The research, led by Hung-Che Wu of Universitas Brawijaya together with Waseem Ul Hameed, Yi-Chang Chen, Raditha Hapsari, and Ananda Sabil Hussein, set out to solve a persistent measurement problem. As artificial intelligence has spread across retail, hotels, banks, and customer service, researchers have struggled to define and quantify what customers actually experience when they move seamlessly between digital and physical channels powered by AI, a setup known as omnichannel service. Without a validated scale, businesses have been essentially flying blind, unable to tell which elements of their AI systems build loyalty and which merely add friction.

To build their measurement instrument, the team followed the rigorous scale-development paradigm first established by marketing researcher Gilbert Churchill in 1979, a multi-stage process that involves generating candidate items from theory and prior literature, purifying them through successive statistical tests, and validating the final structure on independent samples. The researchers collected data from a convenience sample of 496 consumers at an intelligent unmanned store in Beijing, a setting chosen because it compresses the entire omnichannel journey, from app-based entry to computer-vision checkout, into a single visit. The project received ethics approval from the Ethics Committee of Universitas Brawijaya, and the authors report no conflicts of interest and no external funding.

The analytical backbone of the study was partial least squares structural equation modeling, or PLS-SEM, a statistical technique widely used when researchers need to test networks of relationships among abstract psychological constructs that cannot be observed directly. The team assessed the reliability and validity of their measurement scales using established benchmarks, including the Fornell-Larcker criterion and the heterotrait-monotrait ratio developed by Henseler and colleagues, and they checked for common method bias using the full collinearity assessment approach. Confirmatory composite analysis, a newer model-quality test championed by Hair and colleagues in 2020, was also part of the validation toolkit.

What emerged from this statistical machinery was a multidimensional picture of the omnichannel AI experience. The findings contribute to services marketing theory by clarifying the dimensional structure of four interlocking constructs: omnichannel AI experiences themselves, AI experiential relationship quality, AI experiential evaluation cost, and AI experiential engagement intentions. In plain language, the researchers argue that when customers interact with AI across channels, they form a quality judgment about their relationship with the technology, weigh the mental effort required to evaluate it, and only then decide how deeply they want to engage with it in the future.

The chain of relationships the data revealed is as elegant as it is practical. Omnichannel AI experiences were positively associated with AI experiential relationship quality, meaning that smoother, richer, more consistent AI-powered interactions make customers feel a stronger bond with the service. That relationship quality, in turn, positively influenced both AI experiential evaluation cost and AI experiential engagement intentions. The link between relationship quality and evaluation cost may seem counterintuitive at first glance, but it echoes a classic finding from social psychology: Aronson and Mills showed as far back as 1959 that people who invest effort in a relationship tend to value it more. When customers feel connected to an AI service, they appear willing to expend more cognitive effort evaluating and exploring it, treating that effort as an investment rather than a burden.

This effort-based interpretation draws on a long line of consumer research into switching costs and evaluation costs. Building on Hauser and Wernerfelt’s influential evaluation cost model of consideration sets from 1990, the study treats the mental energy customers spend comparing and judging AI services as a real psychological currency. The practical implication is striking: a customer who has built a strong experiential relationship with an AI system is more tolerant of the effort required to use it, and that tolerance feeds directly into stronger engagement intentions, the willingness to keep using, exploring, and advocating for the service.

For the booming unmanned retail industry, the study arrives at a critical moment. Market analysts project substantial global growth for fully automated and semi-automated stores over the coming decade, particularly in Asia where the format is most mature. Yet the sector’s biggest challenge is not technology but psychology: customers who feel frustrated, confused, or impersonally processed by machines simply do not return. The researchers’ results suggest that enhancing omnichannel AI experiences, strengthening AI experiential relationship quality, and effectively managing AI experiential evaluation cost are promising strategies for fostering stronger engagement in unmanned retail environments. In other words, the path to loyalty runs through making the AI feel like a reliable partner rather than a cold interface.

The study also connects to a broader scientific conversation about how humans relate to intelligent machines. Prior research has shown that anthropomorphism, empathy responses, and interaction quality shape trust in AI, and that AI service quality feeds into customer experience and brand relationships. By formalizing the omnichannel AI experience as a validated multidimensional construct and mapping its downstream consequences, the new work gives researchers a standardized tool for comparing findings across industries and cultures. It also extends the experiential relationship quality framework that Wu and collaborators have previously applied to smart hotels, driverless buses, and pet-friendly medical tourism, demonstrating its versatility in contexts where human staff are absent entirely.

Limitations remain, as they do in any single-sample study. The data came from consumers at one intelligent unmanned store in Beijing using a convenience sampling approach, so generalizing to other markets, store formats, or cultural settings will require further testing. The cross-sectional design captures intentions rather than observed long-term behavior. Still, the contribution is substantial: the scale offers retailers, technologists, and psychologists a shared language for measuring the invisible glue between shoppers and the machines that serve them. As unmanned stores multiply worldwide, the study’s central message is likely to resonate far beyond Beijing, that the future of retail loyalty will be decided not by the sophistication of artificial intelligence alone, but by the quality of the relationship customers feel they have built with it.

Subject of Research: Validation of a multidimensional scale measuring how omnichannel AI experiences influence customer relationship quality and engagement intentions in unmanned retail

Article Title: Omnichannel AI experience and engagement intentions: validation of a multidimensional scale

Article References: Wu, H.-C., Hameed, W. U., Chen, Y.-C., Hapsari, R., & Hussein, A. S. (2026). Omnichannel AI experience and engagement intentions: validation of a multidimensional scale. Current Psychology, 45(18), Article 1531. https://doi.org/10.1007/s12144-026-10069-w

Image Credits: AI Generated

DOI: 10.1007/s12144-026-10069-w

Keywords: omnichannel AI, artificial intelligence, unmanned stores, consumer behavior, customer engagement, relationship quality, retail technology, scale validation, services marketing, PLS-SEM, evaluation cost, Beijing