The most common mental model of institutional food transition goes something like this: an institution decides to change. Leadership commits. Staff are trained. Menus are redesigned. Change happens at one site, and then, slowly, individually, at others.
This model is accurate for a significant share of the transitions in the PlantChapters dataset. But it is not the model behind the largest-scale outcomes we have documented. Those work differently.
One decision, hundreds of institutions
In March 2024, Sodexo Campus announced the simultaneous expansion of its DefaultVeg programme to nearly 400 university and college campuses across the United States. Every campus dining hall in the portfolio would have a dedicated plant-based station and an alternating station where plant-based meals were the standard offer unless a student actively requested otherwise. Approximately one million students would encounter plant-based meals as the norm rather than the exception at their dining hall from that point forward.
No university board voted on this. No student union passed a motion. No government regulation required it. A single commercial decision by a food service operator changed the menu architecture at 400 institutions in one announcement.
The Sodexo hospital expansion followed the same logic. Between 2022 and 2025, a plant-based default programme that originated at eleven New York City public hospitals expanded to 131 hospitals across the United States, with a target of reaching all approximately 400 hospitals under Sodexo management by the end of 2026. Sodexo serves approximately 290,000 patient meals per day across its US hospital network. The mechanism was the same: a single operator-level decision propagating through the contracted portfolio without requiring individual hospital approvals at each site.
Together, these two Sodexo programmes represent the largest documented scale outcomes in the PlantChapters dataset. And they share a structural explanation that is almost invisible in standard accounts of plant-based institutional transition.
The operator leverage point
Large contract catering companies occupy a structurally unusual position in institutional food systems. They hold catering contracts with dozens, hundreds, or thousands of institutions simultaneously. Their standard menu designs, recipe databases, procurement frameworks, and chef training systems operate across every client site in the portfolio. When an operator makes a decision to change its standard programme — to add a plant-based station, to reformulate its best-selling dishes, to adopt a default-switch methodology — that decision propagates across the entire contracted network through the same commercial infrastructure that delivers catering services every day.
This is what makes operator decisions structurally different from institutional decisions. An institution that decides to change its own menu affects one site. An operator that changes its standard menu design affects every institution in its portfolio simultaneously. The unit of change is the contract portfolio, not the individual institution.
The PlantChapters dataset documents this logic across multiple operators and institution types.
Eurest, the workplace catering division of Compass Group and the UK’s largest workplace caterer, reformulated the twelve best-selling meat dishes on its lunchtime menu — replacing 40% of the meat content in each with pulses and legumes while maintaining taste and presentation — and rolled that reformulation across hundreds of workplace cafeterias nationally. Sales held steady. Employees did not notice the change. The 47% plant-based menu composition Eurest reported in 2024 is the aggregate outcome of a single recipe change decision, not 47% of individual institutions each deciding independently to become more plant-forward.
Leijona Catering, the state-owned company exclusively responsible for feeding Finland’s conscript army, decided in 2018 to introduce two vegetarian meals per week across all garrison canteens. The decision was announced. A political controversy erupted. A nutrition study was commissioned. The study confirmed adequacy. The change launched across every garrison in Finland simultaneously. Approximately 70,000 meals per day shifted, without requiring approval from individual garrison commanders.
LEOC, a Japanese institutional catering company operating across more than 2,300 facilities, ran its 1,000 Vegan Project differently — driven not by a corporate mandate but by a single executive chef, Hitoshi Sugiura, who personally visited more than 1,000 canteens over 2020 and 2021 to cook plant-based meals and demonstrate recipes. By May 2021, 109,382 plant-based meals had been served through the project. The mechanism was personal rather than corporate, but the structural point holds: one person with authority over a large operator network reached 1,000 sites in the time it would have taken an advocacy campaign to persuade a fraction of that number individually.
Why the operator effect is largely invisible
Despite generating some of the largest-scale outcomes in the dataset, operator-led transitions receive less coverage in advocacy, media, and research accounts of plant-based food change than their scale warrants.
Several reasons explain this.
Operator decisions are often quiet. The Eurest reformulation, replacing 40% of meat content in best-selling dishes with pulses, was described internally as a recipe improvement rather than a sustainability campaign. Diners were not told. There was no press release announcing “Eurest removes meat from a million meals.” The invisibility was deliberate and was part of why it worked. Ingredient substitution at the recipe level is politically safer than publicly declared menu change, particularly in workplace settings where operators are sensitive to client reaction.
Operator transitions are commercially framed. Sodexo’s hospital expansion was announced as a sustainability initiative aligned with patient health outcomes and cost reduction — saving approximately $0.59 per meal by replacing expensive animal proteins with cheaper plant-based alternatives. The cost saving was not incidental; it was the enabling condition that made the operator decision internally defensible without requiring a separate sustainability mandate. That commercial framing makes the transition less legible as activism and therefore less visible in advocacy narratives.
Operator transitions bypass the visible decision points that generate coverage. A university vote, a council motion, a student union campaign — these are politically legible events with named individuals, documented decisions, and often media coverage. An operator reformulating its standard recipe database has no equivalent public moment. The change happens inside a commercial relationship and manifests quietly across hundreds of sites.
The implications for strategy
If the operator effect is real — and the dataset suggests it is — then it changes where leverage lives in institutional food transition strategy.
The standard model of change focuses on institutions: persuade enough hospitals, universities, schools, or councils to commit, and change accumulates one institution at a time. This is not wrong. Direct Institutional Implementation accounts for sixteen Chapters in the PlantChapters dataset, more than any other pathway. But it is linear. Each institution requires its own decision process, its own internal champion, its own implementation sequence.
The operator model is non-linear. Influencing a single operator decision reaches every institution in that operator’s portfolio simultaneously. For advocates, NGOs, and funders, the strategic implication is direct: a relationship with one food service company may be worth more than relationships with dozens of individual institutions, if that company is willing to act.
Greener by Default understood this. Rather than working hospital by hospital to introduce plant-based defaults, GBD worked with Sodexo to embed the default-switch methodology into Sodexo’s standard hospital programme. The result was 131 hospitals changed through a single commercial partnership, with 400 targeted by end of 2026. The same NGO-operator partnership logic underpins the Medirest / Compass Group trial in UK hospitals, now running across six sites with wider rollout expected if results confirm the pattern.
The contract renewal window is the key institutional leverage point for this model. An institution that holds a catering contract with a major operator has, at every contract renewal, an opportunity to embed plant-forward requirements into the new specification. Doing so converts an Operator-led pathway into a Procurement / Policy-led pathway — the institution specifies what it wants, and the operator delivers it across its contracted service. This is the DJI Dutch prison model: a national procurement framework that required contractors to shift menus toward more plant-based protein became the mechanism for changing 25 prisons simultaneously through a single contract award.
The constraints the dataset reveals
The operator effect is real, but it operates under structural constraints that are easy to underestimate.
The first is portfolio dependency. The Sodexo hospital programme changes hospitals where Sodexo holds the catering contract. It does not change hospitals catered by Compass, Aramark, Elior, or in-house food service teams. The scale of any single operator decision is bounded by the scale of that operator’s contract portfolio. In fragmented catering markets — many small regional operators, high prevalence of in-house catering — the operator effect is correspondingly weaker.
The second is the activation problem. The operator effect only activates when an operator decides to act. Operators respond to three types of pressure: corporate sustainability commitments (net-zero targets that identify food as a scope 3 emissions lever), client demand from institutions that specify plant-forward requirements, and evidence of cost-neutrality or commercial benefit. The Sodexo campus programme activated because a peer-reviewed RCT produced evidence that plant-based defaults increased selection rates and reduced costs without reducing satisfaction — evidence that gave the operator a defensible internal business case. Without that evidence, or without equivalent sustainability pressure or client demand, the same operator may not have moved.
The third is mechanism choice. Not all operator-led transitions use the same mechanism, and mechanism choice matters for outcomes. Sodexo’s hospital and campus programmes use Default Switch — plant-based as the standard offer, animal-based available on request. Eurest and UK MOD / ESS Compass Group use Ingredient Substitution — reformulating existing dishes invisibly. Leijona uses Plant-Based Day Model — designated vegetarian meals at specified service occasions. Each mechanism has different visibility, different acceptability profiles, and different replication requirements. The operator effect is not a single model; it is a family of related approaches sharing the structural logic of network propagation.
What is not yet in the dataset
The PlantChapters dataset currently documents operator-led transitions primarily from large multinational operators in high-income markets: Sodexo in the US, Compass/Eurest and Medirest in the UK, Leijona in Finland, LEOC in Japan. Regional and national operators at comparable scale in middle-income markets are almost entirely absent. India’s institutional food transition appears to operate through a CSR-framework model rather than operator networks. Brazil’s transitions are documented through NGO and regulatory models. The operator-led pathway may exist in those markets but is not yet visible in the data.
There are also no documented cases of an operator-led transition failing or being reversed. The dataset captures what worked. The conditions under which operators retreat from plant-based commitments — market pressure, client pushback, leadership changes, cost increases — are not yet documented. That gap matters for anyone trying to assess the durability of operator-led change rather than its scale.
The structural observation
The operator effect is not a theory. It is a structural feature of markets where institutional catering is dominated by large contracted operators with standardised systems. Where those conditions exist, one commercial decision changes hundreds of institutions simultaneously. The organisations best positioned to drive food system change at scale may not be the ones that are most visible in the advocacy landscape. They may be the ones that hold the most contracts.
This is the first 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.