The nonprofit sector has amnesia. Nonprofits lose billions to fraud every year — not because people are dishonest, but because the sector has no memory.
The Association of Certified Fraud Examiners has tracked this for over a decade: the same type of fraud, the same size loss ($75K–$110K), the same 12–18 months before anyone notices, the same root cause — a lack of basic financial controls. And no real progress on fixing the problem. While it may be existential for one nonprofit, it is not existential for the sector. And the closest thing nonprofits have to a clearing system is funders.
How Every Other Sector Builds Shared Memory
Every other sector fixes this problem the same way — by building a shared memory. Banks have regulators and clearing systems. Hospitals have accreditation bodies. Airlines have the FAA and anonymous safety reporting. For them it's existential, so the memory gets built and maintained at the level of the whole industry, not just the individual organization.
Nonprofits have no equivalent. The funder–nonprofit relationship looks like the obvious channel for sector learning, but in practice it makes the problem worse — and it's worth being precise about why.
Why the Funding Model Makes It Worse
When a funder gives money, it typically asks for two things: low overhead and measurable program outcomes. Both incentives push in exactly the wrong direction for organizational resilience. Low overhead means less money for finance staff, internal controls, and systems. Measurable outcomes means organizations optimize for reporting what happened rather than preventing what could go wrong.
Most people have heard of the "Nonprofit Starvation Cycle," named by the Stanford Social Innovation Review in 2009. As much as we wish things had changed, they haven't. Roughly 75 percent of donors still consider overhead ratio when choosing where to give, and foundation grants frequently cap indirect costs at 10 to 15 percent even when true costs run to 25 percent or more. The starvation cycle strips organizations of the capacity to build lasting systems — most small nonprofits spend less than $5,000 a year on all their accounting combined.
Turnover and the Knowledge That Keeps Leaving
But the deeper problem is turnover, and the turnover in this sector is devastating. With 23 percent annual staff turnover and executive directors lasting an average of five years, the people who carry what an organization has learned keep leaving. How do you enter a positive cycle of improvement under those conditions? The knowledge was there. It just couldn't be kept. I call it Nonprofit Sector Amnesia.
The go-to explanation is that the sector is under-resourced and under-regulated. That's true, but it's not the whole story — plenty of under-resourced sectors have built shared learning systems. The real question is what's different about nonprofits.
Independence Is the Enemy of Shared Memory
The answer is that nonprofits are, by design, independent — and independence is the enemy of shared memory. Every nonprofit is its own legal entity, with its own board, its own mission, its own community. This isn't a bug; it's what allows nonprofits to be responsive to specific communities rather than generic ones. The whole point is local specificity.
But that same independence means there's no natural way for operational knowledge to travel between organizations. No centralized body. No systematic way to learn from each other. No reason to report failures for the benefit of others — the way a pilot files an anonymous safety report for the benefit of every other pilot. This is where the usual playbook falls apart: you can't solve a shared-memory problem by asking each organization to build its own memory system it can't afford. The gap between what the sector needs and what any single organization can do isn't closable through more training, more grants, or more regulation.
Where Fiscal Sponsors Come In
This is where fiscal sponsors and intermediary organizations become interesting. A fiscal sponsor that manages dozens or hundreds of sponsored projects sees the same financial management problems repeat across the projects it supports. It has a natural vantage point for pattern recognition — the cross-organizational visibility that individual nonprofits can never have. These intermediaries are, in a sense, the technology.
But in practice, the problem too often gets abstracted to a disconnected tool stack and overstressed sponsor staff. Compliance gets centralized — and so does the risk. That model has a natural ceiling. Fiscal sponsors are absolutely part of the shift toward a better, more compliant sector, but they can't carry it alone. The whole sector needs a different kind of infrastructure: one that doesn't ask already-stretched staff to do more work.
Building Infrastructure That Remembers
What breaks traps like this, historically, is technology that drives the cost of the solution close to zero — not as an add-on, but built into the tools people already use. In practice that means financial controls that are on by default rather than added after something goes wrong, automated separation of duties that doesn't require a dedicated compliance officer to enforce, and transaction monitoring that runs continuously rather than surfacing problems twelve months after the fact.
But the technology alone isn't enough. The deeper unlock comes when built-in controls are combined with a community of organizations that feed their shared learning back through the software — so that a pattern detected in one organization quietly protects the next one before the problem even arrives. That's the break from the current trap. Nonprofits keep their independence and their ability to move fast toward the problems they exist to solve, but they stop learning in isolation, and they stop paying the same tuition over and over for lessons the sector already knows.
Every nonprofit that hits a stumbling block or a compliance failure learns an enormous amount, and right now nearly all of that learning is wasted. The sector keeps relearning the same lessons because it has no way to remember them — not because people don't care, and not because the problems are too hard. The question isn't whether the sector can afford to build a shared memory. It's how much longer it can afford not to.