Communicating operational excellence 5.0: evidence from CSRD sustainability disclosures

Industry in the European Union (EU) faces a dual challenge: regaining its global industrial position while simultaneously addressing the environmental and social imperatives outlined in the 2030 Agenda for Sustainable Development (European Commission, 2021; UN, 2015). Until the late twentieth century, European industry and related services accounted for a significant share of global production. However, in recent decades, they have experienced stagnating output, job losses, and declining competitiveness, largely due to accelerated industrialisation in Asia and other emerging regions. This process of deindustrialisation has been exacerbated by successive shocks, including the 2008 financial crisis, the COVID-19 pandemic, and the energy crisis triggered by the war in Ukraine (Eurostat, 2025).

Between 2000 and 2020, the EU’s industrial production accounted for approximately 20% of its GDP, but its average annual growth was only 0.3%, compared with 3.2% globally. More than 2.3 million industrial jobs were lost across the EU between 2008 and 2023, with Germany, Poland and Romania particularly affected (European Commission, 2021; World Bank, 2025). However, these aggregate figures mask substantial differences across sectors. For instance, while clothing production declined by an average of 6.6% annually, pharmaceuticals grew by 4.5% over the same period (Eurostat, 2025).

This deindustrialisation coincides with growing regulatory and social pressure to address climate change, ecosystem degradation (e.g. desertification, deforestation, and biodiversity loss), sustainable water management, responsible consumption and production, the development of resilient and sustainable cities and access to affordable and clean energy (UN, 2015). In response, the European Commission has proposed advancing Industry 5.0 and adopting Operational Excellence 5.0 (OE 5.0) (European Commission, 2023).

OE 5.0 builds on the long tradition of Lean Six Sigma and operational excellence, extending its fundamental principles, such as waste reduction, process stability and continuous improvement, towards broader strategic objectives (Longo et al., 2020). While Lean Six Sigma has focused on production efficiency and cost minimisation (Oakland, 2014), OE 5.0 incorporates the dimensions of sustainability, resilience and employee well-being into the value creation function (Singh et al., 2024). From a strategic perspective, these dimensions can contribute as organisational capabilities to generate competitive advantages that are difficult to imitate and are integrated into routines and culture, in line with the resource-based view of the firm (Barney, 1991). Furthermore, OE 5.0 takes advantage of technological and organisational opportunities to transform its operating systems to dynamically address sustainability and other environmental challenges (Teece et al., 1997).

Despite its conceptual advantages, commitment to OE 5.0 does not guarantee successful implementation. Previous research documents numerous failures in operational excellence initiatives, often related to cultural resistance, lack of strategic alignment or insufficient integration of human and technological resources (Marolla et al., 2022). Consequently, academics, management practitioners, and policymakers alike face the challenge of assessing how organisations prioritise and communicate OE 5.0-related objectives, especially in a context characterised by increasing regulatory oversight.

From 5 January 2023, the EU began replacing the Non-Financial Reporting Directive (NFRD) with the Corporate Sustainability Reporting Directive (CSRD), which elevates sustainability information to the same level as financial disclosure (BOE, 2022). CSRD requires companies to divulge information in their annual reports on their environmental and social impacts, as well as the effect on their businesses. However, beyond its regulatory compliance function, the CSRD is a powerful institutional mechanism for building corporate image. Rather than neutral descriptions of operational reality, sustainability reports are strategic narratives through which companies signal priorities, manage stakeholder relationships, and respond to regulatory and social demands.

Although the CSRD provides a valuable source of information on companies’ strategic positions on sustainability, their environmental and social impact, and their degree of OE 5.0 implementation, it has been underexplored in the literature on Lean Six Sigma, which has focused predominantly on bibliometric analyses (Prakash et al., 2022). For example, Sakib et al. (2025) extracted emerging trends in production and LSS processes in a bibliometric review of academic publications. Raval and Kant (2017) criticised studies in this field for neglecting the voices of practitioners and the information published by companies. Indeed, recent studies have shown that narratives describing strategic proposals and future actions published in annual reports tend to provide greater informational value to investors and stakeholders than historical accounting data (Gensler et al., 2024).

This study addresses this gap by proposing an analytical framework based on textual analysis of CSRD (or older NFRD) reports to examine their strategic emphasis on the implementation of OE 5.0. It is important to note that the information disclosed in these reports may be skewed to reflect intentions rather than substantial organisational changes (Gensler et al., 2024). Nevertheless, understanding how organisations frame and prioritise environmental, social and governance (ESG 5.0) issues in their official narratives could help predict their future behaviour.

Methodologically, the qualitative study applies dictionary-based text analysis using the Elasticsearch algorithm (Gormley and Tong, 2015) to evaluate the emphasis in CSRD reports on three Sustainable Development Goals (SDGs), a set of proposals introduced by the United Nations as part of the 2030 Agenda. Those three goals are SDG 8 (Decent work and economic growth), SDG 9 (Industry, innovation and infrastructure) and SDG 12 (Responsible consumption and production). The empirical analysis focuses on a single sector, Spanish insurance companies. Unlike manufacturing or logistics firms, insurance companies do not have an intensive operational structure. However, they do provide complementary services by covering operational and workplace risks, such as accidents, damage to property and health-related incidents, thereby helping firms manage uncertainty and maintain safe working environments (Skogh, 1998).

The insurance sector, like other service industries, operates in a context where outcomes are intangible, customer involvement is high, and there are few standardised manufacturing processes. As a result, it entails less operational complexity than manufacturing or logistics (Roth and Menor, 2003), making insurance firms relatively homogeneous and therefore easier to compare. They also systematically publish sustainability information in their CSRD reports.

Previous studies suggest that each industrial sector emphasises different strategic objectives (Han et al., 2017) and, therefore, an aggregate analysis of companies from disparate sectors often produces results with substantial noise and little comparative value (Mizik and Jacobson, 2003). Furthermore, the focus on Spanish companies is also relevant, as they are facing intense pressure to embrace digital transformation, meet environmental objectives and respond to increasing global competition.

Given the exploratory nature of the research, no hypotheses are proposed, and the empirical findings are presented descriptively (Golder et al., 2022). However, the study introduces a methodological process that is transferable and applicable to any sector or institutional context, such as manufacturing, logistics, healthcare services and energy production or distribution. The study combines three areas of analysis: strategic theory, institutional analysis and computational data mining applied to text analysis, contributing to the advancement of research at the intersection of Lean Six Sigma, Industry 5.0, and sustainability reporting. Furthermore, it provides a replicable method of analysis for studying the narratives used in corporate communication of strategic commitments in a context fraught with regulatory constraints (Gensler et al., 2024).

This article proceeds as follows. Firstly, the concept of OE 5.0 is presented, together with its role in value creation. Secondly, the relevance of non-financial information for assessing corporate sustainability, particularly textual disclosures in CSRD reports, is highlighted, and an organisational framework for analysis is introduced. Thirdly, the selected SDGs are described in relation to the development of the analytical dictionary. Fourthly, the Elasticsearch algorithm is explained. Section 5 outlines the methodology, fieldwork and results. Finally, implications for practice and research are discussed, including directions for future research on operational strategies.

Treacy and Wiersema (1993) characterised operational excellence as a commitment to systematically improving processes to deliver high-quality products and services at competitive prices. Central to both operations management and strategy formulation, this approach builds competitive advantage by minimising overhead costs through practices such as eliminating production stages, reducing transaction and friction costs, and adopting management practices based on continuous improvement, standardisation and employee engagement (Treacy and Wiersema, 1993; Oakland, 2014).

Within this tradition, the Lean Six Sigma methodology emerged from a combination of Lean techniques, which eliminate waste and optimise workflow, with the Six Sigma methodology, which reduces variability and defects in processes through statistical control (Oakland, 2014). This approach not only improves product and process quality but also boosts customer satisfaction (Raja Sreedharan and Raju, 2016).

Despite its success, the more traditional approach to Lean Six Sigma has been criticised for its limited focus on production processes, neglecting broader social, environmental and human considerations (Xiang et al., 2023). The emergence of Industry 4.0 further accentuated this disconnect, as the development of technologies such as the Internet of Things (IoT), artificial intelligence (AI) and robotics led to unprecedented levels of automation, but at the same time raised concerns about their side effect on workforce displacement, environmental impact and systemic vulnerability (Breque et al., 2021; Utama and Abirfatin, 2023). As production systems became digitised and globalised, academics and professionals began to recognise that operational excellence could no longer be defined solely in terms of cost reduction and process optimisation.

In response, OE 5.0 has been proposed as a new evolutionary phase of Lean Six Sigma. OE 5.0 maintains the fundamental principles of continuous improvement, process standardisation and data-driven decision-making, but extends them to incorporate sustainability, resilience and human-centred design (Boumsisse et al., 2024). OE 5.0, therefore, represents the combination of the operational rigour established by Lean Six Sigma with the broader social objectives articulated in Industry 5.0 for value creation.

From a theoretical perspective, OE 5.0 can be interpreted from the resource-based view (RBV) of the firm (Kraaijenbrink et al., 2010). This framework argues that sustainable competitive advantage lies primarily in the possession of valuable and unique resources, rather than in external market positioning. OE 5.0’s commitment to integrating sustainable management with workforce engagement, digital coordination and operational resilience seeks to embed intangible operational and organisational routines that are difficult to imitate, making them potential sources of competitive advantage (Barney, 1991).

At the same time, the implementation of OE 5.0 is consistent with the theory of dynamic capabilities (Eisenhardt and Martin, 2017). This theory attempts to explain how organisations sustain their competitive advantage in today’s extremely dynamic environments by sensing, leveraging and reconfiguring resources and competencies to cope with environmental shocks through learning and innovation (Teece et al., 1997). In other words, the adoption of OE 5.0 provides organisations with the robustness to deal more effectively with the challenges of the environment.

However, OE 5.0 extends beyond market and environmental pressures to include those exerted by a broad set of stakeholders and institutional actors. Stakeholder theory posits that value creation (and by extension, long-term viability) depends on firms’ ability to balance the interests of customers, employees, suppliers, public institutions, investors and shareholders (Mahajan et al., 2023; Freeman, 1984). Institutional theory examines why organisations respond to social dynamics rather than allowing rational and efficient operational processes to prevail (Lammers and Barbour, 2006). According to this theory, culture, regulations and social values (institutions) shape organisations to such an extent that they become increasingly similar (isomorphism) in their respective quests to gain legitimacy and social acceptance. But each company may respond differently, ranging from the substantial adoption of new practices to more symbolic forms of compliance, depending on organisational resources, strategic intent and concerns about legitimacy (DiMaggio and Powell, 1983).

In this context, operational excellence is assessed not only through internal performance analysis but also through how companies communicate their financial results, sustainability commitments and future strategic priorities to the external public (Lammers and Barbour, 2006). CSRD, which requires the disclosure of standardised sustainability information alongside economic and financial data, offers a formal channel for such communication. The preparation of these reports provides an opportunity to construct narratives that go beyond reporting past achievements and planned initiatives by also improving an organisation’s corporate image. According to institutional theory, these narratives can be more descriptive in nature, explaining strategies and achievements, or more symbolic, emphasising commitment and intention more than concrete action (DiMaggio and Powell, 1983). However, previous research has highlighted the ability of such narratives to influence stakeholders’ perceptions and expectations (Gensler et al., 2024).

This study considers the adoption of OE 5.0 as both an operational and communicative tool. This approach complements traditional research on Lean Six Sigma, based on process analysis, by considering the arguments used to explain and legitimise the strategic use of OE 5.0 in CSRD reports to stakeholders and under institutional pressure.

Assessing the degree of OE 5.0 adoption, centred on human beings, sustainability and resilience, poses significant methodological challenges as it cannot be determined using traditional operational and financial indicators. Instead, new estimators and indicators need to be developed (Brodny and Tutak, 2025). This study proposes the use of an indirect estimator based on text analysis using dictionaries derived from the SDGs.

The SDGs offer the advantage of specifying regulatory processes, organisational proposals and investment in specific thematic areas (UN, 2015). Given that their approach, in addition to being based on political arguments, is eminently pragmatic, their aims can be considered as estimators of corporate discourse, unravelling the emphasis assigned to different objectives through the description of commitments and narratives.

Although the UN’s 2030 Agenda defines 17 SDGs (UN, 2015) as a broad framework for assessing sustainable industrial transformation, the three most directly connected to OE 5.0 were selected for analysis: SDG 8 (Decent work and economic growth), SDG 9 (Industry, innovation and infrastructure) and SDG 12 (Responsible consumption and production).

This selection can be justified on theoretical grounds. From a dynamic capabilities perspective, these SDGs capture the social and technological challenges associated with OE 5.0, such as labour market transformation and environmental constraints, that require firms to seize opportunities through innovation and process redesign, and to transform operating systems in support of long-term sustainability and resilience. The selection also aligns with stakeholder theory, given the relevance of these SDGs to groups such as employees, regulators, investors, and society in general. Finally, it links to institutional theory, as the descriptions and narratives outlined in the CSRDs fit within a regulatory framework that reinforces institutional and social legitimacy.

SDG 8 captures the human-centred orientation of OE 5.0, as it promotes decent employment and inclusive, sustainable economic growth. Decent work entails productive tasks associated with fair wages, job security, social protection, personal development and social inclusion. This goal is especially pressing given persistent inequalities exacerbated by recent global crises, including high youth unemployment and gender pay gaps (UN, 2015).

OE 5.0 supports SDG 8 by reorienting operations towards worker well-being. Unlike Industry 4.0, which prioritised automation and efficiency, OE 5.0 positions workers as co-creators of value in digital environments (Boumsisse et al., 2024). It promotes ergonomic workplace design, participatory decision-making and lifelong learning. Such measures not only enhance productivity but also reduce the risk of technological exclusion (European Commission, 2021). Training initiatives, including digital skills programmes and interdisciplinary collaboration, further prepare employees for the future.

SDG 9 reflects the technological and infrastructural dimension, encompassing innovation, digitalisation and the resilience of industrial systems. It also promotes interdependence between economic growth, social development and environmental sustainability (UN, 2015). These aims have become critical as manufacturing industries face inflationary pressures, raw material shortages and post-pandemic slowdowns (UN, 2015).

OE 5.0 complements SDG 9 by promoting the integration of emerging technologies such as AI, digital twins, cloud manufacturing and IoT (Xu et al., 2018). It also advocates modular system design (separate, standardised modules or components that can be combined in different ways) and open platforms (software and hardware systems that facilitate interoperability and collaboration). Together, these tools increase flexibility and resilience while enabling real-time analytics, predictive decision-making, and mass customisation. Furthermore, OE 5.0 fosters partnerships between firms, universities, start-ups, and research institutions, supporting innovation and modernisation (Asheim et al., 2007).

SDG 12 focuses on sustainability, particularly resource efficiency, circularity and responsible production practices. It also promotes sustainable consumption characterised by reduced resource use (UN, 2015). Persistent challenges addressed by this SDG include food waste, estimated at 931 million tonnes annually, and the resurgence of fossil fuel subsidies following recent crises (UN, 2015).

OE 5.0 contributes to SDG 12 by embedding circular economy principles into operations. Real-time monitoring improves resource allocation, while environmental metrics track the ecological footprint (Tseng et al., 2018). Closed-loop processes, facilitated by digital twins, enable continuous recycling and reuse, thus reducing waste and reliance on external inputs (Xiang et al., 2023). OE 5.0 also promotes collaboration between firms and consumers, using AI to customise products and engage users in sustainability initiatives. Social media, for example, can disseminate information on product use, maintenance and disposal, aligning with SDG 12’s call for public access to information that supports sustainable lifestyles.

This study employs the Elasticsearch algorithm, a highly flexible text search and analysis engine that operates via the Representational State Transfer (RESTful) protocol. Widely used in web applications for server-client data transfer, Elasticsearch processes full-text inputs and represents them as structured documents, facilitating analyses such as frequency estimation and clustering. Prominent organisations, including Wikipedia, eBay, GitHub and Datadog, use Elasticsearch to search, store, and analyse large volumes of data.

Elasticsearch is built on Apache Lucene, an open-source search engine library written entirely in Java that combines three principal algorithmic components: indexing, query analysis and classification. Its efficiency relies primarily on two elements:

1. An inverted index, which enables rapid retrieval of candidate documents; and

2. A scoring function, traditionally based on Okapi BM25, a probabilistic model that ranks documents according to relevance (Robertson et al., 1994; Manning et al., 2008).

In addition, similarity measures derived from vector analysis are used for document classification, broadening the algorithm’s applicability to semantic search.

Let the collection of documents be denoted by:

where each document d_i consists of a sequence of terms (t₁,t₂,…,t_m).

Elasticsearch replaces each term with a unique, random code (a token) and normalises the text using a parser (stemming to the root form, conversion to lowercase, removal of stop words, etc.), producing a vocabulary:

V={w₁, w₂,…,w_|V|}.

For each term w ∈ V, the system stores a posting list:

P(w)={(d, f _d,w,{p₁,p₂,… }) |w ∈ d},

where f _d,w is the frequency of term w in document d, and {p₁,p₂,… } are the indices of the position occupied by w in d. This mapping of V onto P(w) forms what is called an inverted index (Salton, 1971).

When a query q = {w₁,w₂,…,w_k} is performed, Elasticsearch retrieves the documents ranked by the inverted index and calculates a relevance estimate for each one. The formula used is Okapi BM25 (Robertson et al., 1994), defined as:

where f _d,w is the frequency of w in document d, |d| is the length of the document in number of tokens, avg dl is the average length within the text, k₁∈ [1.2, 2.0] and b ∈ [0, 1] are parameters controlling term saturation and length normalisation, and IDF(w) is the inverse document frequency:

where n = |D| is the total number of documents, and n_w is the number of documents containing w.

This formula balances three elements:

1. Term frequency (how often the word appears);

2. Document inverse frequency (its rarity in the corpus); and

3. Document length normalisation (Manning et al., 2008).

Elasticsearch is used to search and group sets of words either in Boolean form (conjunction/disjunction on posting lists) or in proximity form (using positional indexes) (Salton et al., 1975). However, when working with words, a key difficulty lies in dealing with semantic similarity: documents that use different words to express the same concept may not be retrieved.

To address this limitation, the latest versions of Elasticsearch incorporate vector search, which uses dense numerical embeddings generated by neural language models (Gormley and Tong, 2015; Johnson et al., 2021).

Each document d ∈ D is assigned a vector embedding:

v_d ∈ Rⁿ

Using a trained neural encoder (e.g. BERT, sentence transformer). Similarly, a query q is assigned:

v_q ∈ Rⁿ

The embedding dimension n captures semantic relationships: words or documents with similar meanings are represented close together in the vector space.

In this study, the relevance of a document d to a query q is then calculated based on cosine similarity (which measures the ratio between the adjacent side and the hypotenuse):

Cosine similarity is widely used because the ratio does not depend on the size of the right-angled triangle, but on the angle. In other words, it focuses on the orientation rather than magnitude, which is useful when normalising embeddings.

To evaluate the strategic emphasis on OE 5.0 among insurance companies, annual reports published on their websites from 2023 to 2024 were collected, following Gensler et al. (2024). The content of the CSRD sections was then analysed and classified using the Elasticsearch package (Gormley and Tong, 2015). In line with prior literature highlighting industrial-level variation in strategic applications (Han et al., 2017; Mizik and Jacobson, 2003), the analysis was confined to a single sector: insurance.

However, given that the information published in CSRD reports represents management narratives aimed at multiple stakeholders, the description of strategies and priority objectives does not necessarily translate into operational practices (Gensler et al., 2024). Therefore, the findings should be interpreted as indicators of strategic emphasis, rather than as direct evidence of operational adoption.

The study draws on annual reports published by insurance companies operating in Spain. This industry was chosen because of its relative homogeneity and low operational complexity compared to manufacturing or logistics, making it more suitable for initial testing of a new analytical tool, as it is easier to isolate the tool’s performance from the confusion generated by more complex operating environments (Churchill, 1979). These reports include accounting and stock market information for listed firms, as well as non-financial disclosures concerning strategic directions and sustainability objectives. Companies in SIC code 63 (Insurance and Reinsurance) were selected, provided their reports contained both financial and non-financial information, covering investments, environmental management and sustainable policies for 2023 and 2024. The final sample comprised six firms: Mutua Madrileña, MAPFRE, Generali, Allianz Partners, AXA and VidaCaixa.

Automated text analysis has gained relevance in market research (Berger et al., 2020; Edeling et al., 2021). It has been applied not only to consumer-generated content such as posts or reviews (Moon and Kamakura, 2017) but also to company disclosures, including financial reports (Srivastava et al., 2023).

As Gensler et al. (2024) note, procedures can be bottom-up or top-down and differ depending on the source. In bottom-up procedures, keywords are derived from the data, while in top-down procedures, predefined dictionaries are used (Humphreys and Wang, 2018). For example, in the consumer context, it is common for keywords to be generated inductively from the user’s own language, as in topic or sentiment models (e.g. Hartmann et al., 2019). In contrast, in company documents, such as annual reports, which are typically structured and written in standardised language, dictionary-based approaches are often applied (Nguyen et al., 2025; Srivastava et al., 2023).

Given that the language of CSRD sections is relatively standardised, a top-down dictionary-based method was chosen. This approach ensures transparency, replicability and interpretability, which are particularly relevant when analysing regulated information (Bodnaruk et al., 2015; Loughran and McDonald, 2016).

While more advanced semantic models can capture deeper contextual meaning, e.g. BERT-TextCNN (Liu and Jia, 2025), dictionary-based methods allow for clear traceability between input terms and analytical results, facilitating validation by researchers, practitioners and regulators (Nguyen et al., 2025; Srivastava et al., 2023).

No dictionary currently exists to capture the strategic emphasis associated with OE 5.0. Consequently, a three-stage procedure, consistent with prior research (e.g. Gensler et al., 2024; Homburg et al., 2020), was followed:

Firstly, construct definitions based on statements from the official United Nations definitions of SDG 8 (Decent work and economic growth), SDG 9 (Industry, innovation and infrastructure) and SDG 12 (Responsible consumption and production).

Secondly, an initial list of keywords and synonyms was generated from the definitions of the three SDGs and previous literature on Operational Excellence 5.0 (e.g. Breque et al., 2021; Brodny and Tutak, 2025). This step followed a top-down logic, whereby theoretically based terms were assigned to each SDG category, in accordance with established practices for dictionary development (Berger et al., 2020).

Thirdly, the refinement and validation process (inductive approach) was carried out. The initial dictionary was refined through an inductive review of the CSRD sections of the selected reports to identify context-specific terminology that had not been reflected in the deductive stage. Two OE 5.0 experts independently reviewed the final word lists to assess semantic relevance and confirm correct alignment with the corresponding SDGs. Discrepancies were discussed and resolved jointly (Berger et al., 2020).

The final list of keywords consisted of 90 words, divided into three groups of 30 words assigned to each SDG. Table 1 shows the complete dictionary.

Elasticsearch used the dictionary to analyse the content of the CSRD reports, acting as a search engine, locating, indexing and scoring the relevance of each word in the analysed text. The algorithm not only counts the keyword frequency, but also calculates scores for their relevance based on how often they are repeated within the document, measures their rarity (inverse document frequency) and normalises results relative to document length (Manning et al., 2008).

Because the raw result is an index without direct interpretation, the results were rescaled to facilitate comparison between companies. The scores for each SDG were standardised and converted to percentages, with values scaled relative to the maximum observed score, which was set to 100. An aggregate OE 5.0 index was then calculated as a weighted average of the three SDG dimensions.

The analysis shows that words associated with three SDGs appear in all company reports. Figures 1, 2 and 3 present the estimates generated by the Elasticsearch algorithm. A summary of the findings for each company follows.

Mutua Madrileña mentions labour rights in the context of investment exclusions and lists employees as a stakeholder group. Although the term dignity is absent, references to human rights and labour rights imply respect for this principle. It also cites sustainable development of society and economic growth within responsible investment (SDG 8). Regarding SDG 9, innovation is not explicitly cited, though the firm does refer to the improvement of financial products and integration of ESG criteria. However, terms such as resilience and technology were not detected. For SDG 12, while the circular economy is not mentioned, terms such as sustainability, sustainability risks and sustainable finance do appear, although waste is absent. Scores: SDG 8:60% (Industry, innovation and infrastructure), SDG 9:10% (Decent work and economic growth), SDG 12:70% (Responsible production and consumption); weighted average is 51.9%.

MAPFRE highlights SDG 8 through references to working conditions, gender pay gaps, and financial education. It also emphasises profitable growth and the forecast of economic growth. For SDG 9, the firm explicitly mentions innovation, digitalisation, artificial intelligence and organisational resilience. For SDG 12, although the circular economy is not directly cited, sustainability features prominently, with references to sustainable development, a Sustainability Plan, and responsible investment. Waste management is not explicitly noted. Scores: SDG 8:90%, SDG 9:80%, SDG 12:75%; weighted average is 94.4%.

Generali addresses SDG 8 by promoting talent development, diversity and inclusion, and by highlighting sustainable growth. For SDG 9, it mentions offering innovative and personalised solutions, digitalisation and resilience. For SDG 12, neither circular economy nor waste is mentioned, although there are frequent references to sustainability and green investments. Scores: 85% for SDG 8, 75% for SDG 9, 70% for SDG 12; and the weighted average is 88.9%.

Allianz Partners strongly emphasises SDG 8, highlighting a strong commitment to employment, training, diversity, human rights protection and growth. Concerning SDG 9, it refers to innovative and sustainable solutions, geolocation technologies, and organisational resilience. For SDG 12, it incorporates circular economy practices such as repair over replacement and recycling of unusable parts. Sustainability is also a central theme of the report. Scores: SDG 8:95%, SDG 9:85%, SDG 12:80% and the weighted average is 100%.

AXA Insurance mentions SDG 8 in the context of employee volunteering, equal opportunities, and social development. For SDG 9, it discusses transformation and digitalisation, implying innovation, but technological innovation is not directly mentioned. For SDG 12, it notes practices such as repairing car parts rather than replacing, and green investments, although waste management is not explicitly mentioned. Scores: SDG 8:70%, SDG 9:60%, SDG 12:80%, and the weighted average is 77.8%.

VidaCaixa addresses SDG 8 by referencing labour rights, ethical culture and long-term sustainable growth. For SDG 9, it highlights innovative solutions for designing products that adapt to consumer needs, but neither resilience nor technological innovations are mentioned. For SDG 12, the circular economy is not clearly mentioned, but sustainability is presented as a strategic pillar, with emphasis on responsible investment and ESG criteria. Scores: SDG 8:75%, SDG 9:40%, SDG 12:70% and the weighted average is 70.4%.

Overall, the results indicate varying emphases in how the six insurers communicate OE 5.0 policies. Allinaz Parthers shows the strongest alignment, particularly on SDGs 8 and 12. MAPFRE and Generali also perform well, while VidaCaixa and Mutua Madrileña demonstrate relatively weaker emphasis, especially on SDG 9.

The findings of this study deepen the understanding of OE 5.0 by proposing and empirically illustrating a procedure for analysing how organisations communicate their strategic priorities in sustainability reports, alongside accounting-based information. Although the Elasticsearch algorithm does not allow for the evaluation of operational implementation, it does collect valuable information about the discourse used to explain it in the CSRD.

A first notable finding is the degree of heterogeneity in the importance that each insurer attributes to the different SDGs analysed, despite being subject to a common regulatory framework. This is consistent with institutional theory (DiMaggio and Powell, 1983; Lammers and Barbour, 2006), which suggests that regulation is unable to produce uniform organisational responses, but instead enables different interpretations and strategic positions. Some companies, such as Allianz Partners, opt for broad descriptions that encompass the three dimensions of OE 5.0 (innovation, resilience and circular economy practices), while others, such as Mútua Madrileña, focus their narratives on more specific aspects, including decent work and responsible production and consumption, over others, such as industry, innovation and infrastructure. Differences in how companies frame their arguments around OE 5.0 objectives and the strategies they pursue can also be interpreted from the resource-based view (Barney, 1991) and dynamic capabilities theory (Teece et al., 1997; Eisenhardt and Martin, 2017) as reflecting distinct strategic options for achieving sustainable competitive advantage.

A second contribution highlights the value of automated text analysis for researching operational excellence in organisations. Traditional Lean Six Sigma studies have mainly been based on surveys, case studies or the estimation of performance indicators (Xiang et al., 2023). These procedures are often costly, highly specific to certain contexts, and difficult to replicate (Balata and Breton, 2005; Nguyen et al., 2025). In contrast, the dictionary-based automatic text analysis approach using the Elasticsearch algorithm allows for systematic, transparent and replicable analysis of large volumes of unstructured textual data (Gormley and Tong, 2015). This approach is particularly suitable for analysing texts that are organised in a standardised way, such as CSRD reports, which enable comparative studies of the degree of compliance with objectives vis-à-vis competitors, but also function as an internal control mechanism, by comparing the discrepancy between what has been achieved and what has been reported, as well as enabling the traceability of the origin of the discrepancies (Gensler et al., 2024; Liu and Jia, 2025). This contrasts with the information provided by economic and financial data, which, although quantitative and more specific, has little capacity to describe intentions or future strategic options (Balata and Breton, 2005; Nguyen et al., 2025).

This study also highlights the role that information published in the CSRD can play in shaping a company’s image, as it can act not only as a passive regulatory requirement but also as a strategic communication mechanism that can strengthen the firm’s legitimacy with key stakeholders. Furthermore, the degree of transparency with which reports are presented can also help to reduce reputational risks, as statements of commitment to sustainability are often subject to intense scrutiny, verification and public attention (Nguyen et al., 2025).

Although it has already been noted that the insurance sector is characterised by a relatively simple operating environment compared to sectors such as manufacturing or logistics (Roth and Menor, 2003), this relative simplicity and sector homogeneity make it particularly suitable for initial testing of new analytical tools, such as Elasticsearch algorithms. The literature on the application of new measurement instruments recommends testing first in less complex contexts to reduce confounding effects and isolate tool performance, before extending the analysis to more operationally complex environments (Churchill, 1979).

In summary, the findings suggest that OE 5.0 should not only be understood as an operational strategy but also as a communication tool. By examining how companies articulate sustainability priorities in relation to OE 5.0 within CSRD reports, this study complements the traditional stream of research on Lean Six Sigma.

This study set out to explore how the emerging OE 5.0 paradigm can be examined through regulated sustainability disclosures, proposing a novel analytical approach based on the automated analysis of published text related to its application. By applying a dictionary-based Elasticsearch algorithm to CSRD reports, the study demonstrates that it is possible to operationalise and systematise the strategic emphasis communicated about the execution of OE 5.0-related ESG dimensions in a transparent and replicable manner (Gormley and Tong, 2015).

The main contribution lies in the study’s methodological and conceptual advancement, rather than in empirical generalisation. Conceptually, it positions OE 5.0 as a strategic construct that expands the foundations of Lean Six Sigma by integrating sustainability, resilience and a human-centred approach into operational excellence thinking. Methodologically, it introduces a structured way to leverage mandatory sustainability information as a data source for operational and strategic research, responding to the growing availability of regulated non-financial information in the European context. Importantly, the study demonstrates that although sustainability reports are often viewed as regulatory compliance tools, they can be repurposed as valuable analytical inputs for examining strategic orientation. This helps bridge the gap between research on operational excellence and research on sustainability reporting, two streams that have largely evolved in parallel (Gensler et al., 2024; Liu and Jia, 2025). The proposed framework is particularly promising as a diagnostic and exploratory tool that can support benchmarking exercises, comparative analyses and the initial assessment of strategic priorities.

From a practical standpoint, the findings suggest that organisations should consider CSRD disclosures not only as regulatory obligations but as strategic communication mechanisms that convey their intended trajectory with respect to advanced operational paradigms. For regulators and policymakers, the study highlights the inherent analytical potential of standardised reporting frameworks, which can facilitate more informed oversight and policy evaluation without imposing additional reporting burdens.

Overall, this research contributes to the evolving debate on how operational excellence is conceptualised, communicated and evaluated in the context of Industry 5.0. By offering a transferable analytical framework rather than definitive empirical claims, it lays the foundations for future studies seeking to expand, validate or refine the measurement of OE 5.0 across different sectors, countries and methodological approaches.

Despite its contributions, this study has several limitations that should be acknowledged and point to avenues for future research.

First, the scope of the empirical study is limited, as it focuses on a sample of Spanish insurers that publish annual reports online. While this choice ensured the sectoral and institutional homogeneity of the information, it also limits the generalisability of the results to other industries. Therefore, the contribution lies less in the selected sector than in the procedure used for its analysis, i.e. the methodology. Future research could replicate the analysis in other sectors, particularly manufacturing, healthcare or energy, where OE 5.0 is more crucial to the operational process, and narratives on sustainable implications could be much more closely linked in operational terms.

Second, the analysis is based exclusively on corporate information published in CSRD reports and is therefore more reflective of the proposals and strategies communicated than those actually implemented. Although this limitation has already been considered in previous literature, where it has been pointed out that reports can sometimes prioritise the generation of positive impressions based on symbolic rather than actual compliance, these narratives have also been considered significant indicators of organisations’ strategic intent (Gensler et al., 2024). Therefore, future research should triangulate text analysis with other indicators, such as financial performance metrics, sustainability indices, or interviews with managers and investors.

Third, the dictionary-based text analysis approach, while transparent and replicable, has methodological limitations due to its tendency to overlook contextual subtleties or semantic nuances. Furthermore, although involving experts in the construction of the dictionary can help reduce potential biases, this does not eliminate subjectivity. Recent advances in natural language processing methods, such as transformer-based models (e.g. BERT, FinBERT) or neural topic modelling, offer promising alternatives for capturing more complex semantic structures in greater depth (Grootendorst, 2022). Future research could compare diagnoses generated by dictionary-based programmes with models based on transformers or neural topic networks. It could also explore hybrid designs with multiple programmes that combine interpretability with semantic richness.

Fourthly, the study does not include indicators to estimate the model’s predictive power or external validity. While conceptual validation has been argued through a theoretical basis and the dictionary was reviewed by experts, future research could strengthen explanatory power by using performance indicators. This would allow for comparisons between methods, longitudinal analyses and replication across countries.

Finally, the analysis covers only two years of reporting (2023–2024), which limits understanding of the temporal dynamics of OE 5.0 adoption. Longitudinal studies are needed to examine how sustainability narratives evolve over time and whether emphasis translates into organisational change.