How to give well: the thinking behind Did it work?

About 25 minutes to read. Last reviewed October 2026.

#Why we wrote this

Most giving advice either tells you which charity to choose or tells you how to feel about it. We'd rather show you how we think, so you can judge our recommendations, or anyone else's, for yourself.

Did it work? exists for one reason. When you give, you're trusting someone else to turn your money into good you'll never see directly, and you deserve a clear, honest answer about who will do that best. We don't run a charity, take donations or sell placement. We bring together the published work of independent evaluators, show you where every number comes from, and say plainly what nobody knows yet.

This guide sets out the principles behind every page on this site. It starts from first principles: what a gift is actually for. It ends with a simple process you can use for a gift of any size.

#1. What a gift is for

When you buy something for yourself, you find out quickly whether it was worth the money. A gift to charity is different. The people who benefit are usually not the people who pay, and they are often thousands of miles away. Nothing in the transaction tells you whether it worked.

That makes a charitable gift an act of delegation. You are choosing a steward: an organization you trust to turn your resources into a particular kind of good, competently and honestly, when you can't watch. Seen this way, the job of a careful donor isn't to find the most appealing cause. It's to find the best steward for the outcome you care about.

Three things follow from that.

• Outcomes matter, not activity. A charity can be busy, well run and sincere, and still change little. What you're paying for is the change in people's lives, not the effort.
• Comparisons are unavoidable. Every gift to one place is a gift not made somewhere else. Choosing not to compare doesn't avoid the choice. It just makes it blindly.
• Honesty is part of the product. A steward who overstates results is harder to trust with the next gift. We hold ourselves to the same standard.

#2. The question that matters most: what would have happened anyway?

The single most important idea in effective giving is the counterfactual: the difference between what happens with your gift and what would have happened without it.

Suppose a charity distributes bed nets in a district where the government was going to distribute the same nets next month. The charity's reports would show nets delivered and families protected. But your gift changed very little, because the families would have been protected anyway. Now suppose the nets wouldn't otherwise have arrived for years. The same report describes a much larger difference.

The counterfactual matters in several places:

• The program itself. Would people have got the same help from someone else, such as a government, another charity or the market?
• The evidence. Did the people who were helped do better than similar people who weren't? Well-designed studies, especially randomized trials, compare like with like for exactly this reason.
• The money. If the charity was going to raise the money anyway, your gift may simply free up funds for something else it does. That isn't necessarily bad, but it means the real effect of your gift depends on the charity as a whole, not on the program you pictured.

This is why we lean on independent evaluators such as GiveWell and the Happier Lives Institute. Asking "compared with what?" is central to how they work, and they publish their reasoning so you can check it.

#3. Why the cost of a good outcome varies so much

Two charities can both be honest, well run and helpful, and still differ enormously in how much good a dollar does. That isn't a moral failing on anyone's part. It follows from how the world works:

• Problems differ in how cheaply they can be solved. Some illnesses can be prevented with an inexpensive medicine or a routine vaccine. Others need years of specialist care.
• Places differ in what's already been done. The first nets in a malaria-heavy region prevent far more illness than the last few in a region that already has many.
• Methods differ in how well they work. Some approaches have been tested repeatedly and work well. Others sound sensible but, when studied, achieve little.
• Delivery differs. The same intervention can cost very different amounts depending on how efficiently it reaches people.

Because these factors multiply together, the gap between a good option and an excellent one can be large. Here is what the evaluators we follow have published, as recorded in our data in October 2026:

Modeled, not guaranteed. These are evaluators' best estimates, not prices. Each links to its source in our research library. Our records mark the Pure Earth and Taimaka estimates as lower-confidence than those for the two therapy charities.

Two lessons come out of this table.

First, even among programs that evaluators already rate highly, the cost of a good outcome varies. Within the Happier Lives Institute's own comparison, its highest-rated option is estimated at roughly fourteen times the wellbeing per dollar of simply giving cash, which is itself a respected, well-studied way to help.

Second, the gap between the best and the merely good is usually wider than the gap among the best. The charities in the table have already passed demanding reviews. Most charities have never been assessed in this way at all, so nobody can say where they would fall. That's why we show "not yet rated" rather than a guess.

The practical consequence is simple. For many donors, choosing well can matter as much as giving more.

#4. Evidence versus a good story

Charities need to raise money, and stories raise money. A single photograph of a child who was helped is more moving than any table. But a story tells you that something good happened to someone. It doesn't tell you how often, at what cost, or whether it would have happened anyway.

When we look at the strength of evidence, we ask:

• Has the approach been tested in a way that compares like with like? Randomized trials and well-designed natural experiments are the strongest tools for this.
• Has it been tested more than once, by people with no stake in the result?
• Does it apply here? A result from one country, decade or delivery model may not carry over to another.
• What do the evaluators themselves say is still uncertain?

That last question is one we always put on the page. For example, for the top charities on our homepage, the evaluators flag questions such as how long bed nets keep protecting people and how insecticide resistance affects them (Against Malaria Foundation), whether vaccine efficacy is lower in Nigeria than in other settings (New Incentives), and whether vitamin A still has the effect seen in older trials now that child health has improved (Helen Keller Intl). The Happier Lives Institute notes that some of Friendship Bench's trials carry a risk of bias. None of these are reasons to dismiss the charity. They're the honest edges of what is known, and you should see them.

Each charity on our site carries a plain label for the strength of its evidence, taken from the evaluator's own assessment. Where an evaluator hasn't rated something, we say so.

#5. Can the charity use more money well?

A charity's published cost per outcome is usually an average over a past period. Your gift is a next dollar. They're not the same thing.

A charity can only absorb so much money before it runs out of the most valuable places to spend it. Beyond that point, extra funding may go to less effective work, sit in reserve, or simply replace money another donor would have given. Evaluators call this room for more funding.

For most donors this isn't a worry. For larger gifts, it matters a great deal. We show room for more funding only when an evaluator or charity has published it with a source. Where nobody has, we say "unknown", which is not the same as "no room". Some evaluators publish funding gaps for their recommended charities; others don't publish a live figure for each one.

If you're planning a large gift, three sensible responses are to:

1. check the evaluator's most recent grant pages for the charity;
2. split the gift, or give over time; or
3. use an evaluator's fund, such as the GiveWell Top Charities Fund or the Happier Lives Fund, which direct money to where their researchers judge it is most needed at the time.

#6. The overhead myth

Many donors judge charities by how much of each dollar goes to "programs" rather than "overhead". It's an understandable instinct, and an unreliable guide.

• A low overhead ratio doesn't tell you that anything worked. A charity can spend almost everything on programs and still achieve little if the programs are weak.
• Some overhead is how good work gets done. Monitoring, data, skilled staff, audits and careful evaluation all count as overhead, and they're often what separates a charity that knows it's working from one that only hopes so.
• The ratio is easy to manage on paper. How costs are classified varies between organizations.

The better question is the one evaluators ask: what does the whole operation achieve for each dollar it spends? A charity that spends more on measurement and achieves far more per dollar is the better steward, whatever its overhead ratio.

#7. Uncertainty and humility

Every impact figure on this site is modeled, not guaranteed. A figure such as "$3,500 per life saved" is an evaluator's best estimate of the average cost over a stated period. It is built from studies, surveys and many assumptions. Your gift could achieve more or less, funding needs change, and evaluators revise their numbers as they learn.

That isn't a weakness of the evidence. It's what honest evidence looks like. We hold to a few rules because of it:

• We never invent a number. If a recognized evaluator hasn't published a figure, you'll see "not yet rated". We don't estimate, extrapolate or round a gap into a number to make a page look complete.
• We show where each number comes from. Every figure is cited to the body that published it (GiveWell, the Happier Lives Institute, Animal Charity Evaluators, WHO or IHME), with the period it covers.
• We keep doubt visible. The evaluator's main open question sits next to the headline figure, not in a footnote.
• We expect to be wrong sometimes. When the evidence changes, we change with it, and we keep a record of what was said before (see section 13).

Humility also cuts the other way. Uncertainty isn't a reason to give nothing, or to give at random. A well-studied estimate with honest error bars is far better guidance than none at all.

#8. Choosing between causes when values differ

Some differences between charities are matters of fact: how much a program costs, how well it works. Others are matters of value, and no amount of research will settle them for you.

Is saving a child's life worth more or less than lifting an adult out of severe depression for several years? How much should the welfare of animals count? How should we weigh help today against reducing risks to future generations? Thoughtful people disagree, and the evaluators themselves take different views. GiveWell publishes the moral weights it uses within its own models. The Happier Lives Institute explains why it doesn't take a settled position on how to weigh quality of life against length of life.

Our approach is simple: these choices are yours.

• We keep different kinds of good on separate lists. Lives saved and wellbeing (measured in WELLBYs: one point of life satisfaction, on a 0–10 scale, for one person for one year) are shown side by side, never folded into a single score.
• We never invent a conversion rate between kinds of good, such as how many units of wellbeing equal one life.
• If you care about more than one kind of good, you can split your giving between them. That's a perfectly reasonable way to act on values you hold together.

Today our rankings cover global health and wellbeing, anchored on GiveWell and the Happier Lives Institute. We're researching other causes, including animal welfare and climate, but we don't rank them until the evidence and our approach to them have been reviewed and approved.

#9. Being a good steward of your own resources

Choosing where to give is half the job. Choosing how to give can change how much reaches the charity. The routes below are general guidance only. Your advisor knows your situation, and rules vary by country and by person.

Cash. The simplest route. Give on the charity's own website, or give to an evaluator's fund if you'd rather an expert allocate across the top charities.

Appreciated stock. Gifting long-held shares directly, instead of selling them first and donating the proceeds, may avoid capital gains tax and so let more reach the charity. Not every charity accepts securities, so confirm first. GiveWell publishes instructions for securities.

Donor-advised funds (DAFs). A DAF lets you contribute now and recommend grants later. Any deduction generally applies when you fund the DAF, not when a grant goes out. Your sponsor will need each charity's legal name and EIN; we list these on our homepage. A DAF works best as a holding place on the way to good work. Money that sits in a DAF indefinitely hasn't yet helped anyone.

Retirement accounts. Eligible US donors may be able to give directly from an IRA through a qualified charitable distribution. The rules are specific, so this is a conversation for your advisor and account custodian.

Cryptocurrency. Some evaluators accept crypto through wallets they publish themselves, such as GiveWell. Only ever send to an address the recipient has published on its own site, because transfers can't be reversed.

Bequests. Leaving a gift in your will is one of the most generous things a person can do, and it calls for one extra layer of thought: the best charity today may not be the best one when your bequest is paid. Consider naming a purpose, or an evaluator's fund that will re-decide when the time comes, alongside or instead of a single charity, and work with your estate attorney on the wording. Use the charity's exact legal name and EIN.

Time, skills and introductions. Valuable when a charity has asked for them. They're usually not a substitute for funding when money is what's holding the work back, and they're generally not tax-deductible as if they were cash.

We never take, hold or process money. There's no checkout on this site. Every gift goes directly from you, your bank or your fund sponsor to the charity.

#10. Concentrate or spread?

Many donors instinctively spread their giving across many charities. It feels prudent. But diversification works differently in giving than in investing.

When you invest, you spread money around to protect yourself against loss. When you give, the "return" goes to other people. If you've done the work and one option looks clearly best for the outcome you care about, every dollar moved to a weaker option does less good.

There are still good reasons to spread a gift:

• Your values point in more than one direction. Splitting between lives saved and wellbeing, for example, is a principled way to honor both.
• Room for more funding. A large gift may exceed what the best option can use well right now.
• Real uncertainty. If the evidence can't separate two strong options, splitting between them is reasonable.
• Learning. A smaller gift to a promising but less-proven program can be worthwhile, as long as you treat it as the higher-risk choice it is.

An evaluator's fund is, in effect, a way to concentrate on "the best available" while letting specialists handle the spreading.

#11. Give now or give later?

There's no single right answer, but the question is worth thinking through.

Reasons to give now: problems that are solved earlier often stay solved. Cheap, proven opportunities may be taken by other funders over time. Health gains today can compound into better lives tomorrow. And intentions to give later are easy to postpone indefinitely.

Reasons to give later: you may know more, the evidence may improve, your resources may grow, and better opportunities may appear.

A reasonable middle path for many families is to give steadily each year, review the evidence each time, and use a DAF or a bequest for the share you deliberately set aside for the future, with a clear plan for when it will be granted.

#12. Red flags

None of these proves a charity is bad. Each is a reason to ask more questions before you give.

• Numbers without a source or a date. If you can't see where an impact figure came from, you can't check it.
• Promises. "Your $X will save a life" overstates what anyone can know. Honest estimates come with caveats.
• Results that are only stories. Plenty of photographs and testimonials, but no comparison with what would have happened anyway.
• Pride in low overhead in place of evidence of results.
• Urgency and pressure. "Give in the next hour" is a sales technique, not evidence of need.
• Rankings that can be bought. Ask whether a list takes payment for placement. We don't, and nothing marked sponsored or featured can ever change a ranking on this site.
• Payment details from an unexpected source. Only give through the charity's own website, an evaluator's fund or your fund sponsor. For crypto, only use an address published by the recipient.
• Reluctance to share evidence or answer questions. Good stewards welcome scrutiny.

#13. Did it work? Checking back

Our name is a promise to keep asking the question after the gift is made.

Most giving advice stops at the moment of donation. But evaluators revise their estimates, programs change, and new studies arrive. So we keep a public, dated record of exactly what each evaluator said, with a scheduled date to check again. Each entry carries a digital fingerprint linked to the one before it, so a quiet edit would show. If the estimate changes, the new view is added alongside the old one, and the history stays visible.

The record is new and deliberately modest. Today it holds nine entries, each copied from an evaluator figure we already cite, for charities including Against Malaria Foundation, Malaria Consortium, Helen Keller Intl, New Incentives, StrongMinds and Friendship Bench. The first re-checks are scheduled for after 1 October 2027. Until then every entry is marked pending, which means "not yet re-checked". It doesn't mean the program failed or succeeded.

When the re-checks happen, each entry can come back confirmed (a later evaluator source re-affirmed the estimate), contradicted, or unverifiable. Even "confirmed" means the evidence held up, not that we measured the outcome ourselves.

You can do the same for your own giving:

1. Write down what you expected. Note the figure and source you relied on, and the date.
2. Set a reminder for a year from now.
3. Look again. Has the evaluator updated its view? Did the charity publish results, including disappointing ones? Has its room for more funding changed?
4. Adjust. Giving well is a practice, not a single decision.

#14. A simple decision process

Whether you're giving $1,000 or $10 million, the same steps apply. Only the depth changes.

1. Decide what kind of good you care about most. Lives saved, wellbeing, income, or something else. If more than one, decide roughly how you'd like to split.
2. Set your red lines. Causes, methods, places or organizations you won't support, whatever the numbers say.
3. Start from independent evaluators. Begin with charities they've assessed, or with their funds if you'd rather not choose individual charities.
4. Look at cost per outcome, evidence strength and the open questions together. A slightly higher cost with much stronger evidence can be the better choice.
5. For larger gifts, check room for more funding. Read the latest grant pages, consider splitting, or use a fund.
6. Choose the best route for your gift. Cash, appreciated stock, a DAF grant, a retirement-account distribution or a bequest. Confirm the details with your advisor.
7. Give directly. On the charity's own site, through an evaluator's fund, or through your fund sponsor, using the charity's legal name and EIN.
8. Write down what you expected, and check back in a year.

#15. Questions to ask any charity

Use these with any organization, whether or not it appears on our site. A good steward will be glad you asked.

1. What outcome are you trying to change, and how do you measure it?
2. What would happen to the people you serve if you didn't exist?
3. What's the strongest evidence that your approach works? Has anyone independent tested it?
4. What does it cost you to achieve one unit of that outcome, and how did you calculate it?
5. Has an independent evaluator assessed you? What did they say, including the criticisms?
6. What's the biggest thing you're uncertain about?
7. What would you do with an extra gift of this size this year? Is there a funding gap, and how big is it?
8. Have you ever stopped or changed a program because it wasn't working?
9. How will you report back to me, and will you share disappointing results as well as good ones?
10. Can I give appreciated stock, or a grant from my DAF? What legal name and EIN should I use?

#16. How to use Did it work?

• Start on the homepage. It shows the charities independent evaluators rate highest for saving lives and for mental health, each with its published cost per outcome, its source, the strength of the evidence and the evaluator's main open question.
• Get a sense of scale. The calculator on the homepage divides your gift by each evaluator's published cost per outcome. That's all it does, and it's a sense of scale, not a promise.
• Read the details. Every charity we track has a profile in our research library, with its sources and the date it was last reviewed.
• Find the paperwork. Legal names and EINs for grant forms are on the homepage. Ways to give are under Ways to give.
• Check our rules. Our standards explain exactly how we choose charities, where our numbers come from, and what we'll never do. The FAQ answers the questions donors ask most.
• Use your own AI assistant if you like. Everything on this site is also published in a form AI assistants can read and cite. If an assistant quotes us, it should give you the same sources and caveats you see here.

#What we will never do

• Invent a cost-effectiveness figure, or fill a gap with a guess.
• Present an estimate as a guarantee.
• Let anyone pay to change a ranking, a figure or a confidence level.
• Combine different kinds of good into one score, or decide your values for you.
• Take, hold or process your money.
• Tell you a gift is tax-deductible for you.

A human editor approves every change to which charities we list and how we rank them. Our software does the legwork of gathering and checking. It doesn't decide what matters.

#Sources

Figures in this guide are copied from evaluators' published pages, as recorded in our data in October 2026:

• GiveWell, cost per life saved, 2022–2024 averages: Helen Keller Intl, Malaria Consortium, New Incentives, Against Malaria Foundation.
• Happier Lives Institute, wellbeing per $1,000 and multiples of cash transfers: StrongMinds, Friendship Bench, Taimaka, Pure Earth, and the charities comparison (GiveDirectly benchmark).
• Open questions: each evaluator's review page, linked from the charity profiles in our research library.

Did it work? is an independent research project. We are not affiliated with any charity we list, with any evaluator, or with any donor-advised fund sponsor. Estimates are modeled, not guaranteed. Nothing here is tax, legal or investment advice.