The product portfolios of the world’s top 20 food and beverage companies are dominated by confectionery, savory snacks, biscuits, cakes and soft drinks, and 89% of overall sales were from unhealthy foods that are high in sugar, fat and salt. For every $10 spent on these companies’ products in 2020, only $1.10 was spent on healthier products. Companies dependent on soft drinks and confectionery saw very little – or in the case of Red Bull and Ferrero, none – of their sales derive from brands classed as healthier. Healthier categories, such as fresh fruit and vegetable products and breads and other staples, represented a small proportion of overall sales.

It is increasingly important to protect the fundamental goals and reputations of public health bodies and other organizations from association with food and beverage companies whose revenue derives from products that contribute directly to the ill-health of populations worldwide. The methods presented in this paper provide an objective and transparent means to assess companies’ product portfolios and sales. We propose that there are two main ways this paper can be used by other organizations. The first is that the method could be adapted for use in other countries, with different sales data, composition databases and nutrient profile models that are specific to the organization’s needs and country context. To aid this process, we have presented a detailed step-by-step guide to the method (Fig. 1). The second is that organizations could use the global results presented here in this paper directly – they could be used to inform their engagement policies and identify conflicts of interest. These methods could also be used to inform advocacy groups and policymakers about the extent to which companies are contributing towards poor diets, and better hold them to account. The finance community could also use these methods when considering risk of investment and how vulnerable food and beverage companies are to stricter health-related policies – including fiscal measures – in the future.

However, it should be noted that the method described here has only covered one element of a company’s business practices: the healthiness of its product portfolio and sales. Other elements that may be of importance to other organizations and agencies include the marketing practices, lobbying practices and influence on health and diet policy at an international, national and regional level of each individual company. Existing resources such as ATNI’s Global Index and Country Spotlight Reports [12] provide some insights into these practices. Organizations in the global health and nutrition community could also consider setting a threshold for the proportion of unhealthy sales food and beverage companies must meet in order to engage with them. For example, it could be an organisation says the minority of a company’s sales must be from unhealthy foods (e.g. 49%, 25% or < 10%) or if repeated over time, could be based on an improvement (e.g. a reduction of unhealthy sales by 50% over a given time period). Setting and validating these thresholds is something that warrants further research, and could encourage companies to improve the healthiness of their sales.

While a theoretical typology for public–private engagement in the nutrition sector has been published [26] and UNICEF has published guidance on engaging with the food and beverage industry [27], this is the first time an objective and quantitative tool to inform engagement decisions has been published. This study has wide geographical reach and includes data from seven food and beverage markets from each region of the globe. Including additional countries in the future would give further insight into how leading multinational companies’ sales vary geographically. Repeating this study annually would allow for companies’ progress to be tracked over time. While it is the largest multinational companies that are most likely to seek public–private partnerships and memberships in multi-stakeholder platforms, this study is limited by only including 20 companies. Nationally, a different set of local manufacturers may approach organizations with offers of corporate engagement which are not covered here. We were not able to identify any examples of companies with the majority of their sales coming from healthier products that might be considered lower risk for organizations to engage with. Previous work in nutrient profiling has demonstrated that companies with portfolios dominated by dairy products often have overall “healthier” product offerings [28] and it may be that in countries where dairy companies dominate (such as in Europe), there may be cases of best practice to identify. However, there may be other factors that influence engagement principles with dairy companies, including their compliance with the international code of marketing of breast milk substitutes.

Euromonitor sales data cover the brands sold from packaged food and soft drink categories in 80 countries globally. The main limitation is that brand level sales data (e.g. Yoplait yoghurts) had to be paired with the nutrition information for individual products (e.g. Yoplait original and Yoplait light). This meant that where a brand has multiple product flavors/variants, sales of each individual product was weighted equally, when it may be that one product variant/flavor represents more sales. Ideally, product level, rather than brand-level, sales data could be used, although there are limited data sources for product level sales, and those that do exist have their own limitations, including being expensive, having publication limitations and high levels of imputation [29, 30]. This could be overcome if companies were more transparent with their data, although they are unlikely to want to report individual product sales for fear of disclosing information to their competitors. Reporting percentages instead of absolute values ($) could help overcome this.

There were 316 brands that were excluded from this analysis because they could not be paired with nutrition composition data. This represents 20.0% of the total number of brands, but only 12.1% of total sales, suggesting that these were smaller brands with smaller sales. There was geographical variation, with 97% of sales data matched with nutrition composition data in Brazil, compared to just 62% in China. These differences are down to three reasons: errors in the sales data, where brands that are no longer sold in that market are still listed; brands not being captured in the FoodSwitch database; or differences in the translation of product names between the sales and composition databases. In the future, missing data could be reduced by using product-level sales data that includes barcode, so that products between databases could be better matched, although more granular sources of sales data are often prohibitively expensive [29, 30].

There are several NPMs that could have been used in this study. The WHO has a different model for each of its regions [31,32,33], and the Pan-American Health organization (PAHO) also has an NPM [33]. There are also other widely used models, including the UK Ofcom model [34], Nutri-Score which is used widely across Europe [35], and Australia and New Zealand’s Health Star Rating system [36]. The WHO Euro NPM was used because it is approved by a UN-agency, is regularly used by researchers and using a single model allowed for a unified and comparable method across countries globally. However, depending on the geographies included, a country specific scheme may be preferable. It is important to note that some other models, such as the PAHO NPM, are stricter than the WHO Euro NPM [37] while others e.g. UK NPM, that do not have thresholds for NSS and therefore class zero-calories soft drinks as healthier, are less so.

There are other initiatives that score and rank food and beverage companies based on their performance in nutrition and health which have reached broadly similar conclusions that the world’s largest food and beverage companies are focused on the sale of less healthy products. The Access to Nutrition Initiative (ATNI) [28] aims to rank global food and beverage companies on their contribution to addressing malnutrition, including overweight and obesity, undernutrition and micronutrient deficiencies based on a range of areas, including commitments and practices in terms of governance and management, healthiness of their products based on the Health Star Rating, and how they influence consumer behavior through labelling and marketing practices. ATNI’s most recent Global Product Profile examined the healthiness of the 25 largest companies globally in 25 countries and found that overall 31% of products would be considered “healthier” using the Australasian Health Star Rating system, and that by using the locally-appropriate WHO NPM that only 9% of products would be considered healthier. These results seem in line with the findings presented here and highlight that using a different NPM is unlikely to change the main observation the majority of the world’s top food and beverage companies’ sales are from unhealthy products.

INFORMAS (International Network on Food and Obesity/NCD Research, Monitoring and Action) [38] has produced a series of company scorecards that assess companies’ impact on the food system using the Business Impact Assessment on obesity and nutrition (BIA-Obesity) tool, with a focus on companies’ self-reported nutrition policies, commitments, disclosure and performance [39]. While these metrics are based on business practices and companies’ commitment to nutrition-related policies, they are not directly comparable to the quantitative analysis presented in this study.