The Case for Nonprofit Data Standards: How Shared Schemas Could Transform Sector-Wide AI

The average nonprofit uses five or more separate systems beyond their primary CRM, and these systems almost never share data natively. A typical organization might have one platform for donor management, another for volunteer coordination, a third for program tracking, a grant management tool, and a communication platform, plus spreadsheets filling the gaps where none of the systems talk to each other. Each system uses its own data model, its own naming conventions, and its own export formats. When data needs to move between systems, it usually involves manual export, reformatting in spreadsheets, and manual import, a process that introduces errors and consumes staff time that could be spent on mission delivery.

Research from Heller Consulting published in 2026 found that nonprofits using multiple unintegrated data systems spend a disproportionate share of their evaluation capacity on data cleanup rather than actual analysis. One foundation that moved to standardized data collection with shared identifiers and consistent field definitions reduced its portfolio reporting time from six weeks to two days, a more than 95% efficiency improvement driven entirely by eliminating the reformatting work that had previously consumed most of the reporting cycle. This kind of gain is not exceptional; it is the predictable result of solving a structural problem that most organizations have simply learned to live with.

The AI implications are significant and underappreciated. AI tools are as useful as the data they can access. An AI grant writing assistant that can only read your organization's own historical applications cannot compare your approach to what has worked across hundreds of similar nonprofits. A donor retention model trained only on your CRM cannot incorporate the broader patterns across thousands of comparable organizations that would make its predictions genuinely predictive. A program evaluation tool that cannot aggregate findings across organizations serving similar populations cannot answer the sector-wide question of what interventions actually work. Every AI limitation that stems from data fragmentation is a limitation that shared standards would eliminate.

The nonprofit sector does not need to invent standards infrastructure from scratch. Two other sectors have navigated this transition in ways that offer directly applicable lessons, one through voluntary coordination combined with regulatory mandate (healthcare), and one primarily through regulatory mandate (finance). Both paths have worked, with different timelines and trade-offs.

Healthcare's journey to interoperability began decades ago with HL7 (Health Level Seven), which established standards for exchanging clinical and administrative data between healthcare systems. The more recent and transformative development was FHIR (Fast Healthcare Interoperability Resources), which brought modern API architecture to healthcare data exchange. The US government mandated FHIR-based data sharing through the 21st Century Cures Act, creating regulatory pressure that made adoption economically unavoidable for large healthcare systems. By early 2026, TEFCA (Trusted Exchange Framework and Common Agreement), the national health data interoperability network built on FHIR, reached nearly 500 million health records exchanged across systems that were previously incompatible. HL7 launched an AI Office in 2025 specifically to apply FHIR's interoperability architecture to AI use cases, recognizing that standards-based data is the foundation that makes healthcare AI genuinely useful rather than limited to siloed applications.

The lessons for nonprofits from healthcare's standards journey are specific and actionable. Open, collaborative standards communities create well-designed specifications that find adoption because they serve genuine needs rather than proprietary interests. Involving practitioners, not just technologists, in standards development is essential for adoption because practitioners are the ones who know what data they actually generate and what they actually need from data. Pilot testing before full rollout prevents the specification drift that occurs when standards are written without grounding in real organizational data. And data quality is the foundation on which any AI value depends: healthcare AI experts consistently describe good data quality as a prerequisite for good AI performance, not a nice-to-have.

Finance's path through open banking offers a different but equally instructive lesson. European open banking, mandated by PSD2, standardized financial data APIs in a way that created an ecosystem where third-party services could build on shared data infrastructure. The API call volume generated by this ecosystem is projected to grow from 137 billion in 2025 to over 700 billion globally by 2029, demonstrating the innovation that standards unlock. The key insight from finance is that regulatory mandate can accelerate adoption of voluntary standards that have stalled. The nonprofit sector does not currently have a regulatory equivalent to PSD2, but foundations and institutional funders have significant leverage to create adoption incentives by requiring grantees to use standard data schemas for reporting.

While data fragmentation affects every aspect of nonprofit operations, three domains present the highest opportunity for standards-driven improvement: donor data, program outcome data, and funder and grant data. Each has distinct characteristics and distinct standards implications.

Donor data is perhaps the most immediately tractable problem. The core issue is the absence of a universal donor identity standard: the same person appears as separate, unlinked records across Salesforce NPSP, Blackbaud Raiser's Edge, DonorPerfect, Little Green Light, and dozens of other platforms. There is no equivalent of a shared identifier that would allow organizations to recognize that their mid-level donor who gives annually is the same person who made a major gift to a hospital foundation two years ago. This fragmentation limits the sector's understanding of philanthropic behavior and prevents portfolio-level analysis of giving patterns. Data privacy laws, which as of 2026 have been enacted in 13 or more US states with varying requirements, add compliance complexity to any donor data standardization effort but do not eliminate the opportunity; they just require careful design of consent and governance frameworks.

Program outcome data is perhaps the most fragmented domain of all. There is no agreed-upon definition of a "program outcome" that applies across nonprofit types, program models, or even similar organizations working in the same field. Organizations measuring employment outcomes after job training programs use different definitions of "employment," different follow-up timeframes, and different measurement instruments. Even funders who require outcome reporting often accept whatever framework the grantee prefers, resulting in portfolio data that cannot be aggregated. Multiple frameworks compete for adoption, including IRIS+, the Common Impact Data Standard, Outcomes Star, and Social Return on Investment, without a clear sector standard emerging. The AI opportunity in this domain is enormous: sector-wide outcome data would finally allow the field to answer questions about what works that have been debated for decades without resolution.

Funder and grant data has a more developed standards ecosystem than the other two domains. Candid maintains a comprehensive database of foundation information and grant history with APIs that provide access to over 2,500 data points. The International Aid Transparency Initiative (IATI) provides a standard for publishing development aid data that many international funders use. The combination of Candid's data infrastructure and AI tools like its Anthropic MCP connector represents the sector's most advanced current example of what standardized funder data enables for nonprofits: research that previously took weeks now takes minutes.

The nonprofit sector's standards challenge is not primarily a technical problem. The standards exist. The business case is clear. The barrier is coordination: each organization has limited incentive to invest in standards adoption unilaterally, because most of the value from standards comes from network effects that only materialize when many organizations participate. This is the classic collective action problem, and the nonprofit sector has historically struggled to solve it.

The most powerful levers for changing this dynamic lie with funders and intermediaries rather than individual organizations. When funders require grantees to report outcomes using a specified standard framework, they create an adoption incentive that individual organizations would not otherwise have. When major sector intermediaries like Salesforce.org, Blackbaud, or Microsoft adopt and promote specific standards in their platforms, they reduce the implementation cost for the organizations that use those platforms. When peer learning networks make standards adoption part of their shared infrastructure, they create communities of practice that sustain adoption over time.

Individual organizations that adopt standards early gain competitive advantages in the funding landscape as these dynamics develop. Foundations that are building toward data-driven grantmaking and AI-assisted due diligence will increasingly favor organizations whose data can be aggregated and analyzed without custom formatting work. Organizations that can demonstrate interoperability with sector data infrastructure will have stronger grant applications, clearer impact narratives, and faster reporting cycles than those still operating entirely in proprietary data silos. The case for early adoption is not altruism; it is strategic positioning for a sector that is moving, however slowly, toward data as infrastructure.

For more on the technology infrastructure decisions that underpin AI effectiveness, see our articles on why nonprofit AI tools don't talk to each other, data governance policies for AI, and the clean data imperative for nonprofit AI.