The 90/10 Rule Is a Withdrawal Cap, Not an Expense Rule — and It Rewrites SGO Fundraising Math

Most organizations meet the 90/10 rule as an accounting requirement: spend at least 90% of income on scholarships, keep overhead inside 10%. That framing is accurate but incomplete, and the incomplete version leads planners to two mistakes — imagining an expense-allocation exercise that the rule never requires, and missing a fundraising constraint that it absolutely imposes.

The reframe that fixes both: under the safe harbor Treasury previewed in June 2026, the 90/10 test is a withdrawal cap on each state account. It governs what leaves the account, not where the organization's costs sit. Everything else follows from that.

What the Test Actually Governs

Under the safe harbor, income for the test is the amount held in a state's Section 25F segregated account — qualified contributions plus earnings. Of what is in the account, at least 90% must go out as scholarships to that state's students. Which means each account may release up to 10% of its contents for anything else.

Those releases leave the account and land in the organization's general operating funds. And once there, the money pools. Nothing in the statute or the previewed guidance requires a state's released 10% to be traced to that state's expenses. If one state's account releases $100,000 and another's releases $8,000, the organization has $108,000 in general funds and may lawfully spend it wherever operations require. The constraint is the cap on release, which binds per account — not the destination of the money afterward.

Two overlays temper this in practice: state charitable-solicitation law applies to how funds are raised and represented in each state, and donor expectations constrain what is wise even where the statute is silent. But the federal test itself is a withdrawal cap, full stop.

The Cap Binds Per Account — a Worked Example

Because money never moves between state accounts and the test never aggregates, the binding math happens state by state.

Say an SGO's account in State A holds $1,000,000 and its account in State B holds $80,000. State A can release up to $100,000; State B up to $8,000. Combined releasable funds: $108,000.

Now suppose the organization runs a $150,000 donor campaign covering both states. However the cost is allocated — by contributions, by population, by any reasonable method — the campaign cannot be funded from the accounts: the total available is $108,000, and each account's share of a proportionate allocation would exceed its own cap. The organization is $42,000 short before the first scholarship is affected, and the shortfall must come from somewhere else.

This is the arithmetic behind a rule of thumb that is not actually a rule of thumb: thin state accounts cannot carry their own costs. It is forced by the cap.

The 10:1 Hurdle

Generalize the example and you get the most important sentence in SGO fundraising economics: any fundraising spend that does not return better than ten times its cost — in the same state, within the same test period — cannot be funded from Section 25F money.

Run the failure case. Spend $20,000 on donor acquisition in a state; raise $40,000. A 2:1 return would delight most nonprofit development teams. But the $40,000 sits in the state account, and the account can release only $4,000. The organization is $16,000 underwater on a successful campaign — and the account cannot legally cover the difference. Raise nothing, and it is $20,000 underwater with no release at all.

Paid acquisition — direct mail, digital advertising — rarely returns 10:1 on first-year donors anywhere in the charitable sector. Inside the accounts, it is not inefficient; it is effectively unfundable.

What Clears the Hurdle

Channels with near-zero acquisition cost clear a 10:1 hurdle trivially: a school's bulletin, a principal's email to the parent list, an alumni newsletter, an announcement at a grandparents' day event. The gift arrives because trust and relationship already exist, and the marginal cost of the ask rounds to zero.

This is why school networks, associations, and faith communities are structurally advantaged as SGO operators. They already own the one asset the 90/10 economics reward: free, trusted distribution to people who care about the students being served. A standalone SGO planning to buy its donor base with a media budget is fighting the program's arithmetic; an organization activating existing relationships is working with it.

The Escape Hatch: Money That Never Enters the Accounts

None of this means an SGO cannot market itself. It means the funding source matters. Under the safe harbor, only what is held in the segregated accounts counts as income for the test. Money that never enters them — general operating gifts taken as ordinary charitable deductions, foundation grants for operations, sponsorships, a parent organization's support — is not in the denominator, and spending it does not move the ratio at all.

So the sustainable structure separates two gift types: qualified contributions flow to the state accounts and become scholarships; operating support flows to general funds and pays for staff, systems, and growth. The same donor can do both — $1,700 as a qualified contribution for the full credit, plus a separate operating gift deducted normally. Fundraising campaigns, launch costs, and anything that cannot clear 10:1 belong on the operating side. (Note the caution that comes with it: the safe harbor requires the organization's activities remain largely scholarship-granting, operating gifts must be genuinely separate with no quid pro quo, and large operating donors may still accrue substantial-contributor status under the disqualified-person rules.)

Three Caveats Before You Build the Budget

• The 90% is a spending obligation, not just a ceiling on release. Contributions sitting undisbursed in an account are their own compliance problem. The cap limits what can leave for operations; it does not excuse scholarships that never leave at all.
• Year one is the hard case. Formation, registration, systems, and launch campaigns are all spent before the first qualifying dollar arrives, and a partial first year of contributions cannot absorb them within 10%. IRS Notice 2025-70 asked whether the regulations should provide start-up relief or multi-year smoothing — the question is open, so model as if the answer is no and fund the launch from operating money.
• Fees eat the same 10%. Payment processing competes for the identical allowance, and on card rails it can claim a quarter of it. The payment-rail analysis covers the ACH-first playbook.

The Takeaway

Read as an expense rule, 90/10 looks like a bookkeeping burden. Read correctly — as a per-account withdrawal cap — it is a design constraint on the entire operating model: fund operations from outside the accounts, fundraise through channels that are already free, treat every state account as its own closed economy, and let the 10% releases be a supplement rather than the plan. Organizations that internalize this before launch build budgets that work. Organizations that discover it afterward build deficits. For the broader regulatory picture, start with our walkthrough of Treasury's June preview.