Assessing the sustainability performance of large food retail companies: a generic framework and pilot application to the environmental dimension in Germany

1 Introduction

Food systems are intrinsically linked to nutrition and human health because they determine the availability, accessibility, and quality of foods that provide essential nutrients necessary to sustain human life and wellbeing. At the same time, the food sector exerts wide-ranging effects on the environment, social structures, and economic activity (FAO, 2013; Crippa et al., 2021). In Germany, the food industry is economically important, employing more than 5.1 million people (11.4% of the workforce) and generating EUR 188.5 billion in value added, equivalent to 6.2% of national economic output (Lebensmittelverband Deutschland, 2023).

From an environmental perspective, seven of the nine planetary boundaries have already been exceeded, and food production and consumption are closely linked to these pressures (Richardson et al., 2023; Rockström et al., 2025). Primary production is particularly relevant because of its use of land, water, and nutrients and its associated impacts on climate, soil, water, air, flora, and fauna (Poore and Nemecek, 2018; Crippa et al., 2021). At the consumption level, food accounts for 1.57 t CO₂-eq. (15.1%) of the average annual 10.41 t CO2-eq. greenhouse-gas footprint of a person living in Germany, approximately the same magnitude as mobility excluding air travel (1.62 t CO2-eq.) (Federal Environment Agency, 2025).

Regarding the often intertwined social and economic dimensions of sustainability, the food sector is characterized by precarious working conditions in agriculture and downstream sectors, both domestically and internationally (Benoit-Norris et al., 2012). According to FAO (2013), parts of the food sector generate relatively low added value and wage levels compared with other sectors, while food production remains a primary source of income in many countries in the Global South (ILOSTAT, 2025; World Bank Group, 2025). German households devote a comparatively low share of expenditure to food (11.5% compared with an EU average of 13.6%). At the same time, around 3 million people in Germany experience food poverty, while approximately 76 kg of food per person per year are discarded, around two-thirds of which are considered avoidable (Destatis, 2022; Eurostat, 2022).

Beyond environmental, social and economic dimensions, sustainability assessments of food systems frequently consider consumer health and animal welfare (FAO, 2013). These dimensions are linked through the One Health approach, which conceptualizes human, animal and ecosystem health as interdependent and calls for coordinated action across sectors. Food retailers are relevant to this nexus because procurement standards, product portfolios, pricing, marketing and consumer information can simultaneously influence production practices, animal welfare, dietary patterns and environmental pressures (One Health Initiative, 2020).

Food systems are also central to achieving UN Sustainable Development Goal 12 (“Responsible Consumption and Production”), yet current production and consumption patterns remain incompatible with multiple environmental and health objectives (Romanello et al., 2024). The EAT-Lancet Commission proposed the Planetary Health Diet as a global reference pattern rich in plant-based foods and comparatively limited in red meat and added sugars, designed to support human health while remaining within planetary boundaries (Willett et al., 2019). Its 2025 update reinforces the need for food-system transformation while recognizing that transition pathways must account for regional dietary patterns, affordability, livelihoods and food-system conditions (Rockström et al., 2025). Evidence on sustainable dietary change likewise shows substantial mitigation potential but also highlights context-specific implementation challenges (Jarmul et al., 2020). Because production, retail and consumption are interconnected, such a transition requires action by all major food-system actors.

In economic terms, many environmental and social externalities of food production are not fully reflected in market prices. Costs associated with water pollution, greenhouse-gas emissions, biodiversity loss or land-use change can therefore be shifted to society or future generations rather than being incorporated into the decisions of producers, processors, retailers and consumers. This weakens market incentives for more sustainable production and consumption.

Policy instruments that directly alter prices or restrict particularly damaging practices can face limited public acceptability and political feasibility (Richter et al., 2023). Consequently, benchmarking and disclosure are often considered complementary instruments within broader policy mixes (Gunningham and Grabosky, 1998). Existing sustainability assessments cover farms, dietary patterns and food manufacturers (Vrolijk and Poppe, 2021; Jarmul et al., 2020; Küchler et al., 2023, 2024). For food companies, the BIA-Sustainability approach provides a sector-specific set of environmental indicators (Mackay et al., 2022) and has been pilot-tested in Australia (Serrentino et al., 2024); NGO-led approaches such as WWF Switzerland (2019) and Superlist (Haan et al., 2023) also benchmark retailer practices. However, these approaches differ markedly in sectoral scope, data sources, sustainability dimensions and the extent to which they capture retailers' influence over suppliers, own operations, consumers and policy. This heterogeneity leaves a need for a retailer-specific framework that makes these design choices explicit and can be adapted transparently to different assessment objectives.

In this article, “food retailers” refers to supermarket and discount grocery companies whose primary role is selling food for household consumption; restaurants, cafés, and other out-of-home food-service businesses are not included. Large food retail companies play a particularly important role as gatekeepers within food systems because they operate at the interface between production and consumption. According to a sector inquiry by the German Federal Cartel Office (Bundeskartellamt, 2014), food retailers account for between 65% and 90% of manufacturers' total sales, depending on the procurement market, and around three-quarters of consumers' monthly household expenditure on food is spent in food retail outlets. Retailers can therefore influence sustainable consumption through assortment and pricing, as well as upstream production through procurement requirements. This influence is reinforced by high market concentration: in Germany, the eight food retailers with the highest revenues, belonging to four corporations, generated more than 75% of total food retail revenues in 2022. Vertical integration and the expansion of private brands further strengthen retailers' direct influence on product specifications and supply chains (Jankowski, 2019; Grabs and Carodenuto, 2021).

Food retailers can use their market position to influence food-system transformation through four main spheres of action (Figure 1). First, they can reduce impacts within their own organizations through measures such as building insulation, refrigeration technology, logistics, waste management, staff training, and occupational safety. Second, they can influence upstream value and supply chains through procurement requirements, product specifications, and third-party standards such as organic and Fairtrade certification. Third, they can shape consumer choices through assortment, store and shelf design, advertising, and pricing (Keller et al., 2022). Fourth, as major economic actors, retailers can interact with and influence the policy framework within which the sector operates.

These four spheres of influence represent the fundamental and universal points of contact for large food retail companies, which can be leveraged regardless of the specific sustainability issue – whether it be environmental protection, social responsibility, animal welfare, or health.

Against this background, the study addresses two linked questions: what leverage do large food retailers have to improve food-system sustainability, and how can the extent to which they use this leverage be assessed comparably at company level?

General sustainability frameworks such as life-cycle assessment (ISO, 2006) and the SAFA Guidelines (FAO, 2013), as well as food-company benchmarking approaches, provide important methodological building blocks but do not by themselves capture the full combination of retailer-specific spheres of influence, action-oriented management levers and cross-company comparability required here (Table 1). We therefore developed a generic framework that explicitly structures assessment design around retailers' influence from production through own operations to consumption.

Table 1. Comparison of advantages and limitations of sustainability assessment frameworks for large food retail companies.

The study had three linked objectives:

• To develop a generic framework for comparative sustainability assessment of large food retail companies (Section 2).
• To operationalise the framework for environmental sustainability assessment of large food retailers in Germany (Section 3).
• To apply the resulting methodology to the eight German food retailers with the highest revenues and discuss the findings, limitations and transferability (Section 4).

Sections 3 and 4 draw on an empirical research project commissioned by the German Federal Environment Agency (UBA) and documented in Keller et al. (2022). The technical report provides the full operational details of the German assessment. Building on that project, this article focuses on the generic framework, its operationalisation in the pilot, and the methodological implications for future company-level assessments. Detailed indicator definitions, scoring scales, and weighting factors are publicly documented in Keller et al. (2022).

2 Conceptual framework for assessing sustainability performance of food retail companies

Existing approaches to food-retail sustainability assessment vary substantially in scope, indicator design, and intended use (WWF Switzerland, 2019; Mackay et al., 2022; Haan et al., 2023; Serrentino et al., 2024). The conceptual framework presented here specifies the design decisions required before a retailer sustainability assessment is implemented. It draws on established methodological guidelines and frameworks (Dodgson et al., 2001; ISO, 2006; FAO, 2013) and on previous applications of sustainability assessment in agriculture, food, and nutrition (Schader et al., 2016, 2019; Ssebunya et al., 2018; Curran et al., 2020; Winter et al., 2020; Küchler and Herzig, 2021; Keller et al., 2022; Küchler et al., 2023, 2024; Sander et al., 2025). The components follow an ideal sequence but are intended to be applied iteratively, consistent with the iterative logic of life-cycle assessment.

2.1 Determination of objectives and scope of investigation

2.1.1 Objectives

Explicit definition of the assessment objectives is required before the method is developed. The following objectives are not mutually exclusive:

• Advising individual companies to achieve internal improvements.
• Comparative ranking of food retail companies to create transparency for the public and to incentivise companies to develop further.
• Company monitoring to examine developments over several years.
• Examination of internal and external influencing factors (e.g., political measures) and their effects on company development.

These objectives can be specified further according to the assessment context. The selected objective directly influences the design and implementation of the assessment.

2.1.2 Target groups

The assessment design also requires specification of its intended users. These may include internal company stakeholders, such as management and sustainability departments, as well as external stakeholders, including public authorities, policymakers, consumers, and researchers. The geographic scope of the target audience, for example national or international, likewise forms part of this specification.

2.1.3 Selection of companies

A central design choice concerns the companies included in the analysis. Where several companies are compared, transparent selection criteria are required. These may include, for example, company size, corporate structures or geographical scope of activity.

2.1.4 Companies' spheres of influence

According to Figure 1, the spheres of influence of companies can be divided into the following areas:

• The companies themselves (i.e., branches but also their own production facilities),
• Influence on suppliers and sub-suppliers,
• Influence on consumers,
• Influence on the political framework.

The spheres of influence to be included in the analysis must also be clearly defined.

2.1.5 Product range

Large food retail companies typically offer broad product portfolios that may include food, non-food, and near-food products. The assessment therefore requires an explicit definition of the product segments included. Within food assortments, a distinction is also required between private-label and third-party brands. Because retailers often have more complete data for private-label products, the inclusion of third-party brands needs to be specified for the overall assessment or for individual fields of action.

2.1.6 Sustainability aspects

As explained in Section 1, the decisions made by large food retail companies affect various aspects of sustainability. A clear classification of topics into these aspects is recommended. However, specific topics from individual aspects may also be selected:

• Environment
• Society
• Economy
• Animal welfare
• Consumer health

2.1.7 Reference period

Each assessment requires a clearly defined reference period. Its selection needs to take account of company reporting periods so that the most up-to-date data possible can be used. Several years may also be specified where development over time is the subject of the investigation.

2.1.8 Level of comparison

Defining the level of comparison in advance is an essential design step. Assessments may compare companies at the level of individual indicators, broader topics or spheres of influence, or overall sustainability performance. Any comparison above the indicator level requires explicit weighting and aggregation methods.

2.2 Definition of the evaluation methodology

2.2.1 Use and adaptation of existing methods

Given the large number of sustainability assessment methods and indicators, existing systems provide the preferred basis where they are suitable. This can reduce methodological effort and improve comparability across studies. Departures from established methods may be warranted by specific research questions but require explicit justification.

2.2.2 Definition of fields of action

As discussed by Keller et al. (2022), purely outcome-oriented sustainability assessment is difficult to apply comprehensively to food retailers. In addition to limited data availability, outcome-oriented approaches raise conceptual and methodological challenges. For environmental assessment, these include the choice of functional units, comparability across model assumptions, and allocation methods. Allocation is particularly relevant in food systems because agricultural processes frequently generate multiple co-products, outputs, and functions. For other sustainability dimensions, suitable outcome indicators are often unavailable or insufficient to represent the full consequences of food-retail activities.

The framework therefore adopts an action-oriented assessment structure. It is organized first by the fundamental spheres of influence and then by the corresponding options for action, which can be grouped into strategy, implementation, and communication. A comprehensive company assessment requires fields of action that cover the relevant levers available to retailers while remaining mutually exclusive. Minimizing overlap between fields of action reduces the risk of double counting, particularly when results are aggregated across fields.

In addition to a title, each field-of-action description includes a precise definition of its objectives and a derivation or justification based on scientific findings.

2.2.3 Definition of indicators

The indicator set is intended to capture, as comprehensively as practicable, the extent to which a company uses its scope for action to improve food-system sustainability. Whereas fields of action can be structured independently of a specific sustainability topic, indicator formulation is specific to the relevant topic area. Indicators may be action-oriented or result-oriented in accordance with the DPSIR framework (Spangenberg et al., 2009). When results are aggregated at the field-of-action level, minimizing overlap among indicators helps prevent double counting.

A complete indicator specification includes a title, a defined target, an assessment scale with sufficiently precise descriptors for each scale level, and a methodological specification that supports consistent cross-company comparison. Complex indicators may be subdivided into sub-indicators to improve consistency in data collection and subsequent assessment.

2.2.4 Aggregation method

If there are several indicators per action and/or topic area, an aggregation method must be defined. A basic distinction can be made between additive and multiplicative methods (Gan et al., 2017). Additive methods form an arithmetic mean of the indicator results, while multiplicative methods form a geometric mean. Depending on the chosen aggregation method, the overall assessments can vary significantly. Multiplicative methods are characterized by the fact that individual weaknesses carry greater weight in the overall assessment. For fields of action in which the importance of one indicator depends on the assessment of others, multiplicative aggregation is appropriate. The aggregation method may therefore be defined specifically for each field of action.

For example, in a field of action where company objectives are assessed according to measurability and level of ambition, multiplicative aggregation is appropriate because even an ambitious goal is ineffective if it is not measurable (e.g., because a time frame is missing). If, on the other hand, the proportions of sustainable products in different product assortments are aggregated using appropriate indicators, an additive method is preferable because the product ranges are independent of one another.

2.2.5 Weighting

In addition to the aggregation method, the weighting method must also be defined. If not all indicators are of equal importance for the field of action, weighting is advantageous. Where possible, empirical evidence provides the preferred basis for weighting. For example, sales figures can provide a basis for weighting indicators for individual product ranges.

If there is no empirical basis, expert-based methods can be used to determine weightings. There are a variety of methods with different strengths and weaknesses. For example, the Analytical Hierarchy Process (Handfield et al., 2002) and the Best-Worst Method (Rezaei, 2015) primarily ensure that the evaluations are consistent and coherent. In contrast, the Delphi approach (Delbecq et al., 1975; Becker, 1995; Ziglio, 1996; Hsu and Sandford, 2007; Mukherjee et al., 2015; Küchler et al., 2023) and the Nominal Group Technique (Delbecq et al., 1975) focus primarily on the structured exchange of arguments that are relevant for weighting. For this reason, it may also be useful to combine the different methods.

2.2.6 Review of the methodology by experts and stakeholders

To ensure the relevance and validity of the methodology, it is essential to coordinate the key steps with experts and stakeholders prior to implementation. It is advisable to inform the experts and stakeholders about the project as early as possible and to involve them in the development of the methodology.

2.3 Data collection

2.3.1 Questionnaire development

A questionnaire can be used to collect structured information for the relevant components of the assessment grid. Where practical, its organization can reflect the fields of action and indicators while allowing adaptations to company structures and internal responsibilities. This may enable different sections to be completed by appropriate functions, such as marketing or procurement. Clear definitions of technical terms and, where possible, alignment with established frameworks reduce ambiguity and improve data quality.

Coordination of the questionnaire with participating companies helps ensure that the required data are available or can be made accessible while minimizing the data-collection burden. A pre-test can additionally identify errors before data collection begins.

2.3.2 Data collection and verification

A practical data-collection sequence begins with publicly available secondary sources, particularly sustainability reports (Herzig and Schaltegger, 2011). The information collected can then be entered into the questionnaire and submitted to companies for correction or supplementation. Company-provided data subsequently undergo checks for completeness and plausibility. Precise specification of data requirements is important, and quantitative indicators require an explicit calculation method.

To minimize the workload of data review, it is advisable to limit the scope of the responses. Otherwise, companies tend to provide as much information as possible, only some of which is relevant. Due to the complexity of the data, an iterative data collection process is generally advisable. This means that data collection occurs in several steps, allowing companies to provide data at multiple defined points in time.

2.4 Evaluation

2.4.1 Subdivision of the assessment into sub-indicators

As mentioned above, some indicators may be difficult to capture comprehensively. In such cases, it is advisable to subdivide the indicator into sub-indicators. These sub-indicators are then evaluated independently and aggregated. The aggregation can be carried out either by pragmatic heuristic aggregation or formalized according to a defined aggregation and weighting methodology.

2.4.2 Independent evaluation by several people

Because the information provided often requires qualitative classification, evaluator judgement may influence ratings. Independent assessment by more than one evaluator reduces the risk of individual bias and improves comparability across companies.

Where evaluations differ, evaluators compare the reasoning underlying the alternative assessments. Systematic documentation and, where necessary, discussion support this process until a consensus assessment is reached.

2.4.3 Sensitivity and uncertainty analysis

Where relevant, sensitivity and uncertainty analyses can be used to test the robustness of the assessment to key assumptions. Sources of uncertainty may include indicator data, weighting choices, and aggregation methods.

2.4.4 Verification of the assessment by the respective company

Once the project team has reached consensus on indicator-level performance, discussion of the preliminary results with the companies concerned provides an additional verification step. Returning the assessments to companies for review can help identify errors, misunderstandings, or missing evidence before results are finalized. This verification step is intended to make the final assessments transparent and comprehensible to the companies concerned.

3 Pilot application: methods for assessing the environmental sustainability of German food retailers

The generic framework was operationalised in a pilot assessment of the environmental sustainability performance of the eight food retail companies with the highest revenues in Germany. The procedures used to operationalise the framework are described below; further methodological detail is provided in Keller et al. (2022).

3.1 Determination of objectives and scope of investigation

The study aimed to develop and apply an assessment tool for the comprehensive and comparative evaluation of the environmental sustainability performance of large food retail companies in Germany, thereby increasing public transparency and providing incentives for further improvement.

The target audiences included the participating companies and their stakeholders, such as business partners, consumers, and relevant associations, as well as public authorities and policymakers.

Company selection: Companies were selected according to their share of sales in the German food market. Based on a previous study (Seo, 2020), the eight food retail companies with the highest sales in Germany were included. Because of EDEKA's decentralized corporate structure and the associated data-collection effort, EDEKA Südwest was selected to represent the group.

Sphere of influence of companies: In addition to the companies themselves, their influence on supply chains and consumer decisions was also examined. Their influence on the political framework was not considered, as this would have required a different data basis, the collection of which was not planned within the project's scope.

Product range: Non-food and near-food products were excluded from the assessment. In addition, due to reservations on the part of large food retail companies regarding data collection, only private brands were considered in the supply chain-related fields of action. In the other spheres of influence, third-party brands were also included.

Sustainability areas: The pilot assessment was limited to the environmental dimension, which was subdivided into climate, soil, biodiversity (including deforestation), and oceans.

Reference period: The year 2020 was selected as the most recent calendar year for which complete data were available at the start of the project.

Comparison level: The study compared the companies across individual fields of action. Therefore, the results were not aggregated beyond the field of action level.

3.2 Definition of the evaluation methodology

Literature review: At the start of the project, gray and scientific literature on sustainability assessment in the food sector was reviewed. The review included the World Benchmarking Alliance framework (WBA, 2020), an early working version of WWF Germany's One Planet Business Framework, subsequently published by Metzke and Strahl (2024), and the WWF Switzerland assessment grid for wholesale and food retail (WWF Switzerland, 2019). Relevant international and national frameworks and objectives, including the Planetary Boundaries, SDGs, SAFA Guidelines, the German Federal Government's Sustainability Strategy, and the National Programme for Sustainable Consumption, were also considered when developing the assessment grid.

Definition of fields of action: The fields of action were structured according to retailers' spheres of influence. Within the “supply chain” and “own sites” spheres, fields of action were further grouped into strategy, implementation, and communication; this subdivision was not applied to the “consumption” sphere. The fields of action were derived from the literature review, particularly WBA (2020), WWF Switzerland (2019), and Dräger de Teran and Suckow (2021).

Definition of indicators: Similarly, the indicators were defined as much as possible based on existing assessment methods documented in the literature. The assessment grid comprises a total of 22 fields of action, 43 indicators, and 112 sub-indicators.

The structure of the assessment grid (without sub-indicators) is shown in Figure 2.

Aggregation method: In accordance with the selected comparison level, aggregation was only carried out within fields of action. A multiplicative method was chosen for the aggregation.

Weighting: Indicator weights were elicited using the Best-Worst Method (BWM; Rezaei, 2015). BWM was selected because it requires fewer pairwise comparisons than the Analytical Hierarchy Process and provides a structured consistency check, reducing elicitation burden while retaining transparent expert judgements. Experts were purposively selected on the basis of experience in food retail, sustainability assessment, and the environmental impacts of food systems; sufficient German-language proficiency was also required. Weights were derived from nine independent expert judgements. The purposive panel was designed to capture relevant expertise rather than statistical representativeness. Uncertainty associated with the weighting was examined by testing the robustness of results under alternative weighting schemes.

Review of the methodology by experts and stakeholders: Various experts and stakeholders were involved in the development of the methodology from the outset of the project. Initially, bilateral discussions were held with the developers of the assessment approaches identified in the literature analysis. After a draft assessment grid had been drawn up, a stakeholder and expert workshop was held. The project team presented the developed assessment grid in a workshop attended by civil society actors and assessment experts. The aim of the workshop was to review the assessment methodology and the relevance of the selected fields of action. A total of 24 participants from ministries, federal agencies, NGOs, university and non-university research institutions, planning offices, private sustainability service providers and non-profit institutes took part. The feedback from the workshop participants was incorporated into the assessment grid.

The basic framework of the assessment grid was also presented to the eight participating large food retail companies and the Federal Association of the German Food Retailers (BVLH) for review, to assess the feasibility and compatibility of the available data. Two pilot applications were then carried out with two companies (pre-test) to gather detailed feedback on the application of the indicators and to test the assessment grid. In a multi-stage process, the BVLH's suggestions for reducing the amount of data in the assessment grid were integrated where appropriate.

3.3 Data collection

Questionnaire development: The questionnaire was designed as an Excel file (Supplementary material) and tested with one participating and one non-participating food retailer. Suggestions from the pre-test were incorporated into the final version. The complete assessment grid, indicator-specific scoring scales, and weighting factors are documented in the publicly available technical report (Keller et al., 2022).

Data collection and verification: Data collection proceeded in two stages. First, Excel questionnaires based on the evaluation grid were pre-filled with publicly available company information, including sustainability reports and press releases. The questionnaires were then sent to the companies for correction or supplementation. After 4 months, completed questionnaires were available for all companies; where information remained insufficient for assessment, additional data were requested individually.

3.4 Evaluation

Subdivision of the assessment into sub-indicators: Where necessary for a consistent assessment, the indicators were subdivided into sub-indicators. This mainly concerned indicators relating to the product range. Here, a subdivision was made according to product range categories.

Independent evaluation by several people: Company performance was assessed using the defined indicators and sub-indicators. Each company was classified according to the descriptors for the indicator-specific scale levels on the basis of the information collected. Where indicators had been subdivided, each sub-indicator was assessed separately before aggregation. The standardized rating scale ranged from 1 (“no practice”) to 5 (“best practice”), with indicator- and sub-indicator-specific descriptors intended to support consistent classification.

Sensitivity and uncertainty analysis: Because quantitative data were limited, the uncertainty analysis focused on the robustness of results to alternative indicator weightings.

To improve comparability, each indicator was evaluated for all companies by the same assessor. A second team member independently reviewed the evaluation and proposed revisions where necessary. Indicator-level assessments were subsequently discussed by the full project team and checked for plausibility. Each assigned scale value was supported by a brief written justification.

Verification of the assessment by the respective company: The scales, assessments, and justifications were sent to each company for verification. Under the confidentiality agreement, companies had access only to assessments based on their own data. Each company's results were also presented and discussed in a separate workshop, providing an opportunity to identify errors or inconsistencies. The project team reviewed the resulting comments before finalizing the evaluations.

4 Pilot application: results

This section presents the pilot results by field of action, environmental category, and individual company. Detailed results for all companies and fields of action are provided in Keller et al. (2022).

4.1 Comparison by field of action

4.1.1 Comparison in the sphere of influence of the supply chain

Overall, companies performed moderately in the supply chain sphere of influence: on average, they scored between “No Practice” (scale level 1.0) and “Average Practice” (scale level 3.0) (Figure 3).

Performance was higher in the communication-related field of action (L) than in the strategy- and implementation-related fields. The lowest scores occurred in implementation-related fields of action.

Strategic performance varied across companies. Substantial differences were observed in the use of product-range-specific and comprehensive impact analyses, which can provide a basis for target setting and implementation. These differences were reflected in targets that, in several cases, did not meet the SMART criteria (Rodgers and Hunter, 1992).

Substantial implementation gaps were identified in measures for sustainable product-range design. Measures such as traceability systems, increasing the share of certified products, strengthening procurement guidelines, and implementing sustainability projects were, on average, only partially implemented.

In the communication-related field of action, all companies reported on sustainability activities using recognized standards, including the Global Reporting Initiative. However, the available reporting did not provide sufficient additional information to enable stronger cross-company comparability or demonstrate sector-wide initiatives.

4.1.2 Comparison in the sphere of influence at “Own Locations”

In the sphere of influence “Own sites”, companies scored on average between “Low Practice” (scale level 2.0) and “Average Practice” (scale level 3.0), indicating only moderate use of their scope for action to reduce environmental impacts at their own sites (Figure 4).

The pattern observed in the supply-chain sphere was also evident at companies' own sites: communication-related performance was stronger than strategic positioning, which in turn was stronger than the implementation of substantive improvement measures.

Several issues relevant to environmental performance, including sustainable investment criteria for occupational pension schemes, were not yet integrated into company strategies. Companies also had difficulty quantifying progress for some improvement measures. For example, none was able to provide quantitative information on reductions in food waste at its own sites.

External regulatory and economic conditions, including refrigeration and air-conditioning requirements, energy-efficiency regulations, vehicle-emission limits, and energy costs, created clear incentives for change. By contrast, the scope for differentiation through voluntary measures at the own-site level appeared comparatively limited.

4.1.3 Comparison in the sphere of influence: consumption

On average, company performance in the consumption sphere of influence ranged from “No/Low Practice” (scale levels 1.0/2.0) to “Average Practice” (scale level 3.0) (Figure 5).

All companies implemented environmental or awareness-raising campaigns, but information on investment levels, duration, and scope was insufficient for systematic cross-company comparison. The pattern observed in the other spheres of influence was also evident here: communication activities were more developed than evidence of implementation. Information on the reach of awareness-raising measures and their coverage across product ranges was incomplete in several cases.

The results indicate substantial scope to strengthen support for environmentally preferable purchasing decisions, particularly through product placement, the availability of certified and plant-based products, and consumer information.

4.2 Comparison by environmental category

In the 2020 reference period, climate was the environmental issue receiving the greatest attention across supply chains, own sites, and consumer-facing activities. Most companies sought to derive climate targets from scientific evidence and considered both direct operations and supply chains. Within product portfolios, climate-related topics such as deforestation, soy used as animal feed, the balance of animal and plant-based products, and products from grassland-fed ruminants were addressed by some companies.

Water-related issues, including water quality and consumption, also received considerable attention, particularly for sourcing regions affected by water scarcity. Some companies addressed these issues through third-party certification schemes such as Fairtrade, Rainforest Alliance, organic labels, and Donau Soja, although certified products represented only a small share of most relevant raw-material categories.

Deforestation was addressed by all companies, whereas biodiversity was incorporated into the strategies and concrete measures of only a subset of companies. Potentially relevant certification schemes for biodiversity, including organic labels, Rainforest Alliance, Donau Soja, and ProTerra, were used, but their shares within the most relevant raw-material categories remained small.

Soil received comparatively little attention at either the strategic or implementation level, with only isolated measures reaching the “average practice” category.

Overfishing received greater attention, as reflected in the comparatively high share of certified products in the fish and seafood category. Nevertheless, most related measures were neither systematic nor comprehensive.

4.3 Comparison between companies

Overall environmental performance was relatively homogeneous across the assessed companies. However, discounters such as ALDI SÜD, ALDI Nord, and Lidl performed better in the supply-chain sphere of influence, consistent with their higher share of private-label products and consequently more direct control over procurement of relevant raw materials. In the consumption sphere, full-range retailers performed as well as or better than discounters, particularly through broader environmental campaigns and consumer-awareness activities.

4.4 Summary of pilot results

The pilot results indicate that German food retailers had integrated environmental issues into corporate strategies to varying degrees. The breadth and ambition of several targets suggest awareness of the need for further improvement, and some industry initiatives achieved broad market penetration, including the Science Based Targets initiative and the Too Good Too Waste campaign. In several supply-chain fields of action, discounters performed slightly better than full-range retailers, plausibly reflecting their more concentrated product portfolios and higher private-label shares.

No company demonstrated consistently leading performance across all assessed fields of action. Substantial gaps were identified both in the implementation of measures to reduce environmental impacts and in the integration of sustainability into corporate management. In addition, some targets could not be assessed consistently because the relevant product scope or time frame was insufficiently defined.

Many of the environmental measures implemented by large food retailers were aligned with direct business benefits, such as cost reductions from energy-efficiency improvements or additional revenues from premium product ranges. Reputation-enhancing communication measures were also widely used. Together, these patterns indicate a predominantly business-case-oriented approach to environmental sustainability. By contrast, measures that could conflict with short-term commercial interests or entail economic risk were less common, even where they could contribute substantially to environmental objectives, for example by limiting advertising for animal products or off-season goods.

Differences in assessment results indicate that further improvement is feasible for comparable companies and may confer advantages on firms that act early. Fields in which all companies performed poorly may reflect structural constraints that individual firms cannot resolve or, alternatively, limited use of available company-level options. The former points to a role for changes in the policy framework, whereas the latter indicates remaining scope for corporate action.

5 Discussion

5.1 Methodological limitations and generalisability

The pilot application has several limitations that constrain the interpretation and generalisability of its results. First, it assessed only eight large retailers in Germany, used 2020 as the reference year and focused exclusively on the environmental dimension. The sample covered the dominant firms in a highly concentrated national market, but the resulting performance scores should not be extrapolated to smaller or independent retailers, other retail formats, or countries with different market structures, regulatory regimes and supply-chain configurations. In addition, the breadth of the assessment—43 indicators and 112 sub-indicators across thousands of products—required some fields of action to be assessed at a relatively aggregate level. Individual procurement projects, for example, were not evaluated separately; instead, project portfolios were categorized according to their coverage and expected effectiveness.

Second, the weighting process necessarily involved expert judgement. BWM provided a structured elicitation procedure, but the composition and size of the expert panel may still influence the resulting weights. Sensitivity analyses indicated that relative differences among companies at the field-of-action level were reasonably robust to alternative weighting and aggregation assumptions, although uncertainty remains. Future applications should document expert selection in detail and, where feasible, compare expert-derived weights with empirical or stakeholder-derived alternatives.

Third, data availability and comparability were limited. Companies made substantial efforts to provide information, but gaps remained and some commercially sensitive data—particularly sales and product-group shares—could not be disclosed. Supply-chain assessment was therefore restricted largely to private-label products, although retailers can also influence third-party brands through assortment decisions, procurement conditions and shelf access. This restriction reduces the completeness of the pilot assessment and should be addressed in future applications.

Fourth, part of the evidence was self-reported by companies and could not be independently audited. The project team checked completeness and plausibility, used independent internal review and returned assessments to companies for verification, but these procedures do not substitute for external assurance. An auditing scheme would be required if the results were to support verified marketing or public-performance claims. Future studies may also reduce dependence on self-reported data by combining company submissions with systematically collected external data, including automated web-based sources (e.g., Winkel et al., 2026).

Finally, company structure affects the practical scope for action. Discounters with narrower assortments and higher private-label shares can exercise more direct procurement control than full-range retailers with broader portfolios and larger shares of third-party brands. Such structural differences should therefore be treated as contextual variables rather than interpreted automatically as differences in management quality.

5.2 Comparison with previous research and international benchmarks

The pilot findings are broadly consistent with earlier retailer assessments. The Swiss WWF assessments (Meyer and Rüegge, 2019; WWF Switzerland, 2019) likewise identified substantial differences between companies and uneven progress across sustainability topics. Superlist-related work similarly highlights gaps between corporate sustainability ambitions and implementation (Haan et al., 2023; Winkel et al., 2026). Across these studies and the present pilot, climate strategies, certified sourcing, and reporting are comparatively well established, whereas evidence of comprehensive implementation across product portfolios and supply chains remains more limited.

An important academic comparator is the BIA-Sustainability approach developed by Mackay et al. (2022) and pilot-tested by Serrentino et al. (2024) in Australia. BIA-Sustainability uses 37 indicators across ten environmental domains and emphasizes the comprehensiveness, specificity and transparency of company policies and practices. Both approaches therefore aim to strengthen corporate accountability through independent benchmarking, but they organize the assessment differently. BIA-Sustainability is structured primarily by environmental topic, whereas the framework developed here starts from retailers' spheres of influence and action levers and is designed to accommodate additional sustainability dimensions. The Australian pilot likewise found that formal commitments were often stronger than evidence of implementation and performance, which parallels the strategy/communication-vs.-implementation gap observed in the German pilot.

5.3 Transferability and future research

The context specificity of the German pilot results should be distinguished from the transferability of the generic framework. The numerical scores and company rankings should not be transferred directly to other countries. By contrast, the framework components—including the definition of objectives and scope, spheres of influence, indicator design, weighting and aggregation, data verification, and uncertainty analysis—are intended to be adaptable across contexts. Applications in other countries should therefore re-specify company selection, product scope, regulatory context, sustainability priorities, indicator thresholds, data sources, and weighting through transparent expert and stakeholder processes. Comparative international applications would help identify which components are robust across contexts and which require local adaptation.

6 Recommendations

6.1 Recommendations for large food retail companies

6.1.1 Take responsibility for the entire sphere of influence

Large food retail companies have so far tended to prioritize sustainability measures that are economically viable or entail limited additional costs. To contribute more fully to food-system transformation and reduce environmental impacts across their spheres of influence, retailers should also consider measures that extend beyond a narrow business-case perspective. Priority areas include:

• Taking responsibility for the entire product range, including third-party brand products.
• Developing private brands toward pioneering ecological product lines.
• Initiating industry solutions for measures that would be particularly effective but whose implementation would entail competitive disadvantages (e.g. restricting advertising of [cheap] meat).

6.1.2 Professionalizing sustainability management in corporate governance

Improving sustainability management in food retail companies so that:

• Objectives are set according to the SMART criteria (specific, measurable, ambitious, realistic and time-bound) (i.e., clear product range reference, quantitative targets, clear time frame, etc.).
• Data availability and controlling are improved to systematically collect important key figures, including processed products, third-party brand products and key performance indicators such as the share of sales of animal products.
• Sustainability management should be placed under the responsibility of senior management and more closely integrated with category management and purchasing.

6.1.3 Develop a consistent sustainability strategy

• Establish sustainability as a living corporate culture: raising awareness (e.g., through training measures for the workforce); creating internal incentives; and linking category management and purchasing more closely to sustainability goals.
• Build trust among consumers, suppliers and employees by aligning communication and actions and strengthening transparency and credibility.

6.1.4 Recommendations for policymakers

Policymakers can address structural barriers that limit further progress toward sustainability in the food system and establish framework conditions under which environmentally preferable decisions do not create systematic competitive disadvantages. Potential measures include:

• Developing a coherent food policy instead of treating agricultural, food, health, social, development and economic policy as isolated policy areas.
• Similarly, develop a complex, well coordinated policy mix of regulatory measures and financial incentives rather than individual policy measures.
• A key component of this policy mix should be the internalization of external costs, thereby creating an economic incentive for the production and consumption of more environmentally friendly products.
• Develop additional complementary (regulatory) policy measures, including sanction mechanisms:
• ° Eliminate “dirty ends”, particularly harmful practices such as air transport of food, through bans or industry solutions.
• ° Define minimum sustainability management standards for supply chains in terms of objectives and control options.
• ° Create clear guidelines for communicating companies' sustainability performance to consumers to establish transparency and trust.

Continuous monitoring of food-retailer sustainability performance could complement these policy measures. The generic framework provides a basis for monitoring voluntary sustainability action alongside regulatory instruments. Transparent information and repeated benchmarking can support the discussion, adoption, and diffusion of sustainability management approaches (Windolph et al., 2014). Within a broader environmental policy mix, governments can also create incentives for retailers to invest in measures that reduce environmental impacts.

7 Conclusions

This study developed a generic framework for comparative sustainability assessment of large food retail companies and demonstrated its use in a pilot assessment of the environmental performance of Germany's eight largest food retailers. The pilot showed that an action-oriented, multi-criteria approach can make differences in corporate strategies, implementation and communication visible across supply chains, own sites and consumption.

The main contribution of the framework is its explicit treatment of retailers' spheres of influence and the methodological choices required for a transparent company comparison. At the same time, the pilot results are context-specific: they reflect a highly concentrated German retail market, the 2020 reference period, an environmental focus, restricted coverage of third-party brands and a data base that partly relied on company reporting and expert judgement. The framework should therefore be regarded as transferable in structure rather than as a source of universally transferable indicator thresholds or scores.

The empirical results indicate substantial unused scope for action, particularly in implementation-oriented fields. Differences among companies suggest that some improvements are feasible within existing business models, whereas uniformly weak performance in other fields may point to structural or regulatory barriers. Benchmarking can therefore complement other policy instruments by increasing transparency and identifying both company-level and system-level intervention needs.

A second assessment using 2023 data broadened the approach to social responsibility and animal welfare (Sander et al., 2025), illustrating the modularity of the framework. Future research should test applications in other countries and retail formats, further develop economic and health dimensions, compare alternative weighting and aggregation strategies, and improve reproducibility through open indicator definitions, scoring rules and questionnaires where confidentiality permits.

Generative AI statement

The author(s) declared that Generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher's note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/frsus.2026.1720594/full#supplementary-material

Supplementary material Supplementary material at the source

Footnotes

Edited by: Christian Bux, University of Foggia, Italy

Reviewed by: Pawel Tadeusz Kazibudzki, Opole University of Technology, Poland

Janina Jȩdrzejczak-Gas, University of Zielona Góra, Poland