n.V1-5.06| SELECTED, NOT TOLERATED

01| If funding followed effectiveness, weak organisations would shrink and strong ones would grow. That is what a market does. Instead, organisations of very different quality persist side by side for decades, and nobody in the system appears to be doing anything wrong.

02| There are three places the mechanism could break. Donors might not act on performance. Organisations might not signal it. Or the information might not exist. All three have been tested, and the results are more interesting than the assumption that people are careless.

03| Start with the donors. The most cited measurement is Hope Consulting's Money for Good study, which surveyed 4,000 donors in 2010 and produced a ratio quoted ever since: nine in ten donors say they care about performance, three in ten research the organisations they give to, and three in a hundred take any active step to direct their money to the highest-performing option.

04| The follow-up asked why, and the answer is not indifference. Donors give to organisations they already know, year after year, and when they consider a new one it is usually one that is already well known and, in their view, already vetted by somebody.

05| That deserves respect rather than a lecture on civic virtue. Verification is expensive, and — as the rest of this book part establishes — almost nothing is published in comparable form. Given that, renewing with an organisation you have a long relationship with is a sound heuristic. It economises on a cost the sector has never brought down.

06| The consequence for selection is still unavoidable. If donors mostly renew, then the dominant factor in which organisations receive money is which organisations already received money.

07| The obvious fix is to supply the missing information. That has been tested in the field, at scale. Dean Karlan and Daniel Wood ran a direct-mail experiment with Freedom from Hunger. The letters were identical except that some included a discussion of the programme's impact as measured by scientific research; both arms carried the usual qualitative story about an individual beneficiary.

08| Adding the scientific evidence had no effect on the average likelihood of giving, or on the average gift.

09| The average conceals the finding. Large prior donors, given the evidence, were more likely to give and gave more. Small prior donors, given the same evidence, were less likely to give. Karlan and Wood's reading is that the two groups are doing different things — large donors pursuing altruistic outcomes, small donors pursuing warm glow, the feeling of having helped.

10| Read that from the organisation's side. Putting your strongest evidence in your fundraising material gains you your major donors and costs you part of your small-donor base. Depending on the mix, leading with evidence is either rational or expensive. That is not an organisation failing to communicate its strengths. It is an organisation reading its audience correctly.

11| So what does the audience reward? The best-known answer is the identifiable victim — people give more to one named, pictured person than to statistics about many, and giving does not scale with the size of the problem. This literature needs handling with more care than it usually gets: a 2023 replication published in Collabra: Psychology found no empirical support for the identifiable victim effect. It is contested and should not be leaned on.

12| One result from the original work survives and is worse anyway. Small, Loewenstein and Slovic found that prompting people to notice the discrepancy — teaching them they respond more to one identified person than to many statistical ones — reduced giving to identifiable victims without increasing giving to statistical ones. Total giving fell.

13| If that holds, there is a bind underneath the entire give-donors-better-information agenda. The emotional register that produces giving and the analytical register that produces comparison may not sit comfortably in the same donor at the same moment.

14| There is exactly one piece of comparative information about organisations that donors have reliably used, and it is the overhead ratio. It travelled because it was cheap — extractable from a tax filing, publishable by a rating site, comparable across organisations without any understanding of what they do. And the sector's own reform literature and its main rating bodies now hold that it measures approximately nothing useful, and that pressure to keep it low produces organisations that are worse rather than better.

15| So the market did generate a comparative metric, and did respond to it, and the metric was wrong. That is a sharper diagnosis than donor inattention. When one cheap number is the only comparable thing on offer, it becomes the basis of selection whether or not it means anything.

16| Assemble the pressure and it is describable. The organisation that does well retains habitual donors, tells a compelling story about the problem, reports a low overhead ratio, and does not lead with evidence in mass-market appeals. Every one of those is a rational response to a documented feature of donor behaviour. None of them is impact.

17| And no correction arrives from the other end. A company that disappoints loses customers individually and continuously, and the signal is unmissable. An organisation here has no equivalent. Its beneficiaries cannot withdraw funding. Its donors renew by default. The nearest thing to an exit signal is a single funder deciding to stop, which is one party's judgement rather than an aggregation of many, and usually follows a strategy change rather than a performance assessment.

18| Mediocre organisations are therefore not tolerated by an inattentive market. They are selected by an attentive one optimising the wrong variables, in which every participant behaves sensibly given what it can see. Two honest limits: this does not show the survivors are bad — selection on the wrong variables produces a distribution containing plenty of effective organisations, it just fails to sort them. And it does not show that a market running on measured effectiveness would do better, because measured effectiveness is scarce, contested and skewed toward whatever is easy to count. Nobody has run one long enough to find out.