Comparing Charities: How Big Is the Difference?

Core claim

Donating somewhere is not equivalent to donating anywhere: charities and interventions differ so much in evidence, cost, target population, leverage, and scale that many donors may be able to increase their impact by 100× or more. Good intentions and low overhead are weak guides; donors should instead use independent impact evaluation, seek the places and programs where marginal resources help most, look for leverage, and reconsider whom their moral concern includes.

The headline argument

• Most charities may be well-intentioned, but their results per dollar vary enormously.
• Giving What We Can’s research team believes many donors could “easily” increase their impact 100× by switching toward charities that achieve more per dollar.
• There is no universal multiplier. The difference depends on:
  • the donor’s starting charity;
  • the kind of outcome they value;
  • the quality and age of the available estimates;
  • whether the compared outcomes are genuinely commensurable.
• The key practical point is robust even if a specific ratio is wrong: where a donation goes can matter at least as much as how much is donated.

Three opening comparisons

Donation	Alternative	Claimed difference in outcome
$15 to Toys for Tots	$15 to Evidence Action’s Safe Water Program	One toy for one child vs. one year of clean water for ten children
$2,670 to the Navy SEAL Foundation’s education program	$2,670 to GiveDirectly	Increase one scholarship recipient’s income by $3,892 vs. cover roughly one year of expenses for two households in extreme poverty
$1,500 to Battersea Dogs & Cats Home	$1,500 to a corporate cage-free campaign such as The Humane League’s	Cover roughly half the annual shelter cost of one cat or dog vs. improve conditions for around 15,000 egg-laying hens

• The comparisons are meant to make opportunity cost vivid, not to claim that toys, scholarships, or companion-animal shelters do no good.
• The page adds that GiveWell’s top charities were estimated to be at least twice as cost-effective as GiveDirectly, and cited an average cost of roughly $5,000 to save a child’s life.
• A 2023 study is cited as evidence that donors substantially underestimate variation between charities, consistent with limited performance-based giving and underfunding of some highly cost-effective interventions.

Why charity impact varies so much

Programs and costs differ

• Charities choose different problems, populations, and delivery methods.
• Even organisations addressing the same goal may pay very different amounts for the same outcome or achieve outcomes of different quality.
• As in business, performance has a long tail: a few interventions can outperform typical ones by orders of magnitude.

The charitable “market” has a broken feedback loop

• Consumers personally experience a product’s quality and can stop buying it; firms that consistently fail customers tend to lose revenue.
• A donor usually does not experience the benefit. The beneficiary does, often far away and with no direct way to reward or punish the funder.
• Without independent evaluation, the donor mainly sees the charity’s own marketing.
• A charity can therefore continue attracting funds by telling a compelling story even if its program has little impact. This weak feedback should produce greater variation and inefficiency than in ordinary consumer markets.
• Yet donors often do less research for charity than for personal purchases. The page cites a study in which only one-third of donors researched before giving and only 3% based donations on performance.

Five ways to multiply impact

The article presents each rule of thumb as potentially worth around a 10× improvement from some starting points. The effects overlap, so they cannot simply be multiplied into a 100,000× claim, and not every donor will receive the full gain. But applying even two sufficiently independent principles could plausibly produce a 100× difference.

1. Choose independently evaluated, impact-focused programs

• An impact evaluation asks what a program actually accomplishes, how reliably it produces the outcome, and how much outcome is achieved per dollar.
• This differs from ratings centred on overhead or administrative ratios. Low overhead does not establish that a program works, while some effective programs may need substantial evaluation or operational capacity.
• Independent evaluation is important because plausible stories and prestigious endorsements do not substitute for evidence.

PlayPumps as the cautionary example

• PlayPumps installed playground-style roundabouts intended to let children pump water while playing.
• The concept won a World Bank award, celebrity attention, and millions in US aid; roughly 1,800 pumps were installed before major problems became widely visible.
• Children did not enjoy the sustained labour, some became injured or sick from spinning, and women often had to operate the pumps.
• Manual pumps were reportedly easier to use and produced five times as much water per hour.
• Each PlayPump cost about $14,000. One outside estimate found children would need to operate the planned network for 27 hours per day to hit its water target.
• For the price of one pump, the article estimates that Evidence Action’s evidence-backed safe-water program could have served more than 9,000 people for a year.
• Lesson: an appealing intervention can absorb large resources while underperforming a less photogenic alternative. The true cost includes the better intervention that was not funded.

2. Help where money can help the most

• Geographic income differences mean the same amount of money can produce much larger relative gains in poorer places.
• The page compares a roughly $40-per-day US poverty line with a $2.15 global extreme-poverty line, adjusted for local prices, and infers that doubling the income of a poor American could cost at least 10× as much as doubling that of someone near the global poverty line.
• The mechanism is diminishing marginal utility of consumption: an extra $1,000 changes far more for someone earning $1,000 than for someone earning $100,000.
• Relative rather than absolute income gains may better approximate welfare improvements. Giving to people in extreme poverty can therefore plausibly yield a 10× increase over otherwise similar support in a high-income country.
• The same logic applies across species and animal populations:
  • about 99.6% of animals killed in the US are farmed animals;
  • the page reports that roughly 66% of animal-focused donations go to shelters and only 0.8% to farmed-animal organisations;
  • corporate campaigns can consequently affect far more animals per dollar than companion-animal rescue.
• The argument is not that local poverty or cats and dogs do not matter. It is that impartial donors should consider directing at least some funds to beneficiaries for whom the marginal benefit is much greater.

3. Find the most cost-effective program within a cause

• Cause selection alone is insufficient. Interventions aimed at the same problem also differ greatly.
• Impact-focused evaluators systematically compare programs, evidence, room for more funding, and marginal cost-effectiveness.
• GiveWell uses GiveDirectly-style unconditional cash transfers as a benchmark and recommends opportunities it estimates to be at least twice as cost-effective.
• Combining the page’s rough 10× geographic effect with a further 2× intervention improvement yields its illustrative claim that a GiveWell top charity may be at least 20× more impactful than a charity alleviating poverty in the US.
• The calculation depends on taking GiveWell’s model and the cross-country welfare comparison at face value, so it should be read as an order-of-magnitude illustration rather than a universal exchange rate.

4. Look for leverage points

• Direct service spends the donated resource once. A leverage strategy changes how a much larger pool of government, corporate, investor, or donor resources is used.
• Leverage is only valuable when there is a tractable route to influence; it is not automatically better than direct delivery.
• Examples:
  • Lead Exposure Elimination Project helps governments implement lead-paint regulation, leveraging public authority and budgets.
  • Deworming organisations work through governments, NGOs, and investors to deliver treatments.
  • The Humane League’s campaigns change corporate purchasing commitments, using company supply chains to improve conditions for hens.
  • Effective-giving organisations influence donations already intended for charity; the page says current evidence suggests that $1 spent on some such organisations may move $10 or more to high-impact charities.
• The article notes that malaria bednet distribution may currently be a case where directly funding the Against Malaria Foundation is more sensible than searching for an indirect multiplier.

5. Expand the circle of concern

• The largest shift may come before charity comparison: examine which beings and times are included in one’s values.
• Physical distance need not reduce a person’s moral importance; the drowning-child thought experiment is invoked to challenge the idea that unseen suffering counts less.
• If sentient non-human animals matter, factory farming brings vastly more beings into the calculation.
• If future people matter, moral concern may extend across time just as it extends across space.
• Including animals or future generations can direct donors toward cause areas much larger in scale and more neglected than familiar local problems.
• This is partly a moral argument rather than a purely empirical cost-effectiveness claim. The impact ranking changes with one’s worldview and with how different kinds of welfare are compared.

Detailed case studies

Toys for Tots vs. safe water

• Toys for Tots reported spending about $359.9 million to distribute 24.4 million toys, books, and games to 9.9 million children in the fiscal year ending December 2022.
• Dividing spending by items distributed gives a rough cost of $15 per item.
• Evidence Action estimated its Safe Water Now program could provide safe water for $1.50 per person per year.
• On those figures, the same $15 buys one gift or safe water for ten people for a year.
• The outcomes differ in kind—holiday happiness vs. protection from waterborne disease—so the comparison is rhetorically vivid but not a clean like-for-like cost-effectiveness ratio.

Navy SEAL Foundation vs. GiveDirectly

• Charity Navigator reported that $2,670 to the Navy SEAL Foundation’s scholarship work increases one recipient’s income by $3,892.
• GiveDirectly transfers cash to households in extreme poverty; using the cited delivery ratio, $1,250 in donations generates about $1,000 in transfers.
• The page treats $1,000 as roughly one year of expenses for an average recipient household. A $2,670 gift could therefore cover about a year for two households.
• The nominal $3,892 income gain in the US is larger, but represents only around a month of expenses in many US locations. The comparison relies on the greater welfare value of relative income gains at low income.
• The scholarship may yield educational benefits beyond current income, but GiveDirectly transfers also show benefits beyond cash received, including educational outcomes. Neither side is fully represented by a single income figure.

Battersea vs. cage-free campaigns

• Battersea reported daily care costs of about £46,000 and approximately 7,000 cats and dogs served annually.
• Converting the annual total to roughly $21 million and dividing by animals served gives a back-of-the-envelope annual average of $3,000 per animal; $1,500 therefore funds about half of that cost.
• The estimate is rough because animals spend different lengths of time in care, and institutional costs are not necessarily marginal costs.
• A Rethink Priorities estimate suggested each dollar given to highly effective corporate campaigns improved conditions for 9–120 hens. The page deliberately uses the low end—about ten—to reflect uncertainty and declining campaign cost-effectiveness.
• On that assumption, $1,500 affects about 15,000 hens for their roughly one-year lives by helping eliminate battery-cage confinement.
• Cage-free systems still involve other factory-farm suffering. The relevant comparison is not a fully good life for 15,000 birds versus half a rescued animal, but removal of one severe harm for many animals versus shelter care for part of one animal’s cost.

How to read the numerical claims

• Most figures are back-of-the-envelope calculations, not controlled comparisons or permanent prices.
• Average institutional expenditure is not always the cost of serving one additional beneficiary.
• Charity cost-effectiveness changes as the best opportunities are funded, programs scale, and external conditions shift.
• Several examples compare unlike outcomes, requiring moral judgments about health, income, pleasure, and animal suffering.
• Multipliers overlap. An evaluator may already select globally targeted, cost-effective, and leveraged programs, so counting each feature as an independent 10× gain would double count.
• The page’s strongest conclusion is ordinal—some options are much better than others—not that every donor can obtain a precisely measured 100× gain.

Practical checklist for donors

1. Decide what outcomes and beneficiaries matter to you, including whether distance, species, or time should change their weight.
2. Look for independent evaluations focused on outcomes and marginal cost-effectiveness, not only financial ratios or reputation.
3. Compare both causes and programs within causes.
4. Ask what an additional donation will accomplish, rather than dividing total spending by total historical output without adjustment.
5. Check whether the charity has room to use more money effectively.
6. Consider whether direct aid or a credible leverage strategy offers more impact at the current margin.
7. Inspect uncertainty, assumptions, and possible double counting before trusting a multiplier.
8. Revisit the choice as recommendations, prices, and funding gaps change.

Takeaways

• Charity is a high-variance domain with unusually weak market feedback, making independent impact research especially valuable.
• A compelling mission, celebrity endorsement, good intentions, or low overhead does not establish impact.
• Large gains can come from several levels: choosing a larger and more neglected problem, directing aid where marginal resources matter more, choosing a better intervention, and influencing larger pools of resources.
• Opportunity cost is the moral engine of the argument: funding a merely good option means forgoing what an exceptional option could have achieved.
• Effective giving does not require believing the page’s exact 100× estimate. Even much smaller differences make charity selection consequential.

Notes & Observations

• The page mixes two projects: empirical charity comparison and expansion of the moral circle. The first asks which means best achieve a goal; the second can change the goal and which beneficiaries count.
• Its strongest comparisons are those with similar outcomes, such as relative income gains. Cross-outcome comparisons are useful for intuition but require explicit values that cannot be settled by cost data alone.
• The PlayPumps example illustrates why “innovation” should be treated as a hypothesis to test, not as evidence of effectiveness.
• Some figures use 2022–2024 data, while the page currently points readers to 2026 recommendations. Donors should use the examples to understand the reasoning and consult current evaluators before acting.