Donor Agency Without Allocation Drift: A Family Philanthropy Communication Protocol

A governance use for a fundraising experiment

A family foundation or family office often meets a charitable opportunity through a persuasive appeal. The appeal can emphasize the contribution of one person or the collective effort of a community. Neither framing establishes that the organization is effective, that the proposed grant fits the family mandate, or that a larger gift is prudent. The useful question raised by this study is narrower: can language change giving even when the underlying opportunity stays the same? If so, the family needs a stable allocation process, and the charity needs an honest method of testing communications.

This companion proposes two connected controls: a donor-side decision record that separates message appeal from grant assessment, and a charity-side pilot that evaluates communication without sacrificing consent, accuracy or beneficiary dignity. The proposed controls are editorial recommendations, not interventions tested by the authors. They are relevant to family philanthropy, but the original research did not study family offices, foundations as institutional allocators, or grants made after extensive diligence.

Read the denominator before repeating the headline

Whillans and Dunn report a randomized field experiment during a December 2015 annual campaign at an elite United States business school. The 12,316 alumni on the mailing list received two emails and a letter using either agentic or communal language. The original paper reports average giving of $18.47 versus $10.13 across the assigned groups, including people who gave nothing. The $8.34 difference is the basis for the approximately 82 percent relative increase. It is not an estimate that every wealthy donor will increase a gift by 82 percent, and it is not a forecast for a family-office campaign.

The study did not find a statistically significant difference in whether recipients donated. Among donors, the reported average amounts were higher in the agentic condition, but this conditional comparison should not replace the full randomized-group comparison: becoming a donor is itself an outcome after assignment. The distinction matters operationally. A communication may change the amount given by existing supporters without solving the problem of reaching new supporters. A dashboard that reports only revenue can hide that difference.

Affluence was characterized partly through neighborhood median household income and the business-school setting, not through verified personal net worth. The interaction with neighborhood income was not significant. An exploratory analysis involving past donations should generate a future hypothesis, not a universal threshold for selecting donors. The paper also cannot identify the mechanism with certainty: personal agency, donor identity or a response to the communal comparison could contribute. Longer-term retention and generalization to other charitable causes remain open questions.

Donor-side control: separate attraction from allocation

The family office philanthropy lead should prepare a short grant assessment before presenting campaign imagery or personalized donor language to the decision committee. The record should state the proposed activity, beneficiary need, evidence of effectiveness, implementation capacity, funding gap, restrictions, reporting expectations and material uncertainties. Ask whether the same project would qualify if presented in a less personally affirming way. This is a deliberation prompt, not a claim that donors can remove every framing effect from their decisions.

The family council should identify a budget and a decision authority in advance. A compelling message can justify asking more questions; it should not silently expand the budget or waive conflict checks. If a donor wants recognition or a particular role, record that preference separately from the assessment of beneficiary outcomes. Where recognition imposes obligations on the charity, obtain clarity about those obligations before approval. Pause the grant if the organization cannot explain how the requested funds will be used or if publicity commitments conflict with beneficiary interests.

Charity-side control: test truthful language within a fixed offer

The fundraising director should first confirm that both proposed communications describe the same eligible project, funding destination and donor options. An agentic version can explain what a contribution enables without inventing exclusivity, certainty of impact or urgency. A communal version should not invent matching donors or social approval. Keep differences limited enough that any measured outcome can plausibly be attributed to the planned communication change rather than to different donation buttons, deadlines or benefits.

Before distributing a pilot, obtain the organization's appropriate ethical, privacy and operational review. The original study received ethics approval and a consent waiver; that does not automatically authorize another organization to copy its research arrangements. Respect contact preferences and applicable requirements, and do not collect sensitive wealth data simply to imitate the paper. A single donor should receive a consistent assigned version across planned channels, with a documented process for accidental crossover. The proposal is a communication evaluation, not permission to deceive recipients.

Measurement: count everyone assigned and disclose uncertainty

The evaluation lead should predefine the unit of assignment, eligible audience, campaign window, primary outcome and analysis before the campaign begins. Revenue per assigned recipient, including zero gifts, is a sensible primary measure for this type of question. Report donation participation, amount among donors, refunds, unsubscribes and complaints separately. Costs and staff time should also be recorded; a message may be cheap to change while the campaign itself remains costly. Avoid declaring victory on a large percentage increase when the absolute difference is small or uncertain.

If the organization lacks enough recipients or analytical capacity for a useful randomized evaluation, it should use a modest learning exercise and label the findings descriptive. A before-and-after comparison across different campaigns cannot isolate the effect of language because timing, audience and economic conditions may change. For a randomized pilot, preserve the allocation record and analyze by assignment. Report uncertainty and sensitivity to unusually large gifts rather than quietly removing inconvenient observations. Any exclusion rule must be justified and recorded before seeing the result wherever feasible.

Stopping conditions and durable learning

The responsible executive should set stopping rules before launch for inaccurate claims, privacy failures, unusual complaint patterns, technical errors in assignment or material distress caused by the appeal. These are operational safeguards, not evidence that a particular statistical threshold is mandated by the source. Do not repeatedly inspect results and stop solely when a favorable revenue difference appears. If an interim decision is necessary, state how the monitoring rule changes the strength of the conclusion.

After the campaign, the board or authorized committee should review the complete record: what changed, who received it, what the measured difference was, what uncertainty remains, and what happened to recipient trust. A positive short-term result can justify a further bounded test, not permanent adoption across every cause and audience. Follow retention and complaints over time, and withdraw language that proves misleading even if it raises more money. A fundraising gain is not the same as a beneficiary gain; the grant program still needs its own outcome evidence.

The family office and charity should retain separate decisions. The charity may conclude that a communication is effective, while the family may still conclude that the project is outside its mandate. That separation protects both donor agency and institutional discipline. This is an AI-assisted editorial analysis with no new participant data, no source-author endorsement and no documented independent human acceptance. The study provides a reason to test and scrutinize framing, not a promise of returns or a general psychological profile of wealthy people.

Source, attribution and limits

[1] Ashley V. Whillans; Elizabeth W. Dunn. Agentic appeals increase charitable giving in an affluent sample of donors. PLOS ONE (2018-12-06). https://doi.org/10.1371/journal.pone.0208392. CC BY 4.0.

One US business-school annual campaign; neighborhood income is an imprecise proxy for individual wealth; no significant participation effect; mechanism unresolved; past-donation moderation exploratory; persistence and other causes not established. No family-office or beneficiary-outcome study.

Prepared by the Family Office Private Wealth Journal editorial research desk with AI assistance at the direction of Dr. Freedom Cheteni. Original research authors retain full credit and are not authors or endorsers of this companion. This proposed process has not been evaluated in a new empirical study; no independent peer-review acceptance is claimed. This is educational analysis, not individualized investment, legal or tax advice.

Implementation record

Open the cited sources

1. [1] Ashley V. Whillans; Elizabeth W. Dunn. Agentic appeals increase charitable giving in an affluent sample of donors. PLOS ONE (2018-12-06). ↗