Addressing service grand challenges through inductive research Open Access

From the earliest days onward, researchers and practitioners have engaged with inductive research methods to investigate questions and challenges entailing complex social relations, ranging from the important to the wicked. A recent systematic bibliometric search identified 318 articles employing qualitative data published in three major service research journals (Valtakoski and Glaa, 2024). This rich literature arose from a long history of inductive service research.

Based on her observations as a professional banker, Lynn Shostack pioneered the idea of differentiating services from products by identifying some key dimensions of the former and introducing service marketing as a new domain. Her 1977 paper, published in the Journal of Marketing (Shostack, 1977), has garnered over 4,400 citations. Building on this conceptual breakthrough, she also wrote one of the first service design papers directed to practitioners (Shostack, 1982), which has been cited over 2,200 times. Inductive research is still fueling design in the digital age (Tuunanen et al., 2024). Finally, based on family circumstances, Shostack was also an early promoter of the application of service insights to healthcare, inspiring Len Berry's (Berry and Seltman, 2007) long engagement with this domain.

Moving further through the history of service marketing, based on inductive research, Bitner (1990) developed the concepts of the service encounter to explain the impact of service environments on customers' evaluations of their service experiences. This paper has generated over 10,000 citations. About the same time, Edvardsson and Mattsson (1993) called for inductive work to assess service experience. Following up with a paper conceptualizing and dimensionalizing service environments, Bitner (1992) introduced the servicescape concept in a paper that has inspired over 14,000 citations. Additional work with the doctoral students Kevin Gwinner and Dwayne Gremler (Gwinner et al., 1998) also elaborated the concept of relational benefits of service from inductive analyses. This paper has garnered 4,600 citations.

Building off Holbrook and Hirschman's (1982) concept of consumer experience, inductive research by Arnould and Price (1993) emphasized the concept of service experience. Twenty years on, Helkkula (2011) codified the term service experience, indicative of its conceptual robustness. Another notable feature of Arnould and Price's (1993) work was the use of inductive research to generate scale items that were then directly incorporated into quantitative measures of outcome variables of interest to marketers. This paper has garnered almost 5,000 citations.

Methodologically, inductive research in service has also proved fruitful. For example, the Critical Incident Technique (CIT) that has been used in a variety of service contexts to explore service failure and betrayal, has been instrumental in advancing our understanding of these issues Holloway and Beatty 2008). Bitner et al. 's (1990) foundational work on CIT has garnered almost 9,000 citations. A later article synthesizing the uses of CIT in service research has garnered almost 1,900 citations (Gremler, 2004).

To summarize, many central ideas and approaches to services research – including the service concept itself – arose through inductive research. It is not an exaggeration to add that these conceptual developments were responses to significant, even grand, theoretical and practical challenges in service delivery. A non-exhaustive list of constructs developed through inductive methods could include service design, servicescape, service encounter, service experience, commercial friendships, critical incidents, customer journey, service recovery and emotions and affect in service. Inductive research has also been instrumental in understanding services in vulnerable, marginal and sequestered populations, service relationships with humans as well as robots, and the cultural relativism of service provision and experience. Table 1 summarizes some of the contributions of inductive research to grand challenges in the services field.

As just elaborated, services research has been more open to inductive approaches employing qualitative data than many domains of marketing research, and these methodologies have been foundational to the history of services research. Readers who have taught services management or marketing might posit that the services field is open to qualitative approaches because the essential characteristics of services – intangibility, simultaneity of production and consumption, largely experiential, dependent on credence properties and co-create value – make the process of understanding, managing and delivering services messy, dynamic, interactive and complex (Zeithaml et al., 2020). Since service experiences are specific to times, places and social relations and interactions, contexts are especially important (Akaka and Vargo, 2015; Akaka et al., 2013). Moreover, service researchers' substantive focus has motivated them to go into the field using inductive methods to understand what's going on in the context of real-world problems. Even if later followed up by other approaches, research often has its beginnings talking to service providers to understand their theories in use and observing and talking with consumers to understand their nuanced, contextualized experiences (Zeithaml et al., 2020).

Across marketing, there are more calls for research with ecological value (Van Heerde et al., 2021). In our view, service researchers have always been at the cutting edge in embracing messy, interactive, substantive problems. Services researchers are notable for their early engagement with grand challenges. For example, service researchers quickly catalyzed energy around transformative service research challenges in health care, financial services and social services (Anderson et al., 2013). Inductive methods are particularly well-suited to investigating the myriad grand challenges service researchers address (e.g. Azzari and Baker, 2020; Mende et al., 2025). Management scholars note that inductive methods excel in “situations for which there is limited theory and on problems without clear answers” (Eisenhardt et al., 2016, p. 1113; George et al., 2016). There are several notable recent examples of service papers that adeptly employ inductive methods and qualitative data to address grand challenges such as consumer vulnerability in hospice care, the burdens of negotiating type 2 diabetes in the health care system, and challenges to consumers' agency in residential care facilities (Anderson et al., 2025; Azzari et al., 2021; Sudbury-Riley et al., 2024). Notably, these papers are distinguished by the authors' shared interest in a substantive grand challenge rather than the authors' shared methodological expertise. In each case, the papers include authors with diverse methodological skills.

There are three features of inductive research involving qualitative data and interpretive analysis that make this approach particularly helpful for studying services. First, qualitative data are flexible. This means that the data can take many forms and perspectives. Services are inherently complex, multifaceted, dynamic interactions. A customer's or service provider's experience in any given interaction is shaped by a range of psychological, cultural, social, historical and environmental factors. Quantitative data may not have the flexibility to account for all the necessary perspectives, but qualitative data allow researchers to look at multiple factors from many viewpoints. In contrast, the holistic nature of qualitative data helps to keep a study grounded in the world of experience and sensitive to unexpected context effects, accounting for multiple influences and perspectives that shape services in specific contexts.

Second, inductive research designs are also inherently flexible. A researcher might plan to conduct 20 interviews with frontline employees but realize after the first two interviews that they are asking the wrong questions or that the employees cannot or will not answer these questions. Instead, or in addition to these interviews, the researcher may need to talk to customers or managers. The researcher might also need to observe actual behavior in its natural context. Qualitative projects typically follow an emergent design, which means that it is expected that each point of data collection will require new analysis and revised understandings (Glaser et al., 1968; Miles and Huberman, 1994). Initial data collection may provide information that opens new or unexpected paths of inquiry for a researcher to follow, leading to insights that could not have been anticipated or hypothesized. This is very helpful for building theory about service phenomena, particularly because dealing with complex service phenomena means that the most effective research questions or hypotheses might not be immediately apparent – even after an exhaustive review of the literature. Qualitative data collection procedures encourage researchers to learn and adapt their research as their understanding develops.

Third, inductive interpretive analysis ties insights to embodied human knowledge. Despite the proliferation of advanced quantitative methods for analyzing qualitative data, such as large language models (LLMs), the results of any analysis always require interpretation. Humans and machines can both identify patterns and connections in data, but humans have some advantages when it comes to explaining what these patterns or connections mean. Human interpretation is tied to a researcher's own embodied knowledge, which incorporates both a formal understanding of relevant literature and an informal (and often unconscious) understanding of relevant aspects of human experience. Human experiences are embodied phenomena, which people often understand tacitly but cannot articulate verbally. Even when people can articulate their service experiences, human researchers' empathic, relational qualities and interrogative skills enable them to form relationships with people and communities. Relationship skills help to establish the trust and rapport that enables open or introspective sharing.

Researchers often talk about human judgment and interpretation as “subjective,” which unfortunately carries a negative connotation. Our embodied, experiential knowledge can bias us in ways that can harm others and behavioral economic research demonstrates how heuristics and biases even lead people to act against their own interests (Thaler and Sunstein, 2008). However, this same subjectivity can become a superpower if understood and accounted for through rigorous interpretive methods. For example, an interview informant might say something that completely changes how an interviewer views the world. Lightbulbs are flashing in the researcher's head! But how do they know that the informant's statement is new and interesting? Perhaps it fills a gap in the literature, but not all gaps are important or interesting. More likely, it challenges assumptions that the researcher or even an entire field has about something (Davis, 1971). These insights come from juxtaposing the theoretical world of past literature and well-studied phenomena with the embodied world of experience (McCracken, 1988), which is accessible to humans who have lived it.

Qualitative data include any record or documentation of a phenomenon that is not reduced to numbers. Qualitative data can take many forms. For example, if a researcher was interested in studying how academics discuss research methods in the services field, they might come to a panel discussion at the Frontiers in Service Conference to collect qualitative data. Their data might have included recordings of the panel (audio or visual data), transcriptions of the conversation (textual data), notes taken about the presenters' body language and social dynamics (ethnographic fieldnotes), comments posted on social media about the panel (netnography) or a copy of the conference program (archival data). Any record or form of documentation can serve as qualitative data, so researchers need to focus on data that can illuminate important aspects of their research questions.

Probably the most common source of qualitative data in service research are interviews, which are typically transcribed and analyzed as text. An excellent primer, which is well-cited in marketing and consumer research, is Grant McCracken (1988) classic and very accessible book, The Long Interview. Interviews are a great way to dive deeply into the emotions and cultural meanings that shape service experiences (Arsel, 2017).

Fieldnotes are another excellent source of qualitative data, and they are typically written as part of an ethnographic study that draws on extensive participant observation. A book called Taking Ethnographic Fieldnotes by Emerson et al. (2011) is a great resource with practical advice for how to start and conduct an ethnography. Ethnographic fieldnotes provide a naturalistic perspective on service encounters and interactions, pointing to explanations of phenomena that informants cannot consciously recall or articulate in interviews (Keränen and Prior, 2020; Von Koskull, 2020). Fieldnotes are particularly well-suited for documenting non-verbal behaviors such as emotional responses or interpersonal dynamics, which remain difficult to catalogue and analyze through verbal or numerical data.

There are many other forms of qualitative data that can be helpful for service research. Netnography, a method pioneered by Kozinets (2015, 2023), has been used to examine how virtual communities and online interactions shape customer service experiences (Heinonen and Medberg, 2018). Video recordings can also provide helpful data. Llewellyn (2021) used video camera footage, for example, to analyze how service providers and customers use cultural knowledge to interact through body language. When deciding what kind of data to collect, ask questions like, “What do I need to observe? What are people unaware of in their behavior? Whose perspectives are relevant to the phenomenon? Where and how can I find these perspectives?”

Inductive research is an approach that involves generating new hypotheses or theories, typically from qualitative data. Unlike deductive research, which starts with a theory or hypothesis and then tests it through experimentation and observation, inductive research begins with observations and develops a theory by interpreting thematic patterns as the data analysis proceeds. The analysis typically proceeds iteratively, working back and forth between data and theory-based conjectures. Tacq (2007, p. 191) strongly states the contrast with deductive research:

What about generalization, a common concern for deductive researchers? Tacq (2007, p. 193) explains:

Further, as Tacq clarifies, inductive research does not treat “outliers,” that is, unusual cases, either as expressions of limitation when they are few (i.e. only X % of phenomena conform to the hypothesis with 95% confidence), or when they are abundant (i.e. constituting a rejection of hypothesized relationships). Instead, they call for enlargement of the theoretical framework. This leads to the development of more inclusive, more synthetic theories, which account for observed empirical variability.

In practice, much inductive work also spills over into abductive research practice (Janiszewski and Van Osselaer, 2022). Abductive research is a methodological approach that involves generating the most likely explanations and hypotheses for observed phenomena. Deductive reasoning tests a hypothesis and inductive reasoning builds theories from data, as exemplified by the widely known grounded theory technique (Charmaz, 2024). In contrast, abductive reasoning seeks out the best possible explanation from incomplete or limited observations. Abduction is always iterative, adjusting explanations as new data is collected and analyzed. In our experience of the review process for inductive studies, when reviewers express dissatisfaction with proposed theorizations and call for additional data, researchers often produce different theorizations that incorporate the new data. This is abduction in practice.

Inductive methods typically focus on exploring meanings, implicit patterns and relationships between constructs derived from qualitative data. Some examples of different inductive approaches to data analysis drawn from JOSM and JSR are summarized in Table 2. Interestingly, we found mentions more common than strict applications of these approaches.

Thematic Analysis: Although the identification of themes in data is foundational to virtually all inductive research, in reviewing the Journal of Service Management and Journal of Service Research, we found 49 and 22 papers, respectively, that made explicit mention of thematic analysis. The earliest mention in JOSM is Bettencourt and Gwinner (1996), the most recent, Mele et al. (2025). This method involves identifying and analyzing patterns or themes within qualitative data. Researchers closely examine texts or other data sources to discern recurring themes, which offer insights into the participants' experiences and perspectives.

Grounded Theory: Grounded theory is mentioned over a hundred times in Journal of Service Management and 29 times in the Journal of Service Research, although actual applications are rarer. As mentioned in the section on inductive methods, grounded theory is an inductive method aimed at generating theories grounded in the data itself. Researchers collect qualitative data and progressively develop concepts and theoretical frameworks through an iterative process of increasingly abstract coding, developing themes and comparing data with the emergent theory (Charmaz, 2024; Corbin and Strauss, 2014; Creswell and Creswell, 2018).

Content Analysis: Content analysis involves systematically coding and interpreting textual material to ascertain the presence of certain words, themes, or patterns. While content analysis can be both qualitative and quantitative, interpretive content analysis focuses on understanding meanings and contexts. There are dozens of examples in the major service journals. Beatty and co-authors (Beatty et al., 2016; Bugg Holloway and Beatty, 2008) have made effective use of content analysis combined with the critical incident technique to derive insight into drivers of (dis)satisfaction in off and online service encounters. The emergence of generative AI and LLMs and the proliferation of massive online sources of textual data have meant that quantitative methods have moved to the fore in content analysis. Thus, Hogreve and Beierlein (2023) explore the outcomes of health-care professionals' participation in a vendor-hosted online community by combining qualitative and quantitative data. To analyze the qualitative data, the authors introduce a natural language processing tool that inputs individual statements and uses a rule-sensitive data network to code content to provide a structural and meaningful organization of embedded knowledge. Nevertheless, some experts, as do we, emphasize the crucial intervention of human interpretive skills in developing trustworthy interpretations of contents (Epp and Humphreys, 2025).

Narrative Analysis: We found only one use of narrative analysis in JSR and four uses in JOSM. This method interprets stories or accounts from participants to understand how people construct meaning and identity. It involves analyzing the structure, content and themes of narratives to understand broader social phenomena. The service field is relatively open for the use of more classic narrative analyses based on literary theory (Stern, 1988).

Discourse Analysis: Often used in other subfields, we found just 6 mentions of discourse analysis across JSR and JOSM. Discourse analysis examines language use in texts or spoken words to discern how meaning is constructed and how social realities are produced. It explores power dynamics, ideologies and identities reflected in communication. Formal discourse analyses were not evident in the journals, but Gallan et al. (2024) certainly focused on a close reading of textual data to identify challenging organizational tensions in service firms.

Phenomenological Analysis: Phenomenological analysis seeks to understand and interpret lived, subjective experiences holistically, often referred to as the life world, from the perspective of the participants. It involves a detailed examination of personal narratives to uncover the essence of specific experiences. In service research, providers, customers or managers could all be potential sources of phenomenological research data. We identified 34 mentions of the term phenomenological in JSR and 5 in JOSM, but few applications of phenomenological analysis, as Thompson et al. (1989) delineate the method in consumer research. Because of the centrality of customer experience to service research, many authors have sought to combine essentially phenomenological perspectives with other theoretical perspectives to tackle the elusive concept of value cocreation. Thus, Kelleher et al. (2019) extend research on how customers, including both experienced and novice customers, coordinate with each other in value co-creation. Clearly, the full potential of phenomenological analysis has yet to be tapped in service research.

Case Study Analysis: Case study analysis is a technique often employed in organization studies. Case studies provide an in-depth examination of specific instances or contexts. It is an inductive research approach that provides the opportunity to directly examine phenomena in real-world situations and “natural” contexts. Researchers interpret events, processes or interactions within the case to uncover broader principles or theories. Eisenhardt and Graebner (2007) argue that case studies are particularly useful for generating new theories. By examining detailed cases, researchers can identify patterns, relationships and causal mechanisms that may not be apparent through other research methods (Eisenhardt and Graebner, 2007). We found relatively few examples in the service literature of applications, but more in the early volumes of the service journals as the field was codifying service dimensions and processes. Still, many studies call for more case study research.

Ethnography: If inductive phenomenological research focuses on the micro level, ethnographic research focuses on the meso or collective level. Arnould and Price (1993) may be the first ethnography in a service context, but subsequent ethnographic research is represented by at least half a dozen studies in JOSM and JSR. Ethnography is a family of research methods founded in anthropology, which involves immersing oneself in a culture or social setting to understand the lived experiences and perspectives of the people within that field setting through prolonged observation and participation. Ethnographic data includes many sources, such as interviews, observations, photographs and collected artifacts. Ethnography involves participant observation, with prolonged, active and meaningful or “observant” participation leading to greater access to and understanding of the backstage aspects of a social world (Moeran, 2007). Ethnographers strive to understand social phenomena as enacted within and molded by broader cultural contexts, an approach glossed as holism. They strive to explain social phenomena in terms of general theoretical categories of human organization. Classic ethnographic work often involves years of engagement with specific cultural contexts, while research in western cultures and business settings is often of shorter duration (Arnould and Wallendorf, 1994). Sometimes, research teams can provide cultural insight more efficiently than a lone researcher (Sherry, 2024). Ethnographic service research has produced diverse inductive insights, opening many lines for further research.

Netnography: We found a few mentions of netnography in the service literature, most of them in JOSM. Still, the technique has not yet been fully exploited by service researchers (Heinonen and Medberg, 2018). Netnography is an inductive research method that studies online communities and cultures by adapting traditional ethnographic techniques to the digital realm (Kozinets, 2023). The growth and pervasiveness of the Internet in daily lives have led to online forms of sociality that differ sufficiently from offline worlds that a uniquely tailored methodological approach is necessary to understand them (Kozinets, 2015). Netnography provides “a window into the rich communicative and symbolic world of people and groups as they use the Internet, the Web, and social media, leaving its traces and transmissions for us to discover and decode” (Kozinets, 2015, p. 80).

It might sound cliché, but whenever the “lived experience” of an “actor,” broadly defined, is important to understand, qualitative data and inductive analysis can be fruitful (Thompson et al., 1989). To bring ecological value to the study of a problem requires situating it in lived experience, including actors' personal meanings, feelings, behaviors in a particular context or event. Examples include the lived experience of trying to eliminate plastic bags, house the homeless, care for cancer patients or create community safety (e.g. Gonzalez-Arcos et al., 2021; Berry et al., 2017). As Dubin (1976, p. 18) long ago observed, “It is exceedingly difficult to say something meaningful about the real world without starting in the real-world.”

Lived experience in all its complexity can be studied at many units of analysis: the lived experience of an individual, a family, a home, a waste system, a brick-and-mortar retailer, a community, an ecosystem of any sort. For example, as service researchers, we might seek to understand collective lived experience in a therapeutic servicescape, on a family vacation, in a restaurant that functions as a third place, or at a ritual event following a community disaster (Higgins and Hamilton, 2019; Epp and Price, 2011; Rosenbaum, 2006; Weinberger and Wallendorf, 2012). There are many kinds of data and all kinds of ways of analyzing data, but we always start with qualitative data and inductive analysis. That's because many other analytical techniques are cross-sectional, decontextualized, where relations are narrowly constrained and agency and affordances are assumed. By contrast, lived experience is: (1) interactive – it happens in interaction with people, other actors, places, materials and ideas; (2) temporal – it unfolds, is remembered and misremembered, has a narrative shape, is dynamic and variable; (3) embodied – it is both observable and not observable, it is felt physically and emotionally; (4) agentic – it has causes and consequences that are intentional and unintentional; and (5) afforded – actions are enabled or foreclosed by the set of actors in place.

When thinking about starting a new project that involves qualitative data, we first try to clarify exactly what we're trying to find out and what context is accessible that will enable us to possibly uncover an answer. The unanswered question, often a broad one, almost always comes first. “Context” is a holistic environment within which phenomena are studied (Stremersch et al., 2023). Sometimes we choose a context that seems like it will provide extreme values on dimensions of interest, or a context that isolates a particular construct or process of interest (Price et al., 2024). Of course, sometimes a context just leaps out at a researcher, asking its own questions and demanding explanation.

While still a doctoral student, the third author started a project about repair. This project was inspired by a general interest in sustainability. But more specifically, the author had been frustrated by how much stuff they were wasting due to an inability to repair broken things or find products that were repairable. Conversations with future co-authors led to discussions about companies like Patagonia that were promoting repair services instead of only product replacement. We essentially said, “This seems important. But what's going on here?”

We had the idea of dropping into a shoe repair shop that happened to be down the street from the first author's apartment. A coauthor recommended visiting the shop and coming back to report what was seen and learned. After about an hour at this shop, we came back with more questions than answers! So, we asked what else we need to know, and we looked at existing literature to see what scholars already said about repair (which was not much, particularly in marketing and consumer research). Then, we tried to find a range of repair shops where we could observe and interact with the service providers and customers whose experiences could provide answers. The specific questions and focus evolved as we collected and analyzed new data and used different theories to shine light on the phenomenon of repair services. But the pattern was essentially the same: start with an unexplained phenomenon, see how people experience it in the real world, and then interpret those observations inductively, considering what we already know.

Beginning with where to collect data, it is not clear that there is a single strategy for identifying sites or participants. Going back to Shostack (1977), sometimes sites and participants present themselves through personal observation or interest. For example, a project on commercial friendship (Price and Arnould, 1999) arose from repeated but casual observation in a hair salon: service providers reported that customers often travelled long distances to maintain relationships with stylists. Sometimes researchers are invited by private sector partners or interest groups. One study on extraordinary service experiences was commissioned by the Colorado River Outfitters Association to advise them on marketing communications (Arnould and Price, 1993). Some more recent papers on service atmosphere grew out of career-long consulting relationships between Club Med and one of the researchers (e.g. Rokka et al., 2023). Papers on luxury were made possible by good relationships with alumni networks (e.g. Arnould and Dion, 2023). Obviously, social media, Reddit, Facebook and so forth are useful starting points for contacting people, organizations and communities of interest.

Because a lot of us work in business schools, it is natural to turn to the private sector to access research sites and participants. However, following the example of Hill (2002) above and Ozanne and Ozanne (2021), it would be useful to work with NGOs, municipalities, cooperatives and non-conventional organizations like transition towns, and what we have elsewhere called prefigurative experiments (Arnould and Helkkula, 2024) to address wicked global problems. With the advent of GDPR and other data protection measures, navigating ethical responsibilities toward participants/teachers and their data has become a non-trivial challenge for inductive researchers, given that the data we collect is often highly personal and time- and context-specific.

As for how much data is enough, we often talk about “theoretical saturation,” which to us means collecting data until you start seeing the same thing over and over. Essentially, at saturation, patterns present in new data do not substantially change your current interpretation of the dataset as a whole (Bowen, 2008; Suddaby, 2006). It is important to remember that the unit of analysis is the concept or conceptual framework, not the person, the group, or the case. In this regard, some aspects of the common concern about population generalization are not pertinent.

Classical ethnographers often undergo a kind of data crisis during extended fieldwork. After many months of data collection, ethnographers often experience a moment of clarity when they feel they have worked out the dynamics of the problem under investigation. This is typically followed shortly by the precise opposite conviction that they know absolutely nothing. This speaks to the open-ended nature of inductive analysis that confronts researchers with the infinitude of possible observational data points in social phenomena.

Nevertheless, several studies have attempted to quantify the number of interviews one needs to be confident in one's analysis in a depth interview-based study, making the point that from a quantitative point of view, it is really a matter of how much of the variance in the data one wishes or needs to capture. These studies suggest that from 10 to 30 interviews will likely capture 95 to 99% of the variance in the data (Griffin and Hauser, 1993; Rowlands et al., 2016).

Among inductive researchers, conventional heuristics regarding data saturation include the emergence of robust themes through iterative coding following conventions established in the literature (Charmaz, 2024; Saldaña, 2025) and reaching a point when added interview responses or observations become predictable or fail to yield discrepant results. Miles and Huberman (1994, pp. 245–87) provide a detailed discussion of strategies for drawing and verifying conclusions in qualitative research among which searching for discrepant evidence, triangulation across individuals, times, researchers and data gathering techniques, and “member checks” are useful.

Glaser (2001) renders saturation not as “seeing the same pattern over and over again;” rather, “it is the conceptualization of comparisons of these incidents which yield different properties of the pattern, until no new properties of the pattern emerge. This yields the conceptual density that when integrated into hypotheses make up the body of the generated … theory with theoretical completeness” (Glaser, 2001, p. 191). Low (2019, p. 137) underlines that Corbin and Strauss (2014) speak of robustness rather than saturation:

It is also important to ask whether your conceptual model or theoretical explanation makes “sense given prior research” (Low, 2019, p. 137). However, it is important to bear in mind that theoretical saturation is, in some sense, a red herring because it is always relative to the theoretical framework that often evolves in tandem with other phases of the research process. Thus, a data set adequate to offer a robust theoretical interpretation from one theoretical perspective may not be adequate to offer another interpretation from a different perspective.

We will never forget our first experiences of taking field notes and then trying to make sense of them. For the authors of this article, our approach to analysis is still “old school.” As a recent article on collaborating with AI in Consumer Culture Theory research observes, there are demonstrable reasons to believe old school is still best (Epp and Humphreys, 2025). After collecting a corpus of data, nothing compares with iteratively reading it and making copious notes. Across and between informants and field notes, the goal is not to flatten or smush consumers into categories or themes, as is often the goal in statistical data analysis, and the inclination of AI. Rather, the goal is to ensure that, as Howard Becker suggests, we don't press data into categories too soon, asking instead what category does this data point define? He elaborates: “The data I have here are the answer to a question. What question could I possibly be asking to which what I have written down in my notes is a reasonable answer?” (Becker, 2024, p. 121).

As an example, a current PhD student is working on a project that asks how we can keep products in useful circulation longer. The student is fascinated with the growth in luxury rental platforms and the reuse and repurposing of thrifted objects. This is an important problem since about 80% of recycled garments still end up in landfills. They collected a lot of data on wedding dresses as an extreme case of objects that resist useful circulation, at least in the USA. Brides pay an average of $2000 to purchase a dress and wear it once, then clean and carefully store them in their closet, or under their bed for a lifetime. Each brides' story is so particular, whether they rented, purchased new, thrifted, repurposed or made from scratch. Stories from intermediaries are also unique. Without losing the richness of any one story, collectively the stories weave together into an interpretable pattern that can be used to help rental platforms and second-hand retailers keep objects from getting stuck: stuck because a new owner doesn't want them or stuck because a current owner won't give them up.

It is also important to remember that the veracity of the interpretation is in the weaving together of various and contrasting perspectives and voices rather than necessarily triangulating around one viewpoint (Price and Arnould, 2003). Analysis is always about what Weick and Weick (1995) describe as telling a useful, trustworthy story that is generative and can explain and predict future behaviors and outcomes. There will likely be many interrogations of the data and numerous story versions before researchers land on just the right question that their evocative and engaging qualitative data can answer.

A long time ago, Sutton (1997) wrote about the virtues of closet qualitative research. Historically, including with scholarly leaders such as Robert Cialdini and Robert Sutton, but continuing with experimental powerhouses such as John Lynch, Vicki Morwitz, Joe Alba and others, we see calls for more inductive approaches to knowledge creation, recognizing the limitations of deductive approaches (e.g. Janiszewski and Van Osselaer, 2022). Paralleling this is a call for more contextual studies and empirics first research, including a CFP for a special issue of Journal of Marketing on Empirics first approaches (Golder et al., 2023, https://www.ama.org/2025/04/14/call-for-papers-journal-of-marketing-special-issue-on-empirics-first/). The idea of generating knowledge by first grounding yourself in the phenomena seems intuitive to us, but we continue to be surprised by colleagues and students who believe they can begin a study of a real-world problem just by reading the literature! There are lots of specific ways that the inclusion of inductive interpretations of qualitative data can benefit researchers (Fischer and Guzel, 2023; Epp and Otnes, 2021; Valtakoski, 2020; Valtakoski and Glaa, 2024). We will briefly discuss just two.

First, inductive research and qualitative data provide ecological value. In the context of services research, ecological value is the degree to which research reflects and is relevant to services as they exist and evolve among service stakeholders and service ecosystems (Van Heerde et al., 2021). Ecological value is broader than external validity and involves “infusing a real-world perspective into every stage of the research process,” (Van Heerde et al., 2021, p. 1). The use of inductive research and qualitative data facilitates ecological value by infusing research with a real-world perspective throughout the research process. For example, early in the process, qualitative data can help experimenters develop more realistic stimuli or set up face validity for experimental studies conducted in a campus laboratory with real-world responses to the phenomenon of interest. Inductive research and qualitative data can help address endogeneity issues associated with the analysis of large data sets by showing process and connecting the dots. A situated ethnography may often be more practical and compelling than a field study and sometimes can situate an important direct effect by demonstrating mediating processes. Qualitative data can make real the outcomes of direct effects demonstrated in structural models or experiments by providing specific contextualized lived experiences. For example, qualitative data can combine with experimental and field studies to help us understand the lived experience and stakeholder outcomes of discriminatory practices in financial services (Bone et al., 2014; Scott et al., 2024).

Second, inductive research and qualitative data help generate new theory. New theory can emerge by focusing on new contexts, under-represented actors, processes or interactions that would be hard to uncover in a survey or experiment where participants are randomly assigned and studied in brief temporal interludes. Situations that benefit from closely looking at the tails of any distribution or deviation from the mean, thrive on qualitative data that can uncover what happens at extremes. We could give many excellent examples. One ongoing research project we are part of provides an interesting illustration. It examines a small number of users who actively and vigorously repair bike share platforms. These users are relatively few but have an outsized positive impact on platform sustainability. Knowing what to look for based on small qualitative data helps us find platform maintenance in large, quantitative datasets of user trips. Next, by understanding the key role of these maintenance behaviors and how, when and why they happen, we can understand how platforms prompt and afford beneficial and detrimental user behavior. A recent digital historiography (Sibai et al., 2024) of an online electronic dance music community asks why online consumption communities sometimes become alarmingly hostile, toxic and otherwise verbally violent toward one another. Future experimental research and models could build on this to identify interventions and more directly map consequences. As another example, a recent ethnographic study examines the extreme case of storm chasing teams to inform management for a variety of temporally uncertain settings such as restaurants, entertainment products and seasonal jobs (Kent and Granqvist, 2025). Rich theoretical insights often derive from extreme cases (Wright et al., 2023). Moreover, a nuanced examination of textual data can reveal organic theory and theory in use. With an explosion in access to textual data and opportunities for nuanced analysis, we might similarly expect to see a resurgence in new organic theories in services management and marketing (Berger et al., 2020).

This is not merely a methodological issue but a discussion about divergent philosophies of science. Maxwell (2004, p. 7) argues that “meanings, beliefs, and volitional actions constitute processes that can't be converted to variables, even ‘intervening’ variables, without fundamentally concealing and misrepresenting the nature of this process.” Many of the studies mentioned above show the value of deep, contextual understandings in transformative service research contexts precisely.

Nevertheless, we have found that an inductive understanding of meanings and beliefs in a cultural context can enable the development of robust scales and judicious choice of existing measures if one wants to test relationships in a subsequent deductive phase of research. Moreover, concerns with the “ecological validity” of experimental research suggest that researchers may wish to investigate the how and why of the phenomena they observe in laboratory contexts in the field.

It is often stated that inductive research is not adapted to causal explanation. Some researchers contest this claim (Creswell and Creswell, 2018; Maxwell, 2013). For example, Maxwell (2004, p. 8) argues:

That is, if we look for structural dependencies among elements, for historical chains of relationships, and for antecedents and consequents within the system under analysis, inductive research can provide process theoretical explanations of cause – in other words, how or why something is happening (Giesler and Thompson 2016; Maxwell 2004).

Many scholars grapple with questions about whether and how to incorporate qualitative data and inductive or abductive analysis into their research. We have asked these questions during our careers as well. From our observations and experiences, we would recommend that scholars who are interested in qualitative research consider three questions:

First, what is your research question? This one probably seems obvious, but dig down to fundamental, underlying questions. Do you find yourself struggling to identify your research questions and develop hypotheses around an understudied phenomenon? As illustrated by many studies shown in the two tables, qualitative data, analyzed interpretively, can often help researchers to develop a well-grounded conceptual framework in an understudied area. Are the service interactions you are interested in particularly complex? Qualitative data can often capture services with more contextual nuance than quantitative data, which require the reduction of complex interactions to measurable indicators.

Second, what is your personal philosophy of science? Without diving too deeply into the weeds of different research paradigms, it is helpful to understand how you view the world and your relationship with it as a researcher (Hudson and Ozanne, 1988). If your goal as a scientist is to find one right or true answer, or if you want to develop universal laws that govern service relationships, you might struggle with the relativistic ambiguity of qualitative data collection and interpretation. If your perspective is that service interactions can hold multiple experiences, explanations or even realities, then you're philosophically equipped for the interpretive nature of qualitative data and inductive analysis (Berger and Luckmann, 2016).

Third, how will you get training and mentorship? This might be the most important question. We were incredibly fortunate to receive doctoral training and career mentorship from people and institutions who were leaders in sociocultural theorizing and qualitative research. Importantly, this included training and relationships with scholars outside of marketing departments. These opportunities created a strong foundation for us to learn and practice collecting and analyzing qualitative data. However, we recognize that our experiences of gaining access to these incredible mentors put us in a very fortunate minority. If this is not your situation, there is still hope! Look to adjacent fields for training, such as sociology, anthropology and other social science disciplines with a rich heritage of qualitative data and interpretive analysis. Participate in qualitative data collection and analysis workshops at conferences. The Consumer Culture Theory (CCT) community provides excellent support, including a biannual Qualitative Data Analysis Workshop held in conjunction with the CCT conference. At conferences, build friendships and start collaborations with colleagues and co-authors who can teach you and work with you. You do not have to do a qualitative dissertation to make this type of research a central part of your identity as a scholar after you graduate. Many leading qualitative researchers are testaments to this fact, but these scholars had support from friends and colleagues with experience to guide them through the process. Do not try to “wing it” or push forward alone. Just as you would struggle to design an effective experiment or develop an econometric model after reading just a couple of papers, you will struggle to effectively collect and analyze qualitative data without training and guidance. But help is available, and we have a welcoming community.

We concur wholeheartedly with the importance of tackling sustainability challenges (Helkkula and Arnould, 2022). The United Nations has identified 17 sustainable development goals (SDGs). Each one of these goals presents numerous grand challenges, and every one of these grand challenges has numerous services linked to it. Service researchers have barely scratched the surface of any of them. The immersive tools of inductive research are ideal for investigating these highly significant, complex, uncertain problems that offer no easy solutions. We would like to see more services research tackling all these grand challenges. Of course, that advice is far too broad to be actionable. But all these grand challenges take form in our everyday lives. As such, our advice on what to study next is always the same – follow your own burning questions. Focus your research energy on a thorny problem that you are grappling with in your life, that keeps you awake at night. Investigate an accessible service context that exemplifies why that problem feels so significant, uncertain and without easy solutions for you. Taking on a challenge that provides an opportunity to enhance your well-being and that of those around you with empirical answers to thorny questions is uniquely satisfying.

In addition, there are many topics around which we would like to see more service research. Not all are entirely novel. However, we would like to see more research on non-profit service systems which look at other types of relationships of resource circulation and value cocreation beyond those based on monetary exchange. Gifting in the Maussian sense (Weinberger et al., 2025) is clearly a part of many service interactions, and it seems reasonable to suppose that participation in non-profits depends upon the circulation of gifts in some guise.

Relatedly, we would also like to see more direct attention to the informal service provision of care, which is treated tangentially in some existing studies. However, the voluntary and cooperative provision of care is a huge part of the informal economy as well as an envisioned solution to the global crisis of care (Addati et al., 2018; ILO, 2024; Zigante, 2018). Service research should be more involved.

Additionally, we would like to see a focus on more than human perspectives on service provision and experience, perhaps starting with what are conventionally termed “service animals” (Moraes, 2024). This research can be extended, since the existing work on ecosystem services falls mostly outside of the service community and is also deeply anthropocentric. We think we need to consider more-than-human service interactions as part of a project to address the wicked global challenges of climate change and biodiversity loss, both of which are evidence of value co-destruction.

Technology transforms service interactions, and the proliferation of artificial intelligence is an obvious example of an emerging research area that will undoubtedly have substantial impacts. This is an important area where qualitative data and inductive analysis have so far been used less often than quantitative deductive research. However, researchers should look at customer and employee experiences along the full spectrum of technology mediation – from fully automated interactions to low- or no-tech interactions between humans and service environments. Even in the age of AI, customers are increasingly attracted to “old school” ways of interacting and doing business, as evidenced by the growth of tech-free retreats, no-phone policies in schools, and interest in “dumb phones” and unconnected devices such as typewriters, diaries and day planners. Technological revolutions are shaped not only by innovation, but also by resistance and reinvention. Service researchers can develop a deeper understanding of the culture of service technology by understanding both sides of this constant tug-of-war.

As service research and service experiences move in uncertain future directions, the importance of inductive methods and qualitative data will increase in parallel with the imperative to develop more sophisticated quantitative models and datasets. To keep service research grounded in and relevant to the lived experience of human (and even non-human) actors will require attention to modes of observing and documenting that reflect the nuances of those experiences holistically in both digital and analog worlds.

Inductive analysis allows service researchers to develop a theory that explains phenomena relevant to grand challenges, providing a conceptual foundation for future research to build upon. Very limited success in addressing the grand challenges identified in the UN sSDGs is an invitation to service researchers, who have made strides in addressing issues of vulnerability, to address the challenges identified in the SDGs with service solutions. We invite scholars, regardless of their research orientations, to reintroduce themselves to the contributions of inductive service research and consider how such approaches might help them better engage with grand challenges in services.