Although every IDS is distinct, key differences emerge when comparing systems across three core organising principles: geography (statewide or local), management model (the type of organisation that hosts and governs the system), and purpose (how and why the IDS is used). To support understanding of these variations, AISP created a taxonomy and quality framework based on these dimensions [1, 5], which highlights commonalities and distinctions among IDS initiatives nationwide (see Figure 1).

The field of IDS is not new [6] but has expanded rapidly in the past decade. An initial scan conducted by Actionable Intelligence for Social Policy (AISP) in 2008 identified just 8 U.S.-based IDS. Selection criteria specified that an IDS must have an established legal framework and governance structure and be routinely linking cross-sector data for programme and policy analysis [7]. Since this initial scan, AISP has supported state and local jurisdictions building shared data infrastructure through training, technical assistance, and best practice development. Additionally, AISP convenes a network of IDS across the U.S and conducts a biennial survey that captures the state of the field and growth over time. Despite the challenges of routinely linking administrative data [8–10], IDS efforts across the U.S. have been successful in creating shared data infrastructure that includes collaborative governance, clear legal authority, adaptable technology, and appropriate staff capacity [1].

Record linkage has many potential approaches and core purposes, so it is difficult to determine a valid count of data integration efforts in the U.S. However, the parameters of the AISP Network are explicitly defined. To be a member of the AISP Network, sites must have a clear legal framework and data governance to guide the access and use of cross-sector administrative data, with demonstrated linkage and use of data for impact. The AISP Network now includes over 45 sites [11]. The Network operates on a voluntary, no cost membership model, and Network Sites collectively represent 53% of the U.S. population.

Internationally, AISP’s network building work is comparable to organisations such as Administrative Data Research UK (ADRUK) and Health Data Research Network Canada (HDRN). All three organisations support field building and best practices for networks of data holders across jurisdictions. However, unlike ADRUK and HDRN, AISP is not a data holder, does not operate a trusted research environment (TRE), and does not receive government funding for network activities. AISP’s network activities are funded by philanthropic partners and AISP is university-based, fielding a small team (<10 employees). AISP’s primary work is studying best practices for administrative data reuse and supporting the broader field to implement these best practices, while comparable organisations in other countries focus more on research.

The concept of IDS is quite simple—by linking data across domains at the person level, these systems allow partners to better understand programmes and policies and make decisions that support public good. They are an essential resource for evidence-based policymaking. However, significant barriers remain to their development and expansion into new jurisdictions across the U.S. Currently, the growth of the IDS field is constrained by limited understanding of how these organisations are developed, structured, funded, and staffed. Recent publications have begun to address these questions specific to state longitudinal data systems in the education domain [12, 13] but broader efforts across housing, health and human services, child welfare, and beyond remain under-studied. Drawing on results from two surveys of the AISP network and in-depth qualitative interviews, this article describes the typical development trajectory, budgetary, and staffing structures of IDS and explores how resourcing and the chosen management model affect capacity and IDS impact. In doing so, we aim to provide valuable insights for practitioners, policymakers, and researchers seeking to enhance the effectiveness and sustainability of their administrative data reuse efforts.

Both surveys included three main sections: (i) IDS purpose and management model; (ii) practices and approaches across the five core quality framework components (governance, legal, technology, capacity, impact), [1]; and (iii) current data holdings and practices around data use and access. See Appendix A for an overview of the survey instrument. The first section asked respondents to share their management model (e.g., executive-led, agency-led, university-public partnership, community based organisation led) and how integrated data are used (e.g., conduct research, inform policy, evaluate programmes, support case management and service delivery, etc.). In the second section, respondents were asked about governance structure, forms of legal authority used to facilitate interagency data integration, technical approaches to data integration, information about staffing and budgets, and the observed impact of data use across policy domains. In the final section, respondents reported the types of data that were integrated over the previous two years (e.g., data on vital records, health, child welfare and/or adult protection, early childhood, education, legal and law enforcement engagement, economic security and employment, housing and homelessness, or other local, state, or federal data sources). Multiple-choice and free-response questions were included to enable qualitative explanations.

Survey data from 2023 and 2025 were analysed using descriptive statistics to identify patterns and shifts. When discrepancies emerged between years, 2025 data were used as the default point of comparison to reflect the most current trends. Survey findings were further contextualised with interview data to deepen understanding of participant perspectives.

We conducted semi-structured interviews to contextualise and fill gaps in the information gleaned from the surveys. These interviews were conducted in person during our AISP National Meeting (June 2024) and virtually (June and July 2024). Interviewees represented a diverse group of practitioners and experts from local and state government agencies, universities, and non-profits. The title and duties of interviewees varied, but all led IDS efforts, including some sites that have been AISP Network members since as early as 2009, as well as newer members. After participants provided consent, the interviews, which lasted approximately 30 minutes, were recorded, and notes were taken by the interviewer and/or a dedicated notetaker. Interviewees were not compensated for participation. See Appendix B for a copy of the interview protocol.

Deductive coding was used given an a-priori set of codes derived from findings of the 2023 survey data. The interviewer and one additional author reviewed each interview recording for pre-specified information related to the site’s focus, partnerships, budget, funding, charge model, actual and ideal staff positions, effort, and structure, as well as lessons learned. The lead author (AHN) reviewed the pulled information and resolved any discrepancies. Social desirability was minimised by framing the interview as a means for new data integration efforts to learn from more established ones. The interviewers emphasised that there were no correct answers and ensured anonymity in result dissemination. The sample size was chosen, in part, to ensure sufficient interviews to reach thematic saturation. A condensed version of both the network survey and interview protocol can be found in Appendix A and B.

Our approach of collecting survey data over two years (2023 and 2025) provides a wealth of longitudinal information, but it also introduces analytical challenges. For one, sites changed during this time period, experiencing shifts in funding sources, budget sizes, and staffing structures. Further, the U.S. federal data landscape has changed significantly since January 2025, whereby a new presidential administration enacted dramatic cuts to research funding, national statistical infrastructure, and grantmaking to states and localities. As a result, almost all IDS—especially those operating within academic institutions—are now operating with less funding and staffing than at the time of data collection. At the same time, there has also been an increase in federal demand for and ingestion of state administrative data, leading to increased concerns about privacy, security, governance, and publicly available metadata. In response, AISP has stopped publishing detailed information on the data holdings of Network members, data that were previously publicly available from 2012-2025. Given this broader context, we prioritised reporting overarching trends and have not included site names to protect their anonymity.

Change is a defining characteristic of complex systems. IDS naturally grow and shift over time in ways unique to each context; however, the chosen management model is a helpful entry point to understand the typical paths these efforts take. Where these efforts are located (within or outside of government) influences how they are able to develop, the data that is accessible, the legal challenges they face, and the funding streams they can access. Figure 2 shows the distribution of AISP network sites across the four management models: Agency-led, Executive-led, University-public partnership, and Community based organisation led.

Survey and interview findings indicate that agency-led efforts are the most common approach in the AISP Network (See Figure 2). This is likely because agencies tend to have the capacity to address the complex web of U.S. federal and state privacy laws when building IDS. Management models are also influenced by the purpose they aim to serve (See Table 1). For example, a Health and Human Service Agency may develop an IDS to better understand cross-agency enrollment, programme participation, and downstream social service use. Executive-led models tend to start smaller, are often created for a specific initiative (e.g., improving maternal health outcomes, reducing homelessness), and evolve to support interagency, evidence-based policy making. University-public partnerships and CBO-led IDS tend to be more non-linear in their development, and are often situated outside of government, for reasons such as an interest in integrating data within a neutral institution to improve trust or in response to constrained staff capacity in government.

As Table 1 demonstrates, IDS funding can come from local, state, federal, and philanthropic funds. Additionally, sites can operate in a fee-for-service model whereby they charge for distinct tasks such as preparation of an analytic dataset or completion of a research project. Each funding source has advantages and disadvantages. Philanthropic funding, which supported 11 sites in 2025, is often short-term (one to three years) and tied to a specific focus (e.g., early childhood) and/or geography (Table 1). Local (n = 6), state (n = 17), and federal (n = 13) government funds offer more long-term stability, but come with strict criteria on how funds can be used (e.g., human capital vs. infrastructure). Given the ongoing demands of maintaining institutional knowledge and data system upkeep, tethering efforts to government funding can be precarious due to changing administrations and shifting priorities at the local, state, and federal level. Fee-for-service offers flexibility to support diverse activities, but the overhead costs of governance, maintenance, and integration of new data sources make it challenging to recoup all costs through this model alone. Therefore, all 12 sites that employ fee-for-service do so in conjunction with other funding sources. Given the disproportionate need for staff effort at the outset of IDS creation, a fee-for-service funding model is only possible once the site is past the development stage.

Each site has a slightly different origin story, but philanthropic and federal funding are often involved in the initial development, with sites moving to more diverse funding sources as the effort matures. When describing their site’s mix of funding sources, one participant reflected “So it is totally blended, which I’ll say is a challenge, managing that, but we also see it as kind of protective…regardless of what happens, we know that if any one piece falls out, if any one piece goes away, there is enough other stuff that we feel confident that the core will be sustained.” Most importantly, successful sites use their initial funding to develop relationships, establish data governance, and collaboratively create legal frameworks, rather than building technical infrastructure. This can be a challenge if funders do not see this as a priority, one participant recounts, “so our advice is like, don’t underestimate that time… The relationship building on all different levels, and that is also the hardest fund. Our [funder] realises the immense capacity it takes to build that trust, and then the sharing of data, but not all funders are patient.”

Across the types of funding sources, the surveys and interviews revealed three broad categories of budget size: Small efforts are defined as having annual budgets between $250,000 and $499,000 and included seven sites in 2025 (Figure 3). Five sites were medium efforts, defined as having budgets between $500,000 and $999,999. Large efforts are those with budgets over one million dollars and represented 16 sites in 2025. In general, small efforts tend to serve smaller populations, focus on a specific domain, and employ two to three full-time staff members. Medium efforts often operate at the county- or state level, tend to be based at universities, and employ three to six full-time staff, in addition to student and faculty affiliates. Large efforts tend to operate at the county- or state level, are usually situated within government, and have up to 20-40 full-time staff members.

Interestingly, the costs for development within these bands have remained fairly stable since we began collecting this information in 2008. At their outset, small-sized efforts often operate around $350,000 and as they develop and grow into medium-sized efforts, expand to around one million dollars. Sites that serve larger geographies are associated with larger budgets (Figure 3) and are the sites most likely to integrate data for case management purposes (Table 1). The largest budget in our Network is about $10 million dollars annually while six sites are in early development and currently unfunded and seven have budgets less than $250,000 (Figure 3). As sites mature, the AISP Network has seen a 300% increase in sites that have budgets of more than $5 million from 2023-2025, increasing from one to four sites. These increases were largely due to existing IDS expanding analytic and reporting capacity. We expect that budgets will continue to rise as local and state governments need more access to linked data.

IDS development varies across sites, with almost every respondent describing their structure and approach as different from other AISP sites due to factors like age, staffing structures, or funding mechanisms. This diversity suggests that in practice, there is no single conventional model to IDS development and maintenance, even as many still implicitly reference a perceived archetype. Across contexts, personnel costs consistently represent the largest investment, reinforcing that IDS work is fundamentally a people-centred endeavour. Success relies on relationships, buy-in, and operational capacity, while the technical components are often seen as comparatively easier to resource. Because IDS work develops iteratively, systems must evolve as needs and capacities change. As integrated data efforts grow, some sites aspire to become the state’s singular data hub, while others intentionally define a narrower niche to avoid serving as the sole statewide resource. This landscape can create competitive dynamics, where multiple IDS efforts coexist, underscoring the continually shifting environment in which these systems operate.

While some challenges are universal, there are distinct challenges and opportunities based on the management model. Within the agency, executive, and university setting, there is a persistent need to educate the institution about how IDS work differs from standard processes, causing potential for friction when existing systems and expectations don’t align with what integrated data work requires. For example, sites in these settings face issues of procurement (e.g., How can we craft a sole source contract for this particular approach?) and salary bands (e.g., How can we pay our data scientists close to what private industry offers?). Conversely, this is where CBO-led efforts shine, and the origin story of AISP Network Sites that are situated within CBOs often include anecdotes that the management model was selected to avoid complex bureaucracy and allow more nimble funding and staffing. However, being located in a CBO is often less stable, without economies of scale to facilitate niche staffing positions present in larger university or government institutions (e.g., database architect or ethics officer).

Universities tend to have more stable staffing than government and CBO’s, with greater staff retention, faculty often serving in lifetime appointments, and access to graduate students that may support the work for 2-6 years. Universities also have access to staff with deep expertise that can be utilised for consultation and support through a dedicated portion of staff effort. Universities are used to funding a percentage of a person’s time on a project, an approach that is atypical in other contexts. Universities play a particularly important role in the university-public partnership by offering continuity and stability that governmental partners often lack. Within these partnerships, the university often serves as the source of institutional memory, providing long-term consistency and helping sustain efforts despite turnover, shifting priorities, or structural changes within government agencies. However, it is important to note that public partnerships require consistent engagement and support from government and CBO partners. Data partners must be invested to participate in governance activities, which are often outside of formal job duties.

Funding for IDS work is shaped by a core sustainability lesson: diversify revenue sources. Sites emphasise beginning with investments in partnership development, data governance, and relationship building—rather than starting with infrastructure—because trust and priority alignment are foundational to success. Philanthropic dollars often support early stages through planning grants and start-up operating costs, while federal funds typically finance technical infrastructure and major initial investments, with the goal of eventually shifting to ongoing appropriations. Approaches vary and some sites leverage project funding to strengthen infrastructure, whereas others intentionally pursue infrastructure-specific funds and avoid project-based support. Many sites note that the slow, relational nature of this work is difficult to fund. Even when systems are functioning well, securing new investment can be challenging because leadership may not see an urgent need. One respondent reflected “when something works and people see projects in place. How do you convince them that you actually need to make investments... because they’re like ‘It works. You guys are here’.”

In the interviews, all sites distinguished hard versus soft funding. Hard funding is stable, ongoing support—often from government appropriations or permanent budgets—that an organisation can reliably expect year to year. Soft funding is temporary or uncertain support—like grants, philanthropy, or short-term contracts—that must be renewed or replaced regularly. Sites often weigh the benefits and drawbacks of these funding types. Avoiding constant grant-seeking can be beneficial, but reliance on government appropriations can be precarious amid administrative turnover. Fee-for-service models offer flexibility but reflect a specific perspective on data rights and access. Specifically, several respondents discussed the contradiction of viewing data as a public resource that should be made available regardless of ability to pay, but recognising that sites must cover data management and analytic costs to provide that information.

Respondents frequently discussed the importance of consistent staffing and ensuring transfer of institutional knowledge. They emphasised the importance of documenting processes early and often, both to maintain continuity and to distribute knowledge across the team. Consultants or “pseudo employees” can help fill gaps, but systems should avoid relying on “unicorns”— individuals who hold all the institutional knowledge and who cannot be replaced. One respondent noted “but you don’t wanna have a unicorn, and that’s also the kiss of death. You want a process that’s replicable, and safety nets that people new can come in, learn and follow.” Instead, as the quote suggests, key information should be codified in procedures, shared among multiple staff members, and supported by redundancy and cross-training to ensure the work remains resilient even amid turnover or change. Respondents also discussed the challenge in funding capacity for analysis and dissemination. In small- and medium-sized IDS, staff are driving all parts of the data life cycle. Often this limits their capacity to devote time to important tasks that allow for meaning-making after initial data products are delivered, such as dissemination, strategic communication, and partner engagement. The major distinction between sites lies in their capacity to afford personnel with expertise across stages of the data life cycle, meaning separate staff to support public involvement, conduct analyses, design data products internally, and communications professionals that can use these products to communicate externally and drive action.

In both the surveys and qualitative interviews, respondents were asked about how AISP and its network of sites could help to address needs and challenges. Assistance with capacity constraints was a consistently identified need.

As noted above, sites indicated that they are understaffed across a range of areas—particularly in legal, technical, communications, engagement, and project management roles. Several sites referenced difficulty recruiting and supporting data scientists. Reliance on project-based or soft-money funding undermines long-term sustainability, while dissemination and grant writing remain underfunded. Many expressed interest in strategies such as student engagement, shared services, and multi-site collaborations, along with a need for clear job descriptions and role definitions to streamline hiring. Sites indicated interest in business planning support for fee-for-service or diversified funding models, shared staffing structures across legal, administrative, and technical functions, examples of sustainable funding approaches in universities and government agencies, and a resource bank of job descriptions and hiring strategies.