The study is Cues, Attention, and Charitable Giving, by Luca Henkel, Christoph Oslislo and Frederik Schwerter. It is a preregistered field experiment, run inside a live campaign for a charity that helps children affected by war. The extra paragraph cut the chance of someone signing another pledge by 33 percent. The authors work that out to about 30,000 euros a year. It is a working paper from May 2026 and has not been peer reviewed, which you should know before you forward it to your board.

The second half of the study is the part that matters. They ran the same paragraph past people online and asked them who the charity helps. People who had read it mentioned local recipients much more often than people who hadn’t: 67 percent of them, up to 90 percent. Mentions of the children overseas dropped from 24 percent to 6 percent.

Read that again slowly. The donors were not put off. They were not confused. They read a line about their own neighborhood, pictured a food bank down the road, and the children the charity actually serves went quietly out of mind.

The ask string does the same thing. Indranil Goswami and Oleg Urminsky ran eight studies with 11,508 people, forthcoming in the Journal of Marketing Research. Put any suggested amount on the page and donors became less responsive to everything else on it, including good information about the charity. Without a suggested amount, people gave more to the charities they thought well of. With one, that mattered less.

If you’ve heard my Neurogiving keynote, you probably remember the section on working memory (and the Monty Python clip!) . We write as if the donor reads the whole letter and then decides. She doesn’t. She picks up two or three things, builds a rough picture of who gets helped and what the money does, and decides on that picture.

This is the availability heuristic working inside one letter, email, or conversation. Whatever came to mind easiest stands in for the whole thing. A line does not have to be convincing to cost you a gift. It only has to be one of the two or three things she picked up.

Which is why one specific person in an appeal is so easy to lose. That only works if she holds that person long enough to care about them. Mirror neurons cannot echo someone she never pictured. A second story, a line about the gala, a paragraph on the strategic plan, a highlighted amount: every one of those is competing for the two or three slots, and every one that wins is a person she never met.

Stop saying yes to additions without saying what comes out. Not a ban on editing. One rule in the review cycle: nothing goes in until somebody names what leaves. The sentence your colleague wants to add might be good. That was never the question. The question is what it pushes out, and right now nobody is asking it.

You will get to use this within a week, because somebody is going to ask you to add a line to something.

Here is the whole thing. “Sure. What comes out?” Say it warmly, the first time and the fourth time. It is not a no. It moves the conversation off whether the sentence is good, which it is, and onto what it costs, which nobody has been counting.

You will get one pushback and it is always the same: there’s room. So have the next line ready. “There’s room on the page. There isn’t room in her head.” Then show them the 33 percent.

Sometimes the honest answer is that it can’t come out. The board chair’s note is going in and that’s that. Fine. It’s still a trade, so make it a conscious one. If the note stays, something else goes, and you pick which instead of letting the layout decide for you.

Want to practice first? Take the thing in your current appeal you would fight hardest to keep, and make the case for cutting it. Not because you will. Because if you can’t argue that side, you won’t hold the rule when someone senior pushes.

The Additive Fallacy. You treat the letter as a container, so a good sentence, once added, is assumed to sit there harmlessly next to the others. The letter ends up carrying the story, the impact numbers, the identity language, the match, the anniversary and the board chair’s note. Each one was defended by somebody who was right that it was good. The Henkel result is the cleanest answer to that: one extra paragraph, nothing wrong with it, a third fewer additional pledges. The fix is the rule above, applied before layout instead of after. Every element names the element it will displace. If nobody can say what comes out, nothing goes in.

In Lab Notes Pro

One thing this edition leaves open on purpose. If attention is what runs out, the obvious move is to cut the appeal down to one emotional beat and let the donor’s gut do the rest. I believed that for years. There is a meta-analysis that says the second half of that sentence is wrong, and once you see why, it changes what you cut and what you leave alone.

You could finish the free edition thinking donors are simple, so appeals should be simpler. That is not quite what the evidence says, and the difference determines what you cut.

Where the eye goes first, and it isn’t your logo. Olga Grabowska-Chenczke and colleagues tracked 96 people looking at four food bank posters, in Behavioral Sciences, 2026. Pictures pulled the eye three to seven times faster than logos, and held it for two to seven times as many fixations. Those are the abstract’s numbers. The body is messier: the range covers fixation counts rather than every measure, one poster’s difference in time-to-first-look wasn’t significant, and another went past the top of the range. The direction holds. The precise figures don’t. It is also a small exploratory study of posters in Poland, and nobody donated anything, so treat it as evidence about eyes and not about giving. What you can use: the picture gets looked at first, so the logo you fought for in the header was never going to win. What you cannot use: that sad pictures raise donations.

Fewer choices is not the fix. Amanda Lindkvist and Timothy Luke ran a preregistered experiment with 501 people in Frontiers in Psychology, 2022. They varied how many charities someone could give to, across sixteen set sizes from 2 up to 80. Donation likelihood barely moved. It didn’t move for people who were unsure what they cared about either, and those are the people choice overload says should freeze.

Be careful what you take from that, because it is easy to over-read. Choice overload is a real effect and it shows up in plenty of settings. What this study suggests is that the number of charities in front of someone may not be one of them. That matters, because we reach for the jam-shelf story whenever a donor hesitates, and here it is the wrong story. The donors in the field experiment were not stuck between too many options. One sentence changed who they were thinking about. Different problem, different fix.

And the move everyone makes next is wrong. If attention runs out, cut the appeal down and let the gut decide. Hanna Fromell, Daniele Nosenzo and Trudy Owens pooled 22 experiments with more than 12,000 people in Experimental Economics, 2020. All of them rushed people, loaded them up or tired them out to push them toward instinct and away from thinking. The overall effect was close to zero. They then ran a new experiment built to catch effects hiding under that zero, and found little. Adults do not have a more generous mode you can reach by hurrying them.

Why that is true of adults and not of children. Francesco Margoni and colleagues gave 537 Italian children aged three to ten the same kind of task, under time pressure or with a forced delay, in Nature Human Behaviour, 2026. The young ones were more generous when they answered fast. That gap closed as they got older. By the time your donor is an adult, generosity isn’t something that slips out when you distract her. It’s something she thinks her way into. So your appeal’s job is not to get past her thinking. It’s to give her one thing worth thinking about.

What this adds up to. Supported: attention is what runs out, what she pictures is set by the handful of things she picks up, and adults don’t give more when you rush them. Over-claimed by us: that shorter is always better, that donors stall because we gave them too much to choose from, and that a feelings-first appeal reaches a more generous version of them. Still unknown: which element on your page is the one costing you. No study here can tell you that about your file, which is what the experiment below is for.

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

The Stray Cue. Somebody adds a detail to make the letter warmer or more relevant, and it quietly changes who the donor pictures. Her city, her ZIP code, a local event, a photo of the staff team, a line about the new campus. None of it is wrong, and all of it competes. This is the one the field experiment caught, and it cost real money. The fix is to check every specific detail against one question: does this make her think about the people the gift is for? If it points somewhere else, it isn’t personalization. It’s a competitor, and it costs more than it adds.

The Logo Tax. Brand guidelines put the identity first. Wordmark in the header, tagline, color block, mission statement, all before anyone sees a human face. The eye-tracking work says she looks at the picture first whatever you do, so when the picture is a logo, the first thing she takes in is your institution instead of a person. The fix is to give the beneficiary image the spot the brand currently holds, and move brand identity to where it actually helps, which is reassurance when she’s about to pay, not grabbing attention while she’s deciding.

Tuesday, 6:40 in the evening, phone in hand, pasta water not on yet. She didn’t open it to be persuaded. She opened it because she likes the sender.

First a banner, the name in the colour she knows. Then a photo of people at what looks like a gala. Then a paragraph that starts with the year the organisation was founded. Something has already happened here. The person she was half ready to picture hasn’t arrived, and in her place is a room full of donors in nice clothes.

She scrolls. There’s a story about a girl and it’s a good story, but it starts on the second screen. By then there’s also a line about a match, a quote from the executive director, and the amounts: 50, 100, 250, with the 100 already highlighted. Her eye goes to the 100. The girl is still up there somewhere.

She doesn’t decide no. She decides later, which is a decision made out of everything the email asked her to hold at once. Tomorrow, if she remembers anything, it’ll be the gala photo. Nobody chose that for her. It just got there first.

Additional Research Links

The Subtraction Hypothesis. Taking the non-beneficiary material out of an appeal will raise response rate without lowering average gift, and the lift will be bigger among donors who have given fewer than three times.

The 15-minute version. Pull up your current appeal the way it actually arrives, printed or on a phone. Hand it to a colleague who hasn’t seen it. Eight seconds, then take it back and ask two questions, writing the answers down word for word: who did you picture, and what would the money do? Three colleagues if you can get them. Compare what they said with what you thought the letter was about. That gap is usually specific enough to fix. If they pictured your organisation instead of a person, look at the first hundred words and the first image. If they said the amount before they said the person, look at your ask string.

The full design. Take your year-end control, the package that currently wins. Don’t write a new one. Build the variants by taking things out. Variant B removes everything that doesn’t point at the beneficiary: the anniversary line, the leadership quote, the second story, the mention of her city, the event photo, the brand banner if your platform will let a beneficiary image take its place. Leave the ask string alone. Variant C keeps all the content but removes the pre-selected amount, so the array is there with nothing highlighted. If volume is tight, run B only.

What to measure. Randomise person by person, and split first and second-time donors from donors with three or more gifts, because the prediction is that this matters most when she has the least history to fill in the gaps. Hold send time, sender, subject line and opening sentence the same. Main outcome: response rate. Also watch average gift, and send a two-question survey to a random slice of each group two days later asking who they pictured and what the money would do. Code the answers for whether they named the beneficiary. That survey is how you find out whether the explanation is right.

What would prove me wrong. If B loses on response rate, the material you removed was doing work, and the idea is dead. If B wins on response but average gift drops far enough to lose on revenue, cutting is a trade rather than a free win, and you need the exchange rate before you rebuild anything. If B wins and the survey shows no change in who donors pictured, something else in that material was hurting you and attention isn’t the explanation, which is worth more than a win you don’t understand. Set the threshold before you launch, because everyone whose sentence came out will want it back in January.

The most reliable way to ruin an appeal is to circulate it.

I can predict the whole sequence now. The writer produces something with one person in it and one thing the money does. Then it goes around. Programs wants their initiative mentioned. The ED wants a line. Marketing needs the anniversary lockup. Finance asks if the match can be bigger. Every request is reasonable. Every one gets granted, because saying no to a colleague costs more than saying yes. What comes out the other end is longer, better looking, more complete and worse, and nobody in that room did anything except improve it.

I think we have the donor backwards. We write as if she reads the letter and then decides. She picks up a few things and decides on those, and she doesn’t ask us which ones. That should bother anyone who has ever added a sentence for a colleague. It also makes the job simpler than it looks, because the letter has one question to answer: what is the one thing she’s still holding eight seconds after she looks away? Everything else is either helping that or competing with it.

Here’s where I want to be careful. You can read attention research as a manipulation manual if you want to, and a distressing picture that grabs the eye faster is an invitation to exactly the appeals I’ve argued against in this newsletter. The point isn’t to win the grab by any means available. It’s to stop filling it with things that serve us instead of the person she’s trying to help. Cutting is the honest version of this work, and it’s harder than adding, because it means telling a colleague their good sentence is coming out.

So here’s what I’d ask, and I’m asking it about something I’m revising this week. Take your best appeal and count how many people could point at a sentence and say, that one’s mine. Then ask how many of those sentences she’d keep if she got a vote. If those two numbers are far apart, whose letter is it?