There is a persistent assumption running through most public discourse about plant-based food change: that the primary challenge is motivational. People eat animal products because they want to. They will eat fewer animal products when they want to do otherwise — when they are educated, persuaded, or inspired to change. On this view, institutional food change is fundamentally a communication problem. Provide the right information. Make the right argument. Change beliefs, and behaviour will follow.
The PlantChapters dataset tells a different story.
The highest-uptake mechanisms documented across 57 institutional transitions do not work primarily by changing what people want. They work by changing the structure of the choice — what is presented first, what requires effort to obtain, what is named and positioned as the standard. When those structural conditions change, behaviour changes too. Not because anyone was persuaded. Because the path of least resistance now runs through a plant-based meal.
The numbers that make the case
The dataset contains a natural experiment that illustrates this with unusual clarity.
In Singapore, two hospitals in the National University Health System introduced plant-based protein options through an opt-in model: patients could request a plant-based meal if they wanted one. Uptake was 5–15%.
In New York City, eleven public hospitals under NYC Health + Hospitals adopted a plant-based default model: plant-based meals were the Chef’s Recommendation, presented first to patients, available to every patient unless they actively requested a meat alternative. Approximately 50% of patients accepted the plant-based default — with 90% or higher satisfaction among those who did.
Same institution type — hospital patients, captive setting, no alternative food access, no ideological pre-selection. It is worth noting that Singapore already has one of the highest rates of vegetarian eating in Asia; the cultural conditions for plant-based uptake were arguably more favourable there than in most comparable contexts. If anything, this should have produced higher opt-in uptake, not lower. Opposite approaches to choice architecture. A twenty-fold difference in outcomes.
The difference was not culture. It was not demographics. No patient in New York City was given a more compelling environmental argument than any patient in Singapore. The difference was structural: which meal arrived unless you said otherwise.
What this means for mechanism selection
The PlantChapters dataset identifies sixteen distinct mechanisms through which institutional food transitions occur. Across the full dataset, the mechanisms with the highest documented uptake rates share a common structural feature: they reduce the friction associated with choosing plant-based food rather than asking diners to consciously choose it.
Default Switch is the clearest expression of this principle. When Sodexo Campus expanded DefaultVeg to approximately 400 US university dining halls, the intervention was not a campaign, a communication programme, or an educational initiative. It was a change to which entrée was presented first at designated stations. A randomised controlled trial at three universities — published in the Journal of Environmental Psychology — found that plant-based selection rates rose from 26.9% on control days to 57.6% on plant-default days. Student satisfaction showed no statistically significant decrease, including among regular meat-eaters. The mechanism shifted what more than half of students ate at those stations without shifting what any of them believed.
Ingredient Substitution operates by a different but structurally related logic. Eurest, the UK’s largest workplace caterer, reformulated the twelve best-selling meat-based dishes on its lunchtime menu — replacing 40% of the meat content in each with beans, chickpeas, and pulses. Same dish names. Same presentation. Same price. The reformulation was invisible: employees were not informed, and sales data confirmed no negative impact. The change reduced the animal protein content of meals that hundreds of thousands of employees eat without those employees making any choice at all. Institutional food systems were permanently changed not through the mechanism of persuasion but through the mechanism of recipe redesign.
The UK Ministry of Defence / ESS Compass Group case follows the same logic at the scale of 250 military establishments and 12.4 million meals annually. Over 70 plant-forward recipes were redesigned to reduce meat content by approximately 30% per dish while maintaining total protein — dish-level emissions reductions of 0.79–1.46 kg CO₂e per reformulated recipe were documented. In the first monitored month after launch, 38% of core meals sold were plant-forward and a further 7% were fully vegan or vegetarian. Soldiers did not choose to eat less meat. The food they were served contained less meat.
The identity problem that persuasion cannot solve
Understanding why structural approaches outperform persuasive ones requires understanding a specific barrier that persuasion cannot easily bypass: the identity barrier.
Choosing a plant-based meal in an institutional setting is, for many diners, an act with social meaning. It can signal dietary identity — vegan, vegetarian, health-conscious — that some people actively wish not to signal. In workplace cafeterias serving industrial and manual workers, university dining halls with mixed student populations, or military mess halls where meat-eating norms are culturally significant, the social cost of visibly choosing a plant-based option can be higher than the individual’s actual preference for the food.
The FDF Award that recognised Eurest’s programme specifically cited its success with meat-eaters at industrial sites — historically the hardest group to reach through plant-based promotion. This was not despite the invisible reformulation model; it was because of it. The invisible reformulation model bypassed the identity barrier entirely. Employees who would resist choosing a plant-based dish were served a reformulated dish they already liked, without being asked to make a visible dietary choice.
Default Switch addresses the identity barrier through a different route. When plant-based meals are the default — the Chef’s Recommendation, the standard offer — choosing them carries no social signal at all. It is the default. The social cost has been neutralised by normalisation. The opt-out, by contrast, now carries its own signal: you are the one who asked for something different. Most people do not ask for something different.
This is why approximately 50% of NYC hospital patients accepted plant-based defaults — a figure that rises to 90%+ satisfaction among those who chose it. It is not that 50% of those patients would have chosen plant-based meals from an opt-in menu. The Singapore data suggests far fewer would have. It is that half of all patients accepted the meal that arrived unless they had a specific reason to change it — and most did not have a specific reason.
The persuasion layer still matters
None of this means that communication is irrelevant to institutional food transition. The dataset shows that framing, language, and narrative matter significantly — but at the level of enabling institutional decision-makers to act, rather than persuading individual diners to choose differently.
At NYC Health + Hospitals, the framing of plant-based meals as a Chef’s Recommendation — rather than as a vegan option, a plant-based default, or a meat-free alternative — was a deliberate design decision that neutralised the most common institutional resistance to default models: the perception that a plant-based default removes patient choice. Framing it as a recommendation preserved the cultural intelligibility of the hospital food system. A recommendation is something a chef makes; it can be followed or not. The choice framing remained intact even as the structural default changed.
At the Sodexo campus pilot, academic publication of the RCT results in the Journal of Environmental Psychology was the communication act that mattered most — not to students, but to Sodexo’s internal decision-makers who needed externally validated evidence to justify a system-wide rollout. The persuasion target was not the diner; it was the operator, and the communication form was a peer-reviewed paper, not a campaign.
In the Green Tuesday Initiative across India, the communication framing that unlocked institutional access was not animal welfare messaging but climate action and ESG performance — aligned with the mandatory corporate social responsibility framework under which Indian companies report. The change in what was served in cafeterias resulted from a change in how the decision was framed to corporate sustainability teams. Diners were not persuaded; the institutional rationale was.
This points to a more precise account of where persuasion operates in institutional food transition. It operates at the decision-maker level, before implementation, shaping whether an institution decides to make a structural change at all. Once that decision is made, the structural mechanism — the default, the reformulation, the menu ratio — does the work. Persuasion opens the door. Structure moves the meals.
The mechanisms that require persuasion
Not all mechanisms in the dataset operate through friction reduction. Some require diners to consciously choose differently, and the dataset is informative about when those mechanisms succeed and when they struggle.
Menu Ratio Shift — increasing the proportion of plant-based dishes available — is the most common primary mechanism in the dataset, appearing in fifteen Chapters. But it is also the mechanism whose outcomes are most dependent on whether the new plant-based options are genuinely attractive to the existing diner population. Adding more plant-based dishes to a menu increases the possibility of plant-based choice; it does not change what most people select unless those dishes are competitive in quality, familiarity, and availability with the meat alternatives. The University of British Columbia achieved 80% plant-based meals through a combination of Menu Ratio Shift and deliberate menu positioning — but UBC also invested heavily in culinary quality, cultural diversity in dish design, and a campus sustainability culture that made plant-based choice socially normalised rather than identity-marked.
The lesson is not that Menu Ratio Shift doesn’t work. It is that it works best when accompanied by friction-reducing design — positioning, naming, and cultural normalisation — rather than relying solely on the availability of better options.
Information Architecture — eco-labelling, carbon scores, environmental impact labels — is a mechanism that works explicitly through awareness rather than structural change. Eurest’s partnership with the University of Oxford to implement A–E environmental ratings across its entire recipe portfolio is the clearest dataset example. The evidence for eco-labelling producing meaningful behaviour change at population level is mixed in the academic literature, and the Eurest case does not isolate eco-labelling’s contribution from the simultaneous invisible reformulation. What is documented is that eco-labelling works on a specific segment — environmentally motivated diners who respond to information — while having limited effect on those for whom environmental impact is not a primary decision factor.
A reframe for practitioners
The friction principle, as the dataset suggests it, can be stated simply: structural changes to what is available, what is default, and what requires effort to obtain produce more reliable outcomes than persuasion campaigns directed at individual diner choice.
This reframe has practical implications for institutions, operators, NGOs, and funders considering how to allocate effort and resource in institutional food transition.
For institutions: the design of the choice environment — which meal is presented first, how opt-out processes work, how plant-based dishes are named and positioned — matters more to uptake outcomes than the volume of communication about plant-based eating.
For operators: invisible reformulation of existing dishes is often more effective at scale than visible introduction of new plant-based products. The Eurest model reached more employees with less resistance than any promotional campaign could have achieved, because it did not ask employees to make a different choice.
For NGOs and advocates: the most strategically valuable communication acts may not be directed at the general public at all. The persuasion work that produces structural change happens at the decision-maker level — with food service directors, hospital administrators, university caterers, and government procurement teams — before any diner is involved.
For funders: the return on investment from supporting structural change — default design, recipe reformulation, procurement specification — may be substantially higher than equivalent investment in awareness campaigns, precisely because structural changes are persistent while campaigns are episodic.
What the dataset does not yet show
The case for structural approaches over persuasion is strong in the PlantChapters dataset, but the dataset has a characteristic gap: it documents transitions that worked. The conditions under which friction-reduction mechanisms fail — poorly designed defaults that generate backlash, invisible reformulations discovered and resented, procurement changes reversed under cost pressure — are largely undocumented. The failure modes of structural approaches may be different from the failure modes of persuasion campaigns, but they exist, and a fuller understanding of them would strengthen the practical guidance available to institutions.
The dataset also has limited coverage of institutional contexts where cultural resistance to plant-based food is exceptionally high — highly meat-centric cultures, strongly traditional institutional food systems, contexts where the identity cost of plant-based food is particularly salient. The friction principle may be more robust across cultures than persuasion-based approaches, but the dataset does not yet provide sufficient coverage of high-resistance contexts to test that claim rigorously.
The structural observation
Institutional food transition does not primarily require changing what people want. It requires changing what people encounter. The two are different problems, with different solutions. Most of the highest-scale documented transitions in the PlantChapters dataset changed what people encountered. The outcomes followed.
This is the second in PlantChapters’ Patterns series — cross-dataset structural observations that emerge from analysing transitions collectively rather than individually. The dataset behind this analysis is available at plantchapters.org.