Anya Samek and Chuck Longfield ran field experiments with public television stations and a national nonprofit in which 500,000 new donors were randomized to receive a thank-you call or not, published in American Economic Journal: Applied Economics in 2023. They also asked fundraising professionals to forecast the result first. The professionals predicted the calls would raise donor retention by 80 percent. What the data showed was, in the authors' words, "a precisely estimated null effect of calls on donor retention." Not a small effect. Not an underpowered maybe. A well-measured nothing, nada, zilch.

Then the handwriting question. Maja Adena, Steffen Huck and Levent Neyse ran two preregistered field studies with a leading German opera house across 10,000 donors, published in Experimental Economics in 2025. One study tested a "thank you in advance" line inside the appeal. The other tested handwritten thank-you postcards against printed ones. Neither moved contributions. Two different gratitude strategies, one of them the most sacred artifact in stewardship, both flat. These were single organizations in single contexts, so no one should generalize to every file. But the direction of the evidence is consistent and it is not the direction the sector assumes.

Here is the part that makes this interesting rather than depressing. Gratitude expression does reliably change behavior, and the cleanest demonstration of it happens to have been run inside a fundraising shop. Adam Grant and one of my old teachers, Francesca Gino, in Journal of Personality and Social Psychology in 2010, had a manager visit university fundraising callers and thank them for their work. The callers in the gratitude condition made roughly 50 percent more calls the following week. The mechanism was not confidence in their own ability. It was social worth, the sense of being valued by another person. Gratitude worked. It just worked on the person doing the giving, in a moment where being valued was the thing at stake.

The reward response to being thanked is not a response to information. It is a response to being held in someone's regard, which is why oxytocin and trust show up in this literature and why the felt quality of an acknowledgment matters more than its accuracy. A gift receipt with a warm sentence attached carries the information. It rarely carries the regard.

There is a second circuit running underneath, and it does not feel good. Xiaoxue Gao and colleagues modeled indebtedness across questionnaires, an interpersonal game and neuroimaging in Nature Communications in 2024, and found that receiving a favor produces two separable feelings depending on what the recipient infers about the giver's intent. Perceived care recruits one pattern. Inferred strategy, the sense that a future return is expected, recruits another, and it comes with obligation rather than warmth. Both motivate reciprocity, which is exactly the trap. A donor can give again out of felt closeness or out of felt debt, and the numbers look identical on the report while the relationship underneath them does not.

Which reframes what a thank-you is for. In Neurogiving I treat the acknowledgment as the moment the helper's high either gets confirmed or gets quietly overwritten. The donor arrives at that moment already holding a good feeling they produced themselves. A well-built thank-you says, in effect, you are one of the people this depends on. A receipt-shaped one says the transaction is closed and logged, which is administratively true and relationally expensive, because closing a transaction is the opposite of continuing a relationship.

Here’s a thought experiment before your next acknowledgment review.

Take the letter you send a donor of fifteen years and imagine it arriving from your closest colleague, someone who has worked beside you for a decade, after you covered a shift for them. Read it in that voice. If it sounds strange in that mouth, formal, effusive, oddly grateful for something you both consider ordinary, that strangeness is information. Warmth calibrated for a stranger reads as distance when it lands on an insider.

Then do the small version of the audit. Pull the acknowledgment your twenty longest-tenured donors received for their last gift and mark each one for what it actually asserts. Some assert a fact: your gift was received, here is the amount, here is the tax language. Some assert a claim about the donor: this depends on people like you and you are one of them. Most organizations find they are sending the first kind with adjectives added, and the adjectives are doing none of the work.

Research Sources

In Lab Notes Pro

The Pro edition takes on the finding that explains the nulls: a 2024 paper in the Journal of Personality and Social Psychology showing that expressing gratitude signals a relationship where expectations were exceeded, which makes it less welcome the closer the relationship gets. Alongside it, a January 2026 study where prompting people to thank an old friend measurably reduced their odds of reconnecting, the 2025 registered report finding that the content of a gratitude expression barely matters while its presence does, and the neuroimaging work separating care from obligation. Plus the context breakdown across five channels, three named failure modes, the donor's view from a kitchen table, and an experiment that tests which relational register your acknowledgments are actually written in.

The finding that makes sense of the field nulls comes from Jiaqi Yu and Shereen Chaudhry, published in Journal of Personality and Social Psychology in 2024. Across seven studies combining observational data, real behavior in an experiment and vignettes, they show that expressing gratitude carries a message beyond appreciation: it conveys that relationship expectations have been exceeded. That message is welcome from a stranger and awkward from an intimate, because thanking someone for what you both understood to be their part signals you did not consider it their part. Their test case was culture. People in China, where more is expected of close others as a matter of course, were less likely than Americans to express gratitude to close others, with no cultural difference for distant others, and the gap traced back to differing expectations of duty. These are interpersonal relationships rather than donor files, and the comparison is two countries rather than a general claim about collectivism, so what it licenses is a hypothesis about your own program. The transferable piece is the mechanism: gratitude is a relational signal, and the signal says a duty was not owed.

The behavioral cost of getting that signal wrong shows up in work published this January. Jinal Patel, Kristina Castaneto and Lara Aknin, in Motivation and Emotion, 2026, first coded 194 notes people wrote to friends they had lost touch with and found exactly one containing an unambiguous statement of gratitude. Then they randomized 424 participants to draft a note with or without a suggestion to include an expression of gratitude. The suggestion undermined reconnection efforts. Read against the Yu and Chaudhry mechanism it hangs together: adding thanks to a lapsed relationship marks the distance in it rather than closing it. Anyone whose lapsed-donor reactivation series opens by thanking the donor for their past support should sit with that for a minute, while remembering this is one experiment, on friendship notes, measuring reconnection rather than giving.

Then the question of whether wording can rescue any of this. Anurada Amarasekera and colleagues ran a registered report in the Journal of Experimental Social Psychology in 2025, testing whether gratitude works better when it praises the helper's character or their specific act (Amarasekera, Travis, Castaneto, Cash & Aknin, 2025, verified via the authors' lab record). Experiment 1 found no differences in helping across conditions. Experiment 2 found that people who received any expression of gratitude donated more than people who received just an acknowledgment of their help, with no advantage for either content type. The distinction copywriters argue about for hours did nothing. The distinction between an acknowledgment and an expression did something. Presence over polish, in a preregistered design that could have found the opposite.

The neuroimaging closes the loop. Gao and colleagues in Nature Communications, 2024 validated a model of indebtedness across three studies, and the useful part is the separation. Perceived altruistic intention, which the authors describe as conveying care and communal concern, is associated with one set of activity patterns; inferred strategic intention, which conveys expectations of future reciprocity, is associated with a different set and produces obligation rather than warmth. Both predict reciprocity. This is a laboratory game with modest samples, so hold the mapping to donor mail loosely. But it names something practitioners feel and cannot usually articulate: two acknowledgments can produce the same second gift while installing entirely different reasons for it, and only one of those reasons survives a bad year.

The honest synthesis is narrower than either the sector's confidence or a cynic's dismissal. Gratitude expression is real and its mechanism is social worth. What has repeatedly failed to move giving is the industrialized gratitude ritual, the call and the handwritten card and the advance thanks. And there is now a plausible reason for that failure beyond bad execution: a generic thank-you sends a relational signal about expectation and duty, and the donors receiving the most of them are precisely the ones for whom that signal is wrong.

The principle is constant: reduce caution triggers, and substantiate rather than embellish. How it lands depends on the room you are working in.

The Receipt Reflex. The organization treats the acknowledgment as a confirmation of a completed transaction, then adds warmth on top: gift amount, date, tax language, and a sentence of appreciation appended like a signature. The result satisfies compliance and delivers no regard, which is the one ingredient the mechanism actually needs. The corrective is to write the assertion first and let the receipt information sit underneath it, so the document's primary claim is about the donor's place in the work rather than about the money's arrival in the account.

The Debt Ledger. Stewardship language slides, usually without anyone deciding to, from care into expectation: your continued support is what makes this possible, we are counting on you, we hope we can rely on you again. The indebtedness research says this is not a neutral rhetorical choice, because inferred strategic intent produces obligation rather than warmth even when it produces the same next gift. The corrective is a language audit for future-tense claims on the donor. Gratitude that points backward at what happened is care; gratitude that points forward at what is expected is an invoice with a nicer font.

The Stranger's Thank You. Every donor receives the register designed for a first gift, so the people most fused with the cause get told, several times a year, that their participation was above and beyond. The Yu and Chaudhry mechanism makes this a distancing signal rather than a warm one, and the Patel result suggests the cost is behavioral rather than merely aesthetic. The corrective is to segment acknowledgment register by tenure and depth, which is an unglamorous data project and probably the highest-leverage stewardship change available to most shops this year.

He gave in March, the way he has given every March since his daughter's first season, and the letter arrives eleven days later. It is a good letter. Someone clearly worked on it. There is a line about how his generosity makes the work possible, and a line thanking him for choosing to support them, and his name is spelled right, and the amount is correct.

He reads it standing at the kitchen counter and feels something small he would not be able to name if asked. Not offense. Something closer to the feeling of being introduced to a person you have known for years. Choosing to support them. He did not experience it as a choice. It was March. He has never once considered not doing it, and the letter seems to think he considered it and decided generously, which means whoever wrote it does not know the shape of his life with them in it.

He puts it in the pile. Next March he will give again, because the giving was never about the letter. But something did not get built that could have been built, and nobody at the organization will ever know it, because the gift renewed and the report will say the stewardship worked.

Additional Research Links

The Relational Register Hypothesis. For long-tenured donors, an acknowledgment written in a partnership register will outperform one written in a gratitude register, and the effect will reverse or disappear among first-time donors.

Run this on your automated or batch acknowledgment for a period with real volume, ideally a spring appeal or a year-end wave large enough to split. Randomize at the donor level within tenure strata, holding channel, timing, sender and gift-receipt content identical. The only thing that varies is the assertion at the top.

Arm A is your current thank-you, whatever register it is in. Arm B is a partnership register: no thanks in the opening, no reference to generosity or to choosing to give, and instead a specific statement of what happened and where the donor sits in it, closing with what comes next. Arm C, if volume allows, is your current letter with the exceeded-expectation language intensified, because the prediction is directional and you want to see whether pushing it further makes things worse among your most committed donors. Stratify into first gift, two to four years, and five years or more before randomizing, since the interaction is the whole point.

Primary outcome is next-gift rate at twelve months, because that is where the field nulls were measured and you want a comparable bar. Secondary outcomes worth capturing: replies to the acknowledgment, which are cheap to count and are the closest available proxy for felt regard; opt-outs and complaints; and, if you can survey a subsample thirty days out, a single item on how well the donor feels the organization understands their relationship to the cause.

If B wins among five-year-plus donors and A wins among first-gift donors, you have replicated the register interaction in your own file and the implication is a tenure-segmented acknowledgment program rather than a better letter. If nothing separates, that is a genuinely useful null and it belongs in the same category as the two field studies above: proof that acknowledgment copy is not the lever, and permission to move the resource somewhere that is. If C wins anywhere, something in your donor relationships works differently than this literature predicts, and that is the most interesting result available to you.

Gratitude is the one thing in fundraising nobody is allowed to test, and I think that is because we do not really believe it is a tactic. We believe it is a virtue. Suggesting the handwritten note might not work feels like suggesting we be rude, so the conversation stops before it starts, and a practice with two well-run nulls against it keeps consuming Saturday mornings.

I have written those notes. Hundreds of them, some of them at a kitchen table late at night with a glass of wine and a stack of donor lists, and I would defend the practice for reasons that have nothing to do with retention. It changed me. The gratitude research on the person doing the feeling is actually the strongest part of the literature, and a fundraiser who spends an hour thinking specifically about forty people is a better fundraiser on Monday. That is a real benefit. It is just not the benefit we claim in the board report, and the difference between those two things is where our credibility leaks out.

What frustrates me is the substitution. We have decided gratitude is the answer to a retention problem that is really a relevance problem, and gratitude is easier, so we do more of it, faster, with better paper. Meanwhile the donor at the kitchen counter is reading a letter that thinks she chose to be generous when she experienced no choice at all. Nobody is being manipulated in that exchange. Something subtler is happening, which is that we are describing her relationship to us incorrectly and she can hear it.

So the question I do not have a settled answer to, and would like yours on: if you accept that a thank-you is a claim about the relationship rather than a courtesy, then most organizations are making the same claim to a first-time $25 donor and a twenty-year trustee. Which of those two is your current letter actually written for, and what would you have to know about your donors to write the other one?