Introduction
Data collection represents a substantial investment for nonprofit organizations, e\ncompassing financial, operational, compliance, perceptual, and effectiveness data (Ahearn & Mai, Reference Ahearn and Mai2025; Benjamin et al., Reference Benjamin, Voida and Bopp2018). These datasets are critical for accountability, understood as the processes through which nonprofits provide accounts of their actions, decisions, performance, and impacts to those who review them (Ebrahim, Reference Ebrahim, Renz and Herman2016). This includes both obligatory accountability within principal–agent relationships, such as reporting to funders, and voluntary forms of accountability to broader stakeholders including service users, communities, donors, staff, partners, and the broader public (de Boer, Reference de Boer2023; Ebrahim, Reference Ebrahim, Renz and Herman2016). Subsequently, these datasets form the empirical foundation for much of the nonprofit research field (Ho et al., Reference Ho, Duffy and Benjamin2023). Yet, a key dimension often overlooked is the rich and dynamic origins of data well before it is entered into a spreadsheet and analyzed for accountability purposes: namely, the frontline context in which information is first exchanged between clients and social workers (Benjamin & Campbell, Reference Benjamin and Campbell2015; Bradt et al., Reference Bradt, Roose, Bouverne-De Bie and De Schryver2011).
Although data is critical for achieving accountability, collecting it is contentious for resource-strained frontline staff who understand its necessity for contractual compliance but also worry it may misrepresent their practice or disadvantage clients (Høybye-Mortensen & Ejbye-Ernst, Reference Høybye-Mortensen and Ejbye-Ernst2018; Mathys et al., Reference Mathys, Raeymaeckers, Suykens and Van Steenlandt2024). Managers and leaders, on the other hand, view data as an asset for planning, accountability, tendering, research, and advocacy (Adams & Simnett, Reference Adams and Simnett2011; Benjamin et al., Reference Benjamin, Voida and Bopp2018; Schoech, Reference Schoech2010). Funders, in turn, rely on data both to hold nonprofits accountable and to justify their own decisions to the public (Fyall, Reference Fyall2017). Such varied expectations of data are typical of the ecosystem in which nonprofit organizations operate (Lebec & Dudau, Reference Lebec and Dudau2024).
Although previous research has examined the varied expectations placed on data across frontline practice, organizational management, and funder accountability, less attention has been paid to how these expectations operate simultaneously through shared data collection processes (Benjamin & Campbell, Reference Benjamin and Campbell2015). Adopting institutional ethnography in a case of data collection and use in the homelessness sector, we aim to gain a better understanding of this ecosystem, operationalized as four processing interchanges.
Our analysis uncovers the dynamic ecosystem in which data circulates within social service nonprofits, along with the complexities this ecosystem imposes on how data is collected and used. We show that data functions not only as a central nervous system within this ecosystem, but also that the very nature of the data collected is itself shaped by these surrounding dynamics. Our findings therefore present a double-edged sword for the future use of data. On the one hand, we demonstrate its vital role and show how, through advances in interoperability and digital proficiency, nonprofits can achieve greater autonomy and ownership over their data, making it more useful to their core mission. On the other hand, we highlight how divergent stakeholder perceptions about why data is collected and how it should be used affect the character of the data produced, thereby threatening its validity. Importantly, this ecosystem and its associated pressures are not new; rather, what this ethnography enables is a close-up view of the ongoing push and pull of these expectations as they play out in real time.
We begin with an outline of key terms of interest in this study, including how we approach data as a social practice and its relation to positive social change.
Background
Nonprofit data
Among nonprofit research scholars, the term data carries various meanings (Ahearn & Mai, Reference Ahearn and Mai2025; Widyawati, Reference Widyawati2020). Nonprofit data serves both functional and strategic purposes, enabling organizations to pursue their social mission within inherently resource-constrained environments (Barman, Reference Barman2007; Benjamin et al., Reference Benjamin, Voida and Bopp2018; Dahler-Larsen, Reference Dahler-Larsen2014). The ways these data are collected and used are a central focus of research inquiry (Ho et al., Reference Ho, Duffy and Benjamin2023).
Data collected by nonprofits typically serves five purposes: financial, operational, compliance, perceptual, and effectiveness (Ahearn & Mai, Reference Ahearn and Mai2025; Benjamin et al., Reference Benjamin, Voida and Bopp2018). Different domains of nonprofit management and the associated research fields rely on different types of data. For example, accounting focuses on financial data, human resources on operational data, fundraising on perceptual data, and evaluation and performance on effectiveness data (Ahearn & Mai, Reference Ahearn and Mai2025; Moxham, Reference Moxham2014). Yet, these domains rarely operate in isolation. For instance, financial data without reference to operational outgoings or demonstrated effectiveness has limited utility (Adams & Simnett, Reference Adams and Simnett2011). Equally, perceptual data is often shaped by public statements of accountability and compliance, which are built upon effectiveness and financial data (Shukla et al., Reference Shukla, Geetika and Shukla2022).
The focus on funder reporting requirements as the primary driver of data collection (Benjamin et al., Reference Benjamin, Voida and Bopp2018; Carnochan et al., Reference Carnochan, Samples, Myers and Austin2014; Cordery et al., Reference Cordery, Goncharenko, Polzer, McConville and Belal2023; Ahearn & Parsell, Reference Ahearn and Parsell2025; Carman, Reference Carman2011; Cordery & Sinclair, Reference Cordery and Sinclair2013) is increasingly expanded to include research on alternative motivations (Ebrahim, Reference Ebrahim2005; Lebec & Dudau, Reference Lebec and Dudau2024; Mathys et al., Reference Mathys, Raeymaeckers, Suykens and Van Steenlandt2024). Nonprofits are not only accountable upward to funders, but also downward to clients and communities, and internally to their staff and decision-makers (Ebrahim, Reference Ebrahim2005). Multiple accountabilities explain the breadth of data required for nonprofits to pursue strategic purposes. As Lebec and Dudau (Reference Lebec and Dudau2024) note, nonprofit accountability research has often imported theories and models from business research, bringing with them dyadic assumptions that do not fully align with the broader accountability arrangements in which nonprofits operate. This is particularly important for understanding nonprofit data, which is rarely produced for a single audience or purpose, but expected to support accountability, service delivery, organizational decision-making, and sector-level advocacy at once (Benjamin et al., Reference Benjamin, Voida and Bopp2018; Cordery et al., Reference Cordery, Goncharenko, Polzer, McConville and Belal2023; Ebrahim, Reference Ebrahim2005; Lebec & Dudau, Reference Lebec and Dudau2024; Schoech, Reference Schoech2010).
Data as practice in social service organizations
Data is not a static product solely to be funneled upwards to funders as evidence of effectiveness. How data are collected, analyzed, and reported depends on which stakeholders are involved and the requirements that shape how they interact with and value data (Benjamin & Campbell, Reference Benjamin and Campbell2015; Mathys et al., Reference Mathys, Raeymaeckers, Suykens and Van Steenlandt2024).
The intake and assessment stage of social work practice is often the first point of data collection, and in most cases, the first interaction between social worker and client (Scherz, Reference Scherz1952). This stage demands substantial training and expertise in identifying the types of information required to support clients within a given context, and knowledge of what supports are available and effective (Loewestein, Reference Loewestein1974). Most importantly, it requires the ability to build trust and rapport while progressing information gathering and action planning (Benjamin & Campbell, Reference Benjamin and Campbell2015; Bradt et al., Reference Bradt, Roose, Bouverne-De Bie and De Schryver2011).
Collecting standardized and routine data through digital systems has become increasingly common in social work practice; however, standardized data collection tools are also highly contested (Høybye-Mortensen & Ejbye-Ernst, Reference Høybye-Mortensen and Ejbye-Ernst2018; Fritsch et al., Reference Fritsch, Hiler, Mueller, Wu and Wustmann2017; Zuboff, Reference Zuboff2019). While some argue that the use of standardized measurement tools to inform assessments has improved the professionalism of the field (Bradt et al., Reference Bradt, Roose, Bouverne-De Bie and De Schryver2011; Gillingham & Graham, Reference Gillingham and Graham2017; Høybye-Mortensen & Ejbye-Ernst, Reference Høybye-Mortensen and Ejbye-Ernst2018), others regard them as mandated, top-down performance measurement instruments. Standardized metrics are completed by workers to demonstrate client need and eligibility, as well as trace the effectiveness of their performance and that of their program (Ebrahim, Reference Ebrahim, Renz and Herman2016; Mathys et al., Reference Mathys, Raeymaeckers, Suykens and Van Steenlandt2024). These measures are criticized as overly prescriptive, narrowing down frontline workers’ discretion in practice, and reinforcing managerial and bureaucratic approaches to service delivery (Bradt et al., Reference Bradt, Roose, Bouverne-De Bie and De Schryver2011; Høybye-Mortensen & Ejbye-Ernst, Reference Høybye-Mortensen and Ejbye-Ernst2018).
Within the homelessness sector, top-down measurement is especially fraught. As Pawson et al. (Reference Pawson, Parsell, Clarke, Moore, Hartley, Aminpour and Eagles2024, p. 11) observe, the homelessness sector is “funded by government at a level inadequate to keep pace with rising demand, yet also officially assessed on whether this demand has been met.” They argue that this creates a vexed position for frontline workers and risks alienating them from the very systems intended to support improvement. Identifying measures that can meaningfully capture this complex, taxing, and often structurally constrained work is especially contentious when housing outcomes are limited more by the absence of available housing than frontline service delivery (Stambe et al., Reference Stambe, Plage, Parsell, Kuskoff and Wagland2026).
Recent research suggests that frontline practitioners are not powerless to these pressures (Bakko, Reference Bakko2026). Mathys et al. (Reference Mathys, Raeymaeckers, Suykens and Van Steenlandt2024), demonstrate how social workers adapt funder-mandated performance indicators in their everyday practice. Their study shows that both social workers and clients exert influence over the data collection process, such that top-down indicators become interpreted and enacted in ways that reflect professional judgment and practice realities. The discretionary space enacted by frontline social workers in support of their clients is not a new observation; however, its interaction with data collection practices introduces additional complexity to the use of data. Contradictory expectations about what a measurement tool is designed to do often result in inaccuracies in the data collected (Dahler-Larsen, Reference Dahler-Larsen2014).
In the case of state-funded social services aimed at alleviating or ending homelessness, the discretionary space within measurement indicators is particularly complex (Morris et al., Reference Morris, Clarke, Robinson, Idle and Parsell2023). Clients of these services often face the most challenging manifestations of vulnerability and, when presenting to support services, are frequently overtired and cognitively diminished by the act of surviving (Parsell & Clarke, Reference Parsell and Clarke2022). In such a high-stakes context, social desirability bias in the data collected is a valid concern. Typically, social desirability bias suggests that individuals try to portray themselves in a more positive light (Kithulgoda et al., Reference Kithulgoda, Vaithianathan and Parsell2024). However, in some examples of homelessness services, both clients and support workers recognize that presenting an opposite image, that of extreme vulnerability, may be more advantageous. This ‘race to the bottom’ occurs when clients and their advocates strategically emphasize their extreme physical, emotional, and financial precarity to secure priority access to scarce housing resources (Morris et al., Reference Morris, Clarke, Robinson, Idle and Parsell2023). Data collection thus becomes embedded within assumptions of who is and who is not deserving.
Thus far, we have outlined the potentials and challenges of data in relation to practice, rapport-building, and case advocacy for individuals and families. Yet a defining feature of social work practice is its operation across multiple levels starting with individuals, through to broader political and legislative advocacy aimed at systemic change (Fyall, Reference Fyall2017; Lens, Reference Lens2004). More recently, the role of data in this wider arena has been emphasized, particularly through the aggregation of evidence to demonstrate consistent patterns of systemic failure (Parsell et al., Reference Parsell, Kuskoff and Ahearn2025). Data-informed approaches to social change, for example, cost-benefit analyses, can be especially persuasive for audiences who may not be swayed by social justice arguments alone (Gillingham & Graham, Reference Gillingham and Graham2017). Taken together, this highlights that nonprofit data collection is never a neutral activity. Through the discretion of managers and frontline staff, it actively shapes policy outcomes, underscoring the central insight of the advocate–provider framework (Fyall, Reference Fyall2017).
We engage with these complexities to address the research question: How does frontline social work practice generate and shape data within a centralized homelessness data system, and how is this data subsequently mobilized for accountability, collaboration, and organizational processes across the nonprofit ecosystem?
Method
Research context
This study is an institutional ethnographic case study of the Australian Homelessness Vulnerability Triage Tool (AHVTT) as it was applied within a large Brisbane-based homelessness organization, the broader Brisbane Advance to Zero (AtoZ) movement, and the Australian Alliance to End Homelessness (AAEH). The AHVTT is used by frontline social workers at intake to identify client vulnerabilities (Australian Alliance to End Homelessness, 2024). The resulting data informs practitioner assessment, management decisions, service coordination, and AtoZ advocacy.
Unlike clinical assessment tools, which often seek to provide a detailed and highly sensitive assessment of a specific vulnerability domain, the AHVTT uses a brief screening format across multiple domains through 57 questions. It typically asks one to two questions about each domain, with yes/no/skip response options. For example, respondents may be asked: “Have you experienced violence in the last year?” or “Do you have a diagnosed mental health condition?” The main exception to this format is a set of service-use questions, which ask, for example, how many times a person has been transported to hospital by ambulance in the past year.
Together, the AHVTT and the AtoZ Digital Database function as representations of individuals and families experiencing homelessness, enabling their circumstances to travel through the housing ecosystem. Affiliated service providers access and update these records in real time, including AHVTT data. In most cases, AHVTT data is entered once at intake, or during a later follow-up to ensure that all data and required forms are complete. The tool may be readministered where there has been a major life change, a long period out of homelessness followed by re-entry, presentation to a different support agency, or a new health diagnosis.
AHVTT data therefore functions as operational and effectiveness data: data nonprofits use to measure, evidence, refine, and articulate their work. By focusing on AHVTT data, we are particularly interested in how these data, generated through frontline encounters between workers and service beneficiaries, become available for multiple and sometimes competing purposes, including service coordination, organizational accountability, performance measurement, advocacy, and sector-level learning.
Institutional ethnography
Institutional ethnography examines how institutional work is enacted and coordinated through text-based knowledge such as documents, records, and other written processes; the design of written texts or data may matter greatly, including what is included or omitted, how information is framed, and how it is aggregated or passed on (Smith, Reference Smith2005, Reference Smith2006). These features sustain institutional norms and logics (DiMaggio & Powell, Reference DiMaggio and Powell1983) and allow such norms to spread seamlessly through organizations without individuals needing to consciously enact them (Zimmerman & Zeitz, Reference Zimmerman and Zeitz2002).
Institutional ethnographers pay particular attention to processing interchanges that are “organisational occasions of action in which one practitioner receives from another a document pertaining to a case and then makes something of the document, does something to it, and forwards it” (Smith, Reference Smith2006, p. 30). Here, we attend to texts created about a person experiencing homelessness that move through the various institutional stages and requirements while implementing support (DeVault & McCoy, Reference DeVault, McCoy and Smith2006).
Fieldwork and data collection
The ethnography was conducted over 13 months across three service settings: a street-front crisis hub, short-term emergency accommodation centers, and office-based organizational and sector meetings. Table 1 provides an overview of the forms of data collected at each setting, their volume, and their analytic purpose.
Qualitative corpus summary
Table 1. Long description
The table consists of four columns: Data source, Site/participants, Volume, and Analytic purpose.
* Row 1: Observation of street-front crisis center. Participants: People experiencing homelessness and frontline workers. Volume: 110 h. Purpose: Examined how A H V T T was introduced, explained, completed, adapted, and interpreted in frontline encounters.
* Row 2: Observation of outreach shadowing. Participants: Outreach settings with frontline workers. Volume: 36 h. Purpose: Traced how A H V T T-informed practice extended beyond the street-front crisis hub into street-based and mobile service delivery.
* Row 3: Observation at advocacy meetings and sector events. Participants: Organizational leaders, sector actors, advocacy forums. Volume: 21 h. Purpose: Examined how A H V T T data was mobilized for advocacy, systems change, and sector-level accountability.
* Row 4: Observation at interagency service coordination meetings. Participants: Practitioners and representatives from collaborating organizations. Volume: 16 h. Purpose: Traced how A H V T T data moved across organizational boundaries and supported service coordination.
* Row 5: Observation at internal organizational meetings and training. Participants: Staff, managers, internal implementation settings. Volume: 36 h. Purpose: Examined organizational interpretation, training, implementation fidelity, and management use of A H V T T data.
* Row 6: Interviews with people experiencing homelessness. Participants: 17 service beneficiaries. Volume: 17 interviews. Purpose: Explored client perceptions of A H V T T and the experience of being documented through the tool.
* Row 7: Interviews with frontline practitioners. Participants: 9 practitioners. Volume: 9 interviews. Purpose: Examined how workers collected, interpreted, adapted, and acted on A H V T T data.
* Row 8: Interviews with managers. Participants: 6 managers. Volume: 6 interviews. Purpose: Explored how A H V T T data informed organizational decision-making, accountability, advocacy, and strategy.
Together, these settings enabled observation of the AHVTT across multiple points in the homelessness service ecosystem: direct intake, ongoing social work, interagency coordination, and advocacy. Data collection was immersive and iterative rather than sequential. Observations of AHVTT administration and everyday service provision, practitioner and management interviews, and observation of training, meetings, and other sector activities occurred concurrently throughout much of the fieldwork. Movement between these settings allowed insights emerging in one context to inform attention and interpretation in others.
Fieldnotes were kept throughout and supplemented by audio recordings of observations. In addition to observations, we conducted in-depth interviews with 17 participants experiencing homelessness, focusing on their perceptions of the AHVTT. We also interviewed nine frontline practitioners and six managers. Additionally, ongoing conversations with staff were constant and integral to the fieldwork. This combination captured the perspectives of those who collected, interpreted, and acted on AHVTT data, as well as those whose lives it documented.
As institutional ethnographers, our target was to identify and understand the processing interchanges of the AHVTT. Beginning at the street-front crisis hub, we followed the collection and subsequent movement of data as it changed hands and advanced through the organization and broader ecosystem. At each interchange, we conducted observations and interviewed individuals involved, mapping how the tool was enacted across different sites of practice.
Ethical considerations
The study involved observation of AHVTT administration in sensitive frontline service settings, where participants were often experiencing homelessness, crisis, or long-term vulnerability. Human research ethics review and approval were granted by the University of Queensland Human Research Ethics Committee. Observations were conducted overtly, with participants informed of the researcher’s presence and able to decline or withdraw from observation at any point. Fieldnotes focused on how the tool was administered, including clarifications, points of confusion, and interactional dynamics, rather than recording clients’ personal responses or identifying information. This approach enabled the research team to examine the AHVTT in practice while minimizing intrusion into social work relationships and protecting participant confidentiality.
Researcher positionality
The study formed part of a broader research agenda examining homelessness service provision in Queensland, in which all authors undertook fieldwork across homelessness organizations and sector settings. Accordingly, all members of the research team had exposure to the AHVTT, its use in practice, or its flow-on effects within organizational and interorganizational processes. The first author conducted a broader study investigating the use and appropriateness of the AHVTT. They therefore led the data collection specifically focused on the AHVTT, undertaking the interviews reported here and the majority of sustained observations of its administration and use. The team regularly, but informally, shared their observations and interpretations of the AHVTT administration and use.
A defining feature of the study was the positionality of Author 1, who was both embedded as a researcher and engaged in providing ongoing advice on AHVTT design and implementation. This hybrid role extended access and created opportunities for frank discussion about how the tool should evolve. It also encouraged openness from practitioners, who saw value in long-term reflection, while her outsider status allowed for more candid feedback from clients and practitioners. Author 1 consistently maintained a scientific position toward the AHVTT, emphasizing to research participants that her job was to empirically examine how the AHVTT was used. This scientific position helped build trust among the stakeholder group as there were many who either strongly supported or vocally criticized the AHVTT.
Data analysis
Fieldnotes, interview transcripts, and observation records were imported into NVivo and analyzed using a thematic approach (Padgett, Reference Padgett2008). The iterative analytical process involved memo writing, theoretical engagement, and ongoing dialogue with practitioners. This dialogue involved authors ERA and CP returning to field sites to discuss emerging interpretations of how AHVTT data moved across practice settings. Through the coding and our iterative and reciprocal discussions, four key processing interchanges were identified. An overview of these processing interchanges can be found in Table 2. Interview transcripts, fieldnotes, and observational records were coded against these processing interchanges. A second stage of coding examined how data uses were described and contested within each interchange.
Processing interchanges
Table 2. Long description
The table consists of four columns and four rows of data.
Row 1: Social work practice. Stakeholders include Clients and frontline social worker. Data transformation involves physical reality and lived experience translated into numeric data through the A H V T T. Data purpose is for assessment, prioritization, and service planning.
Row 2: Case advocacy. Stakeholders include frontline social workers and practitioners across the network. Data transformation involves A H V T T numeric data escalated through comparison. Data purpose is to prioritize individuals with the most severe A H V T T results.
Row 3: Cause advocacy and systems change. Stakeholders include leadership and management across the network. Data transformation involves data aggregated beyond individual cases to reflect the housing system. Data purpose is to identify where the system is failing clients.
Row 4: Ecosystem advocacy. Stakeholders include leadership, management, and frontline social workers. Data transformation involves data recontextualized at the organizational and practice level. Data purpose enables organizations and workers to evidence their contribution, defend practice, and hold funders or system actors accountable for inadequacies beyond organizational control.
Findings
Processing interchange 1: Social work practice
The first, and arguably most influential, processing interchange occurs between client and frontline social worker when AHVTT data is collected. This process turns the physical reality of the individual who is homeless into documentation. That documentation then “walks” them through the virtual and aggregated homelessness sector as data representing them. This data capture is crucial not only for ensuring best practice in service delivery but also for generating accurate data to support other processing interchanges.
This sequence, however, relies on a reciprocal exchange of data between social workers and clients. In Australia’s drastically under-resourced social housing sector, the collection of personal data to inform service provision and target allocations to the most disadvantaged households has become commonplace (Morris et al., Reference Morris, Clarke, Robinson, Idle and Parsell2023). This was evident in the way clients approached the street-front crisis hub with an expectation to complete paperwork. For example, after Author 1 commented on how many forms and questions one particular client had just completed, she replied, “It was a lot hey, but if they need it so they can help us out whatever, it’s fine.” This response was typical of many interactions, where clients expressed both awareness of the administrative load and resignation to it as part of receiving assistance.
Some staff aimed to alleviate the tool’s administrative load by completing it collaboratively, for example by sitting alongside clients to fill out the form together rather than conducting a question-and-answer style exchange. Workers who used this approach felt it fostered equality and camaraderie. Observations of practice support this: the form often appeared as an irritating formality imposed by an omnipotent third party, one that worker and client had to complete together before moving forward.
Although staff sought to balance the need for data collection with the burden it placed on clients and the potential damage to rapport, the AHVTT and its questions increasingly came to be seen as part of their practice framework.
When I first started, I was like what is the practice framework?… Like how are we doing assessments? So, coming in as a social worker I just felt like there wasn’t enough structure or theory to what we were doing. But then AHVTT came in and I was like ah yeah cool that’s it, that’s some sort of foundational basis. (Service Manager).
Staff described the practice framework as encompassing the data generated by the tool, which shaped the areas of need they could focus support toward, as well as how questions were asked. Delivered in a nonconfrontational and nonjudgmental way, these questions could open up further conversation and understanding with clients.
Reflecting previous research (Kithulgoda et al., Reference Kithulgoda, Vaithianathan and Parsell2024; Mathys et al., Reference Mathys, Raeymaeckers, Suykens and Van Steenlandt2024; Morris et al., Reference Morris, Clarke, Robinson, Idle and Parsell2023), we observed frontline workers exercising discretion in how questions were asked, reframed, or revisited to support clients in providing responses that reflected their circumstances and needs. Specific traits of the individual client coupled with the parts of the system most salient at the time influenced how data collection unfolded.
On the one hand, in some cases the “race to the bottom” demands were observed whereby workers directed questions and responses to demonstrate a high level of vulnerability and need for rapid and holistic supports. We witnessed exchanges between workers and clients where vulnerabilities were not adequately expressed, and workers followed up with probing questions to ensure that the answer to the instrument indicated the level of need the worker observed. For example:
Frontline Worker: Other than money for housing, do you have enough money every fortnight to take care of your day-to-day needs?
Client: Yep.
Frontline Worker: So by the end of the fortnight do you have some money left over or have you usually run out?
Client: Oh yeah I always run out.
Frontline Worker: So that sounds more like you don’t have enough to get by?
Client: Yeah yeah, nah ok that one should be no.
On the other hand, research demonstrates that, particularly for women and families, some vulnerabilities may be withheld for fear of statutory intervention or child removal. Data analysis of AHVTT responses revealed gender differences on items such as substance use and gambling, which suggests that concerns about disclosing potentially stigmatized or intervention-triggering vulnerabilities may also have shaped responses beyond anticipated gendered experiences.
That said, in our observations, clients rarely voiced such concerns to their workers, except for occasional questions like, “Do the police get access to this?” Concern about these sensitivities was primarily expressed by social workers. In discussion with senior leadership about the AHVTT’s use in practice, Author 1 noted the concern that families with children may report having sufficient income because of concerns that disclosing financial hardship could attract scrutiny from child protection authorities.
Senior Leader: Yes, for good reason. But like we say, we don’t share the data with anyone.
Author 1: Yeah, but I don’t think that’s being explained well enough at the start, and some staff have said they don’t quite understand where the boundaries on that sit. Like, to what extent does the data just stay within the database or—?
Senior Leader: Yeah, it’s staying within consent to share, because other homelessness organizations need it—that’s the point of the database. But even if we share summary information or aggregated data, ultimately there’d never be any actions taken. So while it might be shared with other organizations, nothing in this triage form would cause action for child safety or police or anything. So, we would never give it to a statutory authority to look at the raw data.
This explanation of where the data sat was reinforced in training and practice supervision sessions to alleviate worries about jeopardizing rapport with clients and exposing them to potential bias, discrimination, or punitive consequences. However, staff trust in, and perceived value of the tool strongly influenced how often AHVTT was completed.
Processing interchange 2: Case advocacy
The interagency coordination meeting was the core processing interchange where data generated in frontline practice was elevated to a systems level. A facilitator explained, “our service coordination is the Brisbane zero meeting, every Friday Brisbane City Council, Corrections, QPS [police], Housing, and a bunch of support services all come together in the same room.” These meetings brought “everyone” together to prioritize individuals, discuss options for housing allocations, and share updates on the people represented in the data who had not been seen for some time. Workers made follow-up requests and checked directly with the housing authority regarding the availability of social housing and the status of individual clients.
AHVTT served as a critical infrastructure here. To advocate effectively in these meetings, workers needed intimate knowledge of the individuals they represented, anticipating any questions and framing the case persuasively to secure resources. The AHVTT was valued for ensuring they had “all of the information they could possibly need” (Frontline Worker), and the form was occasionally referred to directly.
Social Worker: Do they have any debts they need help sorting out?
Key Social Worker: Ahh let me just check I don’t think so… [consults completed AHVTT]. No they didn’t mention any.
Although the tool was only occasionally invoked directly, its value lay in the creation of a shared, data-driven language that proved especially persuasive in coordination meetings with housing authority staff. Social workers regarded this common framework as a key benefit of the AHVTT in enabling coordinated practice. As explained during a training session,
What the AHVTT can do that other assessments can’t is have us [the homelessness sector] all speaking in a common language and coordinating a common way. So that means, if somebody has an AHVTT score in your organisation, that means the same thing if they showed up to another organisation.
In this way, the AHVTT illustrates how professionalization and standardization of social work practice can generate legitimacy.
Processing interchange 3: Cause advocacy and systems change
AHVTT data are aggregated beyond individual cases to reflect on the broader housing system and provide a critical foundation for cause advocacy by demonstrating scale and characteristics of homelessness while also exposing systemic gaps. As was explained in training sessions:
Step one: know who is actually homeless. So who is out there? If somebody is case managed by Organisation A and then they show up at Organisation B, and then Organisation C sees them on outreach is that three people, or is that one person? How many people are out there? What’s the actual size of the problem, and what are the kind of demographic? You know, are we seeing a bunch of women fleeing domestic violence? That’s going to be a different problem to solve from if we’re seeing a bunch of veterans experience post-traumatic stress, right? So who are we seeing out there? What do they need? Then how do we work together to figure out what to do? (AHVTT Trainer)
This perspective underscores how data aggregation prevents double counting, clarifies the scale of homelessness, and makes visible the demographic patterns and needs of those experiencing homelessness. By linking disparate encounters underlying unmet needs, aggregated data expose where the system is failing to deliver lasting solutions. In advocacy settings, such data strengthen calls for policy and service reform by shifting conversations from isolated anecdotes to evidence-backed understanding of systemic shortcomings.
A strong example of this arose following a cyclone event in Brisbane, when the City Council sought to remove all people sleeping rough in parks. The Council’s claims quoted in news media (Messenger & Ore, Reference Messenger and Ore2025) implied that all rough sleepers had been offered shelter and relied on stereotypes about behavior and choice to explain why people remained in the streets. Challenging this framing of homelessness as a lifestyle decision rather than a structural constraint, the CEO, Karyn Walsh, of the leading AtoZ organization drew on data derived from the AtoZ to contest this narrative (Messenger & Ore, Reference Messenger and Ore2025). She reported that of the 261 people who sought accommodation during the cyclone, 146 were denied due to lacking identification, while 44 others with identification remained on waiting lists. Only 71 had secured accommodation. By drawing on the data elicited by social workers on the ground, she was able to frame homelessness as the predictable outcome of a system under severe strain, where bureaucratic requirements and insufficient housing stock prevented access even for those actively seeking shelter.
This exchange underscores a tension central to the ecosystem: the Council’s reliance on stereotypes to legitimize regulatory interventions contrasts sharply with the data-driven, systemic critique enabled by AHVTT data aggregated into A to Z. While the Council’s narrative centers on compliance and personal responsibility, the sector’s response demonstrates how quantitative evidence can be mobilized to reveal structural barriers and expose the inadequacy of existing policy responses.
Processing interchange 4: Ecosystem accountability
The final series of processing interchanges we outline concerns the exchange of data collected by frontline social workers with their management and leadership teams, who then use the aggregated data for multiple accountability functions. The role of the AHVTT as a performance measurement tool is complex, but typical of, and arguably necessary within the context of homelessness services. Performance measurement instruments are often understood as either activity-based measures such as hours of service delivered, or outcome-based measures, in which change over time is assessed (Ahearn & Parsell, Reference Ahearn and Parsell2026). The AHVTT, however, captures vulnerability and, by extension, the complexity of the work being undertaken by frontline social workers. In doing so, it enables the organization to account for the intensity of frontline work within an underfunded service system. The data can simultaneously be used to hold that system to account, by making visible the levels of need that homelessness services are inadequately resourced to address. Further, organizations use AHVTT data to enact accountability to frontline staff themselves, including by monitoring workloads in relation to caseload and complexity. Together, these interchanges contribute to a tentative balance of accountability between stakeholders in the ecosystem, moderated by the AHVTT data.
Unlike most previous research which investigates funder-mandated or top-down measures, the AHVTT and the AtoZ database that hosts it are sector-owned and operated. Organizations within the housing sector were still required to fulfil top-down reporting measures for contractual accountability to funders, in many cases the housing authority. Although the AHVTT was not designed with this purpose in mind, it provided an additional benefit to frontline staff, as the information collected through it was more comprehensive than that required by funders. As a result, staff often used AHVTT scores to meet their reporting requirements:
Before AHVTT you’d go to do your [funder] report and realise oh shit I forgot to ask about that and then you’d have to try and follow up with the participant but like, you know, you mightn’t find them again or you know so yeah with AHVTT I’ve always got all of the information I need in there.(Frontline Worker)
The depth and accuracy of the AHVTT and the AtoZ database were broadly recognized across the sector as superior to government-held homelessness data. However, this granularity was only explicitly leveraged when the data were strategically deployed for cause advocacy efforts.
Beyond fulfilling contractual requirements, the AHVTT data were also used to hold the housing authority and other government departments to account. One sector representative explained that, although government agencies fund the development of tools such as the AtoZ Digital Database, tensions can arise when the resulting data are later used in the media to criticize those same agencies. However, as they clarified, “that’s not the intention, the intention is to make systems accountable.” In one training session it was explained how this dynamic unfolded in practice:
If it’s seen that government is giving a lot of funding to the specialist homelessness services then the health system and the justice system will see this, and they’ll just try and push all their problems onto us. But if we have clear data and clear service coordination we can see when that’s happening and hold those other systems to account because it shows up in the data. (AHVTT Training)
While holding the systems to account in this way was certainly intended to support people experiencing homelessness, relieving the burden on the sector and frontline workers was also a strong focus of leadership. Indeed, one aspiration was for AHVTT data to directly inform individual workload allocations. As described in training:
If one worker has 10 clients and they’re all low AHVTT scores and another has 10 clients and they’re all high AHVTT scores who do you think is going to burnout first? Right so that’s not very fair and if we can see that then we can get in and adjust the case allocations. (AHVTT Training)
Ultimately, the AHVTT data were used to fulfil accountability relationships across the ecosystem, thereby enabling trust and legitimacy within it. In these “institutional relations,” the data functioned as “a central nervous system running through and coordinating different sites” (DeVault & McCoy, Reference DeVault, McCoy and Smith2006, p. 33). By being data-driven and producing evidence, the AHVTT also created the foundation for holding other systems to account, while simultaneously supporting frontline workers to carry out their roles effectively. Yet the primary accountability to which all of these processes ultimately return is the commitment to supporting people experiencing homelessness. In a complex ecosystem, this central mission can only be achieved once multiple layers of accountability have been negotiated and fulfilled along the way.
Discussion and conclusion
Through our ethnographic case study, we found that data actively facilitated the nonprofit ecosystem through four processing interchanges. First, the collection of data from service beneficiaries was central to frontline social work practice, while simultaneously being shaped by the discretion of individual workers and their professional practice. Second, data functioned as a mechanism for enabling the complex coordination and collaboration required between organizations to provide care. Third, aggregated and shared data became a resource for cause advocacy, both across organizations and in communication with the media and government. Fourth, data supported advocacy for the homelessness sector itself by evidencing its challenges and effectiveness, legitimizing the work of staff, and holding other parts of the ecosystem accountable when they constrained outcomes.
These findings challenge the prevailing perspective of nonprofit data that center primarily on dyadic reporting relationships between nonprofits and funders (Carnochan et al., Reference Carnochan, Samples, Myers and Austin2014; Cordery et al., Reference Cordery, Goncharenko, Polzer, McConville and Belal2023; Polonsky et al., Reference Polonsky, Landreth Grau and McDonald2016; Suykens et al., Reference Suykens, Meyfroodt, Desmidt and Verschuere2022). While these relationships remain important, our findings show that they represent only one dimension of the accountability arrangements through which nonprofit data circulates. In this study, data were mobilized across multiple forums: to coordinate care, support advocacy, account for the complexity of frontline work, manage workloads, and challenge systemic constraints on service outcomes.
This analysis reminds us that accountability data ultimately originate in practice. Our research demonstrates that data should not be understood only as a reporting artifact produced after practice for external scrutiny. Rather, accountability data are shaped from the outset by professional judgement, relational work, care, and institutional pressures. In homelessness services, this is especially important because what is recorded may reflect not only client need or service effectiveness, but also the strategic and relational dynamics through which vulnerability is made visible in a resource-constrained system. For example, where access to scarce housing resources depends on demonstrating acute vulnerability, both clients and workers may have reason to emphasize the severity of need (Kithulgoda et al., Reference Kithulgoda, Vaithianathan and Parsell2024; Mathys et al., Reference Mathys, Raeymaeckers, Suykens and Van Steenlandt2024; Morris et al., Reference Morris, Clarke, Robinson, Idle and Parsell2023). Rather than narrowly viewing data as a measure of effectiveness, we need to recognize the frontline decisions and practice that determine what is recorded in the first place. Situating data within social work practice therefore makes visible how data reflect, influence, and circulate through practice, sustaining multiple forms of accountability along the way.
Within this web of accountability relationships, data do not simply flow upward to funders, downward to clients, outward through voluntary public reporting, or horizontally across organizations as separate dyadic exchanges (Benjamin et al., Reference Benjamin, Voida and Bopp2018; Cordery et al., Reference Cordery, Goncharenko, Polzer, McConville and Belal2023; Ebrahim, Reference Ebrahim2005; Lebec & Dudau, Reference Lebec and Dudau2024; Schoech, Reference Schoech2010). Rather, the same data may be generated in one relationship and then transformed and remobilized across multiple accountability forums. This increased mobility of data has been largely enabled by rapid advancements in digital infrastructure (Ahearn & Parsell Reference Ahearn and Parsell2025; Ahearn et al., Reference Ahearn, Mai and Marston2026; Rana & Cordery, Reference Rana and Cordery2024; Schoech, Reference Schoech2010). Schoech (Reference Schoech2010) describes this as interoperability: the electronic exchange of information that is automatic, instantaneous, seamless, and secure across organizations based on agreed data definitions, standards, and protocols. Such interoperability is likely to become more common, given pressures on nonprofits to maximize resources and the increasing capabilities of technology (Benjamin et al., Reference Benjamin, Voida and Bopp2018; Schoech, Reference Schoech2010).
Our research documents an important progression in nonprofit data and digital data systems, highlighting a new stage of interoperability. The AHVTT was neither imposed by funders or regulators, nor developed solely bottom-up within a single organization. Rather, it was horizontally standardized across participating homelessness services, which collaboratively agreed to collect and share common data to make vulnerability, complexity, and service demand visible across the sector. This horizontal standardization means that the AHVTT is collectively required by organizations as a condition of shared data use. This complicates the usual opposition between top-down and bottom-up measurement by showing how sector-led interoperability generates its own standards, expectations, and forms of accountability. This is particularly significant in a resource-constrained sector where organizations must collaborate to address service gaps and coordinate care, while also operating within funding regimes that can place them in competition for scarce resources and legitimacy (Curley et al., Reference Curley, Levine Daniel, Walk and Harrison2021; Raeymaeckers & Dierckx, Reference Raeymaeckers and Dierckx2012).
The findings presented here are drawn from a case study of one sector of nonprofit social services in Australia. It is possible that these organizations are further advanced than others in terms of data literacy and digital infrastructure. Consistent with naturalistic conceptions of transferability (Lincoln & Guba, Reference Lincoln and Guba1985), we therefore do not assume that the particular configuration observed here is universally generalizable. Rather, by specifying the characteristics and conditions of this ecosystem, we aim to enable readers to assess the extent to which the dynamics identified may be transferable to other nonprofit contexts characterized by similar relationships between practice, management, advocacy, funding, and governance. It is notable that homelessness services, despite being among the least funded parts of the sector, appear to have reached this point, perhaps out of a necessity for survival rather than through abundant resourcing. Indeed, future research should be directed to understanding why more complex and advanced data collection systems are often observed within the homelessness sector (Benjamin et al., Reference Benjamin, Voida and Bopp2018; Schoech, Reference Schoech2010), and whether the dynamics identified here operate similarly in sectors with different levels of data maturity and infrastructure.
Previous research has often framed nonprofit data primarily as a mechanism for reporting to funders. Our findings expand this conceptualization of both “use” and “accountability” by showing how data is mobilized across relationships that extend well beyond funders, shaping coordination, advocacy, workforce management, and frontline practice itself. In this context, accountability cannot be reduced to providing an account to a principal; it is enacted across a wider ecosystem of clients, staff, partner organizations, government, media, and the broader sector. This function of nonprofit data has been obscured by the tendency to view management through a private sector lens. As data become increasingly interoperable, aggregated, and repurposed across different interchanges, it is essential to remain attentive to the assumptions, discretion, and institutional pressures embedded in their production. Further research should examine how nonprofit data travel across accountability forums, particularly in sectors or programs where data are tied to funding, performance incentives, or payment-by-outcomes models. These contexts offer important sites for understanding how data may work in unintended or unforeseen ways, and for recognizing the limits of what multipurposed nonprofit data actually represent.
Data availability statement
The data that support the findings of this study are available from the corresponding author upon reasonable request.
Funding statement
This research was supported by the Australian Government through the Australian Research Council’s Centre of Excellence for Children and Families over the Life Course (Project ID CE200100025) and an ARC Industry Fellowship (IM230100850).
Competing interests
The authors declare no conflicts of interest.
AI statement
Artificial Intelligence has been used for proofreading this manuscript.
Open access
