How much should a nonprofit spend on fundraising?

Your finance committee read somewhere that fundraising should stay under a third of what it brings in. Your development director wants a second gift officer and a bigger mailing. Both sides quote numbers, and the numbers answer different questions.

Three standards get quoted, and only one of them is a budget

The watchdogs publish ceilings. The BBB Wise Giving Alliance caps fundraising at 35% of related contributions, CharityWatch draws its highly efficient line at $25 spent per $100 raised, and Charity Navigator gives full credit at 20 cents per dollar of contributions. Those are tests you pass, not plans you write. Form 990 filings give medians, and the tables below have them by budget band. Those tell you where you sit. Only the third question sets a budget, and nobody publishes it: does the next dollar you put into a given line bring back more than a dollar. Take the three in that order and expect the last one to move your number most. This piece sits inside our fundraising strategy coverage, which settles the revenue mix the budget follows from.

What the three watchdogs actually require

The three agencies do not measure the same thing, which is why a search on this question returns three different percentages attributed to the same handful of organizations.

Sources: BBB standards 8 and 9, the CharityWatch rating process, and Charity Navigator’s Rating Methodology Guide.

Two differences matter more than the numbers. The first is the denominator. CharityWatch divides by related contributions, the money fundraising actually brought in, and says plainly that dividing by total revenue “often has the effect of making a charity appear to be a more efficient fundraiser than it actually is.” Charity Navigator divides by total contributions on Part VIII line 1h, which sweeps in government grants and donated goods that no fundraiser solicited.

The second is that all three are ceilings. Not one publishes a floor. An organization that spends nothing on fundraising because it has no fundraising program passes every watchdog test with room to spare.

What Form 990 filers report, by budget band

The IRS Statistics of Income annual extract, the usual source for Form 990 benchmarks, cannot answer this question. Its data dictionary carries Part IX column (A) only, so it holds no fundraising expense, and ProPublica’s Nonprofit Explorer API republishes the same fields with the same gap. These tables come from the IRS Form 990 series e-file downloads instead. Every Form 990 in the calendar year 2024 and 2025 release files was read, 739,997 returns in all, then filtered to tax year 2023 filings from 501(c)(3) organizations with total functional expenses of at least $1M and contributions above zero, and collapsed to one filing per organization. That leaves 85,514 organizations. Fundraising is Part IX line 25 column (D), total expenses is column (A), contributions is Part VIII line 1h.

Read the spread, not the median. Inside a single band the gap between the quartiles is several times the gap between bands, so budget size explains almost nothing about fundraising spend. Revenue mix explains it, and the filing does not report revenue mix.

Two in five report no fundraising expense at all

The most useful number here is not a median. It is the share reporting a zero, which runs from 37.7% in the $3M to $10M band up to 46.7% above $50M. Those organizations have at least $1M in expenses and contributions on the books, and they are telling the IRS that raising those contributions cost nothing.

An organization that spends nothing on fundraising because it has no fundraising program passes every watchdog test with room to spare.

The share also rises with budget size, which is not what an understaffed development office looks like. Composition explains it. Above $50M, filers reporting zero book contributions worth a median 1.5% of their expenses. Filers in the same band that report a cost book 21%. Half of what the zero reporters book is government grant money, and 42% of them operate a hospital. They are institutions funded by fees and government, not by donors.

The Form 990 instructions explain how. Section 501(c)(3) organizations must complete Part IX columns (A) through (D), but on allocation the instructions say that if the accounting system does not allocate expenses, “the organization can use any reasonable method of allocation.” Reasonable is doing a great deal of work in that sentence. A development director’s salary can sit entirely in program services if someone decided those donor visits are community outreach. The same instructions send the direct costs of a gala to Part VIII line 8b rather than Part IX, so the catering and the venue never touch the fundraising column.

Two things follow. Your own ratio is comparable to the benchmark only if your accountant allocates the way most filers do, and you cannot know that without asking. And any ratio built on these columns, including the ones on this page, measures accounting practice as much as fundraising.

Cost per dollar raised, and which dollar you are counting

Among filers that do report a fundraising cost, here is what a dollar of contributions costs them. The second column uses the same denominator Charity Navigator uses. The third strips out government grants on line 1e and donated goods on line 1g, which is the closest Form 990 gets to the related contributions CharityWatch insists on.

The two columns tell opposite stories. On total contributions, cost per dollar barely moves, seven cents in the first three bands and eight in the largest. On private cash, it climbs steadily with size, from eleven cents to nineteen. The first column stays put because large organizations carry large government grants that cost nothing to raise. Strip those out and the pattern reverses: the bigger the organization, the more it pays for each dollar an actual donor gives it.

That also settles an apparent disagreement in the published figures. Marts and Lundy’s Fundraising Investment Study, published 27 January 2026 from 243 organizations across seven countries, reports a median cost per dollar raised of $0.18 and a median return of $4.50 for every dollar invested in fundraising. That is well above the seven cent Form 990 median, and within a penny of the nineteen cents the largest Form 990 band pays for private cash. Both numbers are right. They count different dollars.

Channel matters more than either. AFP puts direct mail acquisition at $1.00 to $1.25 per dollar raised and a renewal mailing to those donors at about $0.20, with events returning less than major gifts and a new planned giving program possibly returning nothing for years. Compute your figure per line. A blended number hides an expensive gala behind a cheap major gifts program.

The 33 percent rule is a public support test, not a spending cap

This one surfaces in every board discussion and it is not about spending at all. Under the Schedule A instructions, a public charity has to show that at least 33 1/3% of its support over five years came from governmental units, the general public, and other public charities. Whatever a single donor gives above 2% of total support is stripped out before the count. A charity that misses 33 1/3% can still qualify at 10% if the facts and circumstances support it. It tests where your money comes from, and failing it costs you your public charity status. It says nothing about what you may spend to raise it. If someone cites the 33 percent rule against a fundraising budget, they have the wrong rule.

The more common failure is spending too little

Every published standard is a ceiling, so the whole apparatus pushes one way. Nothing in it tells a board the development office is understaffed, and boards read ratios far more often than they read pipeline reports.

The corrective is the marginal question. AFP’s acquisition figure is the clearest case: the first mailing to people who have never given loses money by design, and the donor file it builds pays for it over the following years. Judged on the ratio alone, acquisition is the first line a finance committee cuts. Judged on what the next dollar returns across three years, it is often the last. AFP’s own advice is that boards “should determine a reasonable rate of return on investment for their own organization based on prior results”, which is a polite way of saying the benchmark cannot do this for you.

So run the test on one line before setting next year’s number. Find what you spent on that line last year and what it raised, then estimate what one more increment would return. If the answer is above a dollar and the money renews, the ratio is the wrong thing to look at.

A worked example at $6M

A hypothetical human services organization with $6M in expenses, 45 staff, and a development team of two. Every figure below is invented and carries no source.

Three ratios come out of that, and they disagree.

Nothing about the fundraising operation changed between those three lines. Only the denominator did. Compute all three, label the denominator each uses, and never quote one alone.

Then the line that actually sets the budget. Last year this organization moved $40k out of the gala into a monthly giving push. It returned $96k in first year recurring gifts, $2.40 on the dollar, from donors who renew. The last $40k of gala cost returned $52k gross and consumed 400 staff hours. That comparison decides next year’s budget. The ratio does not.

Three lines for the board

Boards ask for a percentage because a percentage looks like accountability. Give them three lines on one page, once a year, at budget approval.

The third line is the one that changes the conversation, because it is the only one that implies a decision. Put the benchmark in a footnote with your band’s median and the note that two in five filers report nothing, so nobody reads a median as a target. What each board member owes the effort is a separate document, and Board member fundraising expectations that hold has the wording. The revenue lines this budget attaches to belong in the fundraising plan, where each one already carries a target and an owner. If the board wants a second ratio to worry about, the technology spend benchmark is built the same way from the same filings.

When not to do this

Skip the benchmark comparison entirely if your fundraising column has been zero or inconsistent for three years. It will flatter or alarm you for reasons that have nothing to do with your fundraising. Ask your accountant how the allocation is done at the next close, fix it, and compare a year later.

Skip it during a capital campaign and for the two years after. Campaign costs land in one or two fiscal years while the gifts land across five, so every ratio will be wrong in a direction that changes annually. Track campaign cost against campaign goal separately until the campaign closes.

Do not run the marginal test on a line younger than two years. A first year of monthly giving, a first planned giving mailing, and a first grant writer all look like failures on a twelve month view, which is why AFP flags the new planned giving program as potentially returning nothing at first. Give a new line three years or do not start it.

And do not use any of this to settle whether to hire a development director. A ratio cannot answer that. It can only tell you, once the person is in post, whether the year they arrived will look unusual to a board member with a calculator.