Audencia Business School

This study examines how consumers from individualistic and collectivistic consumer markets form recovery expectations and loyalty intentions in response to service recovery promises and bystander comments on social media. Across two experiments, we find that complainants’ origin fundamentally shapes how they interpret recovery efforts. Individualistic consumers exposed to non-accommodative recovery promises respond more favorably when accompanied by positive consumer comments. In contrast, collectivistic consumers report higher recovery expectations and intentions when the same promises are paired with negative comments. Importantly, bystander comments exert greater influence when companies issue non-accommodative rather than accommodative responses across both markets. This research is the first to reveal that complainant origin (culture) and bystander presence jointly moderate the effects of recovery communications. The findings challenge uniform global customer behavior assumptions and offer actionable insights for managing service failures in global markets. The study advances online service recovery theory.

This paper explores how the legacy of state-owned enterprises (SOEs) can simultaneously offer advantages and impose constraints, shaping a firm's strategic development over time. Using the transition from Communism in Central and Eastern Europe as a backdrop, we examine the case of Sberbank, a Russian SOE. In particular, we show how legacy artifacts rooted in the firm's historical circumstances and experiences shaped Sberbank's approach to the present and the future, operating as both a positive inheritance and a negative burden. As a positive inheritance, Sberbank's SOE legacy provided significant market dominance inherited from its historical role under state ownership, giving it a competitive edge in the post-Communist era. However, persistent interference and strong entanglements with the state constrained Sberbank's strategic autonomy substantially. Drawing on interviews, archival materials, and a comprehensive database of corporate communications, we show how Sberbank's management engaged with this challenging legacy. Their main strategy was to incorporate new initiatives that allowed them to decouple and address the negative aspects of state involvement while capitalizing on the market advantages inherited from their past. This study highlights how the concept of legacy is central to understanding how non-family businesses, and especially SOEs, navigate the tension between inherited strengths and constraints in their strategic evolution.

Sentiment analysis has become vital for understanding consumer attitudes, guiding product development, and informing strategic decisions. Although LLMs such as GPT-3.5 and GPT-4 deliver strong zero-shot performance, they can be cost prohibitive and raise privacy concerns. In contrast, Small Language Models (SLMs) provide lighter and more deployable solution, but their ability to match LLM accuracy, especially in zero-shot scenarios, remains underexplored. In this experimental study, we examine whether ensembles of zero-shot SLMs can serve as a viable alternative to proprietary LLMs in sentiment classification tasks. We investigate five commonly used SLMs (Phi2 Mini, Mistral, Llama, Gemma, Aya) and compare them to GPT-based models (GPT-3.5, GPT, 4, GPT-4 omni, GPT, 4, omni mini) across seven English-language datasets. By automating prompt generation and filtering responses based on a strict output format, we maintain a purely zero-shot approach. We form SLM ensembles via majority voting and evaluate their performance on accuracy, weighted precision, and weighted F1. We also measure inference time to assess cost and scalability trade-offs. Results show that SLM ensembles as a form of decision fusion, consistently outperform single SLMs, significantly boosting metrics in zero-shot settings. In contrast with GPT models, the ensemble achieves accuracy comparable to GPT-3.5 and even rivals GPT-4 on certain prompts. However, GPT, 4, retains a slight edge in both precision and F1 score. Moreover, local SLM ensembles incur higher latency yet offer potential advantages in data privacy and operational control. This experimental study's findings illuminate the feasibility of employing lightweight, zero-shot SLM ensembles for sentiment analysis, providing organizations with an effective and more flexible alternative to exclusively relying on large proprietary models.

While literature reviews are now well established in purchasing and supply chain management, there is ongoing debate about how they contribute to theory, which is the core aim of this special issue. In this editorial, building on the methods and content of the articles included in the special issue, we reflect on this along two dimensions: (1) the theoretical stance taken shows that different tools can be applied in systematic literature reviews. This is particularly true in data analysis, which can be positioned along an inductive-to-deductive dimension. We reflect on how the papers in this special issue apply different logics of theory development, modification, refinement or extension. Almost obviously, there is no one best way to organize related arguments. (2) Turning to a content-based approach, the accepted papers are linked to the wider field that is in scope for the Journal of Purchasing and Supply Management. Here, four related clusters are present: (1) Individual Learning, Knowledge, and Behavior, (2) Power and Dependence, (3) Corporate Sustainability and Supply Chain Due Diligence, (4) Information and Signaling Mechanisms. The contributions include reflections on both established and emerging topics in the field, which serve as a foundation for reflections on missing topics and suggestions for future research directions.