Authors
Dr. Ana Isabel Lopes
Department of Communication Science, Vrije Universiteit Amsterdam, The Netherlands
a.i.loureiro.lopes@vu.nl
Dr. João Gonçalves
Department of Media and Communication at Erasmus University Rotterdam, The Netherlands
ferreiragoncalves@eshcc.eur.nl
Corporations are pulling back on public diversity commitments to minimize political risk and stakeholder backlash, shifting toward strategic silence and treating diversity messaging as a flexible reputational tool rather than a permanent corporate value.
We analyzed the social media posts of top 20 European and US companies for their Diversity related content before (2024) and after (2025) the election of Donald Trump. We find that there was a decline in diversity related content in 2025 both in the US and in Europe, with the biggest drop being registered in the US. From all social media content, 6,7% of posts by US based companies in 2024 included diversity related topics, while in 2025 only 1,3% were about diversity. For European companies, the scenario is similar. In 2024, 12,6% of social media posts by European companies touch upon diversity content which dropped to 7,0% in the same period of 2025.
In general, European companies posted more on diversity than American companies both in 2024 and in 2025. These results show us that while the drop in diversity related talk by corporations was more accentuated in the US, other geographies seem to also replicate the trend. The Figure below illustrates these numbers.
Prevalence of Diversity in Corporate Social Media Posts
EU vs US, 2024 vs 2025
- EU 2024
- EU 2025
- US 2024
- US 2025
View the data as a table
| Topic | EU 2024 | EU 2025 | US 2024 | US 2025 |
|---|---|---|---|---|
| Diversity (Overall) | 12,6% | 7,0% | 6,7% | 1,3% |
| Sexual Orientation | 10,4% | 13,3% | 2,7% | 2,9% |
| Gender | 5,6% | 3,8% | 3,3% | 0,6% |
| Ethnic / Cultural | 7,1% | 3,5% | 3,6% | 0,7% |
Context
Corporate communication on diversity has shifted from a peripheral concern to a central expression of organizational values, legitimacy, and social responsibility. This evolution has been particularly evident among large European and American corporations, whose public messaging increasingly addresses diversity, equity, and inclusion in response to social movements, stakeholder expectations, and regulatory and cultural pressures. The election of Donald Trump in 2024 marks a significant new inflection point in this trajectory, occurring within a context of renewed political polarization and intensified debate over the role of corporations in addressing social issues.
This whitepaper examines how diversity-related communication by European and American corporations evolved before and after the 2024 Trump election. Through a comparative transatlantic analysis, this white paper explores how shifting political climates influence corporate discourse and how firms navigate contested diversity narratives across different institutional and cultural environments.
Donald Trump was elected in November 2024 and Assumed office on January 20, 2025. To compare pre and post Trump periods, data was collected from Facebook for the periods of March 2024 – June 2024 and March 2025 – June 2025. The companies under study are the top 20 European and 20 US based companies that are on the stock market at the time of data collection (August 14th, 2025).
Taken together, these findings suggest a notable retrenchment in corporate social responsibility (CSR) signaling and public-facing commitments to diversity in the wake of a shifting political climate. The sharp decline in diversity-related communication, especially among U.S.-based corporations, indicates that many firms may be recalibrating their Public Relations strategies to reduce perceived political risk, controversy, or stakeholder backlash, rather than maintaining consistent normative commitments, which is considered a form of corporate washing (Lopes & de Keyzer, 2026). For CSR, this raises questions about the depth and durability of corporate engagement with diversity: whether it reflects long-term responsibility embedded in organizational practice or a more contingent, context-sensitive form of reputational and stakeholder management. From a public relations perspective, the results highlight how corporations actively adjust discourse in response to political leadership and polarization, prioritizing strategic silence or neutrality over advocacy when social issues become more contested. The transatlantic differences further imply that institutional and cultural environments continue to shape how confidently firms communicate social values, even as global corporations face shared pressures to balance authenticity, legitimacy, and risk in their external communications.
The table below shows the companies under analysis, as well as the number of posts and the percentage of diversity content in 2024 and 2025.
Companies under analysis
European companies
| Company | Number of followers | Number of posts 2024 | Number of posts 2025 | % Diversity content 2024 | % Diversity content 2025 |
|---|---|---|---|---|---|
| Astra Zeneca | 72K | 18 | 10 | 5,6% | 30,0% |
| HSBC | 3.1M | 11 | 9 | 36,4% | 33,3% |
| Shell | 4M | 1 | 5 | 0,0% | 20,0% |
| Roche | 105K | 31 | 25 | 22,6% | 12,0% |
| Novartis | 428K | 18 | 10 | 0,0% | 0,0% |
| Nestlé | 11M | 26 | 31 | 7,7% | 0,0% |
| ASML | 158K | 37 | 24 | 2,7% | 8,3% |
| SAP | 1.4M | 35 | 14 | 17,1% | 0,0% |
| SIEMENS | 1.6M | 87 | 50 | 9,2% | 12,0% |
| Deutsche Telekom | 539K | 103 | 73 | 14,6% | 6,8% |
| LVMH | 359K | 52 | 85 | 9,6% | 7,1% |
| Hermès | 3.9M | 19 | 24 | 0,0% | 0,0% |
| L’Oréal | 2.8M | 12 | 44 | 8,3% | 9,1% |
| Novo Nordisk | 756K | 17 | 2 | 23,5% | 0,0% |
| Airbus | 2.8M | 26 | 50 | 0,0% | 0,0% |
| Allianz SE | 3.9M | 23 | 2 | 13,0% | 50,0% |
| Accenture | 1M | 41 | 15 | 17,1% | 26,7% |
| Unilever | 3.9M | 39 | 27 | 0,0% | 3,7% |
| Arm Holdings | 118K | 7 | 7 | 0,0% | 0,0% |
| Schneider Electric | 5M | 213 | 178 | 18,3% | 5,1% |
US companies
| Company | Number of followers | Number of posts 2024 | Number of posts 2025 | % Diversity content 2024 | % Diversity content 2025 |
|---|---|---|---|---|---|
| NVIDIA | 2.4M | 115 | 118 | 0,0% | 0,8% |
| 34M | 145 | 9 | 10,3% | 0,0% | |
| Amazon | 29M | 37 | 151 | 2,7% | 0,0% |
| Meta | 94M | 40 | 33 | 5,0% | 0,0% |
| Walmart | 32M | 30 | 42 | 6,7% | 0,0% |
| JPMorganChase | 465K | 37 | 27 | 16,2% | 7,4% |
| Oracle | 2.8M | 165 | 139 | 1,8% | 0,0% |
| Eli Lilly | 260K | 28 | 31 | 35,7% | 12,9% |
| Costco | 3.1M | 160 | 128 | 0,0% | 0,0% |
| Johnson & Johnson | 1M | 23 | 20 | 21,7% | 0,0% |
| Home Depot | 5.8M | 113 | 107 | 1,8% | 1,9% |
| Bank of America | 2.9M | 68 | 66 | 4,4% | 1,5% |
| Chevron | 1.2M | 57 | 28 | 8,8% | 3,6% |
| Coca-Cola | 1.2M | 7 | 12 | 0,0% | 8,3% |
| UnitedHealth | 156K | 16 | 19 | 25,0% | 0,0% |
| Cisco | 2.5M | 198 | 57 | 9,1% | 1,8% |
| Wells Fargo | 1.1M | 48 | 37 | 12,5% | 8,1% |
| Philip Morris International | 1.4M | 58 | 37 | 32,8% | 0,0% |
| AMD | 3M | 26 | 48 | 3,8% | 0,0% |
| Sales Force | 794K | 252 | 113 | 0,8% | 0,0% |
| Morgan Stanley | 137K | 37 | 27 | 16.2% | 7.4% |
Notes on Methodology
The original list of companies contained other companies which data was not scraped because they had no Facebook page, the Facebook page had less than 50K followers or there were no posts in at least one of the periods under study. When the companies had different pages for each region, the Default Page was selected. The final list of companies is presented on the table above. Having selected the companies that fit the criteria, we manually scrapped the text of all the Facebook posts (including shares from other pages) published in the designated period (since the data was collected in August 2025, we do not have access to posts that might have been deleted). Some of the companies had at least some of the posts in other languages than English. To make data analysis possible, these posts were translated to English using the Google Translate API.
To automatically classify social media posts for their diversity we relied on the training data of an existing classifier and definitions1 for social media diversity (Hofhuis et al., 2024) To maximize accuracy and reliability, we applied a state of the art BERT-NLI classification approach (Laurer et al., 2024). The classifier showed good reliability statistics with a macro F1 score of .85 for the presence of diversity and an accuracy of 89.1%. Reliability values for gender (F1 = .71; Acc = 93.2%), sexual orientation (F1 = .73; Acc = 97.4%) and ethnic/cultural (F1 = .67; Acc = 88.2%) were lower, reflecting training data sparsity.
Data is available here: https://doi.org/10.5281/zenodo.21980361.
Footnote
1 The concept may include any mention of diversity in any of its dimensions, or of diversity management practices and policies in the organizational context. The dimensions of diversity may include gender, religious beliefs, race, marital status, ethnicity, parental status, age, education, physical and mental ability, income, sexual orientation, occupation, language, geographic location, etc. ↩
References
Hofhuis, J., et al. (2024). “Examining strategic diversity communication on social media using supervised machine learning: Development, validation and future research directions.” Public Relations Review, 50(1): 102431.
Laurer, M., Van Atteveldt, W., Casas, A., & Welbers, K. (2024). Less annotating, more classifying: Addressing the data scarcity issue of supervised machine learning with deep transfer learning and bert-nli. Political Analysis, 32(1), 84-100.
Lopes, A. I., & De Keyzer, F. (2026). Communicating Corporate Social Responsibility and Advocacy on Social Media as a Driver for Societal Change. In The Impact of Marketing Strategies on Society: Examining Critical Aspects of Marketing and their Implications for Societal Well-being (pp. 307-326). Cham: Springer Nature Switzerland.
How to contact us?
For questions regarding the whitepaper or the CSR Science Hub please contact Dr. Ana Isabel Lopes (a.i.loureiro.lopes@vu.nl).
©2026 CSR Science Hub. All rights reserved. This whitepaper is protected by copyright law. You are permitted to share, copy, and distribute this document in its original, unedited form for non-commercial, educational, or informational purposes, provided full credit and citation are given to the authors. You may quote or reference short passages from this whitepaper, provided that appropriate attribution is included along with a direct link or reference to the original document. No part of this publication may be modified, adapted, translated, or commercially exploited without the prior written permission of the copyright holder.