When a purchase triggered a contribution to a cocoa farmer, participants in a controlled experiment tended to reduce the amount they gave directly, researchers found.
The study raises a question for businesses and nonprofits that link sales to a cause: how much of a campaign’s reported donation is additional, and how much replaces a contribution someone might otherwise have made?
The result comes from a stylized experiment with 89 people, not an analysis of shoppers in a real store. It does not establish that commercial charity campaigns generally fail to increase giving.
What purchase-linked giving means
In cause-related marketing, a business may promise to contribute an amount to a nonprofit or another cause when customers buy a product. For example, a seller could pledge a fixed sum for each item sold. The pledge, recipient and calculation should be clear enough for a buyer to understand.
A campaign can generate money for the recipient and help a company distinguish its products. But its total impact depends partly on what customers would have done without it.
If a customer who would have donated separately gives less because a purchase already triggers a contribution, economists call the reduction in direct giving crowding out. The size of that effect can vary.
How the researchers tested the trade-off
Nishita Sinha, Marco A. Palma and Richard T. Woodward report the research in the Journal of Economic Behavior & Organization, published in May 2026.
The 89 community participants in Texas made nine allocation decisions using funds provided for the experiment. They could buy abstract tokens for themselves or contribute directly to an individual small-scale cocoa farmer in Ecuador. A foundation also contributed under rules that varied between tasks.
In some tasks, the foundation’s donation rose with each token purchased, representing a purchase-linked contribution. In others, its contribution was determined separately. One decision was randomly selected to determine actual earnings and donations.
Tokens were used instead of a real consumer product so that preferences for a particular brand or product would not dominate the comparison. That design helps isolate the decision but makes the setting less like a normal shopping trip.
What the findings do and do not show
In the researchers’ model, higher purchase-linked contributions were accompanied by lower direct contributions. For one comparison, the researchers could not rule out a full offset of the additional donation. Failing to rule out a full offset does not prove that the two amounts were exactly equal.
The paper also examines the satisfaction people may receive from giving, sometimes called the warm glow of donation. The authors interpret the pattern as consistent with participants valuing both direct and purchase-linked contributions. Their model is an explanation of choices within the experiment, not a direct reading of each person’s motivation.
The participants came from one area, the recipient was a specific farmer and the decision involved tokens rather than familiar products. Different causes, campaign terms, customer groups and competing opportunities to give could produce different results.
How to assess a campaign
A company can state the amount or share donated, identify the recipient and explain any minimum, maximum or time limit. A nonprofit can track what it actually receives and consider whether the campaign reaches new supporters or changes other giving.
Where practical, comparing a campaign with a similar group that did not see it can help distinguish extra contributions from money that would have arrived anyway. Businesses should also count campaign and administrative costs before presenting results as net benefits.
Related MBN reporting on how advertising tests can compare different audiences as well as messages explains another reason to examine the comparison carefully.
A purchase-linked donation can still provide valuable support. The experiment shows why the amount printed on a campaign report and the additional benefit to a cause are not necessarily the same calculation.