical wording can signal explanation, knowledge or reasoning, while visual cues associated with competence may subtly change how the surrounding promotional content is processed.
For marketers, the result challenges the idea that food advertising should automatically maximise emotional language and warm colours. A post built only around appetite and friendliness may attract attention without giving users enough reason to redistribute it.
The practical implication is not that every restaurant should redesign its identity around blue and green. The study instead suggests that food cues work in combination with the rest of the message. Explanatory wording such as why a product is different, how something is made or what makes an offer useful may provide a competence layer that supports sharing.
What the study cannot establish
The scale of the dataset is a major strength, but the research remains observational. Brands did not randomly assign warmth and competence cues to otherwise identical posts. Other characteristics of campaigns, audiences and brands may therefore contribute to the observed sharing patterns.
The analysis also focuses on Facebook and on quick-service restaurants and delivery services in the United States. Consumer responses may differ on more visually driven platforms, in other countries or in categories where warmth is central to the brand relationship.
Colour should also be interpreted cautiously. Blue and green tones were operationalised as competence cues and red, orange and yellow as warmth cues within the researchers’ framework. Colour meanings can vary with culture, brand identity and context, so the results do not establish a universal psychological rule that cool colours always outperform warm ones.
Nor is sharing the only useful marketing outcome. A campaign designed to create affection, stimulate immediate purchase or reinforce a distinctive brand identity may reasonably optimise different cues. What this research isolates is the behaviour of passing content onward.
Food alone may not be enough to earn a share
The central finding is unusually consistent across text and imagery, restaurant brands and delivery services. Showing food mattered, but the surrounding signals changed how effectively that food content translated into sharing.
Across 22,475 observations, competence repeatedly strengthened the association while warmth tended to weaken it or make little difference. In a social feed crowded with polished food photography and emotional captions, credibility and explanation may therefore offer brands another way to earn the more demanding form of engagement: getting a consumer to put the message in front of someone else.
Source Information
Study Title: Competence cues drive engagement with food posts on Facebook: Convergent multimodal evidence from the field
Authors: Uchitha Wijesinghe, Frank Mathmann and Di Wang
Journal: International Journal of Hospitality Management
Volume: 140
Article: 104849
DOI: 10.1016/j.ijhm.2026.104849
Study design: Two observational field studies using text processing, automated image analysis and negative binomial regression
Total observations: 22,475 Facebook text posts and photo observations from quick-service restaurants and food-delivery services
