Why Shuttering AmazonSmile Was a Mistake

In 2013, Amazon launched a program called AmazonSmile as a way to make it easier for their customers to give back. Under this program, customers who chose to visit an alternative version of the online marketplace — AmazonSmile.com rather than Amazon.com — could designate a charity of their choice. Amazon would then donate a small percentage of their purchases to that charity. This initiative was what we term a “giving-by-proxy” program: Amazon was acting as a proxy for their consumers, making charitable donations in their behalf.

AmazonSmile contributed a tremendous amount to the collective good. The company reports that AmazonSmile directly led to half a billion dollars in charitable donations over the past 10 years. As more and more organizations are working toward boosting their societal impact, the success of AmazonSmile therefore highlights that giving-by-proxy programs may be a value path through which organizations can give back. But in January of this year, Amazon announced that they would be closing the AmazonSmile program. “The program has not grown to create the impact that we had originally hoped,” the company explained in an email to AmazonSmile customers. “With so many eligible organizations — more than 1 million globally — our ability to have an impact was often spread too thin.” This news was met with disappointment from many members of philanthropic community, who saw great value in the $449 million that Amazon donated, but according to a company spokesman, the average annual AmazonSmile donation was only about $230 per charity.

Our research suggests that Amazon’s concerns about the impact of AmazonSmile were misguided, in two ways. First, it suggests that the impact of the program was meaningfully larger than Amazon realized. And second, it suggests that the company could have retained the AmazonSmile program — and continued to capture its benefits — by making a simple modification that would have directed the program’s efforts at a smaller number of charities.

Indirect Benefits

Why might AmazonSmile’s impact have been bigger than Amazon realized? Because giving-by-proxy programs like AmazonSmile tend to have benefits that extend beyond their direct impact.

Giving-by-proxy programs are growing increasingly popular, in not only consumer contexts but also workplace contexts. For example, employers are beginning to shift away from incentivizing employees solely through traditional bonus structures and have started providing them with “prosocial bonuses” (that is, employee bonuses spent on others, often in the form of charitable donations).

In recent research, we explored whether these giving-by-proxy programs, beyond their direct impact, might inspire consumers and employees to engage in other prosocial behaviors down the line. For example, after participating in giving-by-proxy experiences, might consumers and employees be more likely to make charitable donations outside of the giving-by-proxy setting?

To find out, we conducted a series of experiments. We recruited participants online — in total, over 3,000 participants across four experiments — and randomly assigned each participant to a “giving-by-proxy” condition (in which the participant engaged in a giving-by-proxy experience) or a “control” condition (in which the participant did not have such an experience). Then we gave each participant a chance to behave charitably or selfishly, which allowed us to assess whether giving-by-proxy experiences inspired participants to behave more charitably.

In three of our experiments, we explored the impact of giving-by-proxy experiences designed to mirror workplace contexts. To this end, we recruited participants to complete puzzles for us as our temporary employees. In the giving-by-proxy condition, we rewarded them for completing the puzzles with a prosocial bonus, in which we donated money to charity in their behalf. In the control condition, by contrast, we did not provide a bonus for their work.

In a fourth experiment, we instead attempted to mirror giving-by-proxy programs in consumer contexts. We recruited participants to view a selection of products from Amazon and choose one product to purchase. In the giving-by-proxy condition, participants were linked to products on AmazonSmile (and read some information about AmazonSmile — this experiment was conducted before the program was terminated). In the control condition, participants were linked to the same products, but on the standard Amazon platform, where they did not read anything about AmazonSmile.

Critically, across each of these four experiments, giving-by-proxy experiences inspired subsequent charitable behavior. For example, in our AmazonSmile experiment, 39.5% of participants in our giving-by-proxy condition (the participants directed to AmazonSmile) chose to donate the additional bonus they received to charity at the end of the experiment, while just 29.2% of participants in the control condition (who were directed to Amazon.com) chose to donate.

These results highlight the potential power of giving-by-proxy initiatives to inspire downstream charitable giving, in both consumer and workplace settings, and therefore to have an indirect impact that adds to their social value. Our results also speak to the psychological processes through which giving-by-proxy experiences may inspire subsequent charitable behavior. Notably, after participating in giving-by-proxy experiences, participants in our experiments reported feeling like more charitable people and expected that others would see them as more charitable. And these feelings were positively correlated with donating to charity at the end of the experiment.

This work suggests that, beyond its direct impact, AmazonSmile may have indirectly inspired consumers to behave charitably in other ways. Ethics scholars have recently made a call for treating moral decision-making as a design problem, suggesting that policymakers and organizational leaders should work to create contexts that encourage ethical action. We believe that giving-by-proxy programs may be one important strategy that organizations can use — by designing prosocial bonuses and giving-by-proxy marketing programs — to help create contexts that will inspire their employees and consumers to give back.

Less is More

Our research also explored the question of how to optimally implement giving-by-proxy programs. In particular, we were interested in whether incorporating autonomy into such programs is important: When making charitable donations in behalf of consumers and workers, is it important to let these individuals choose the charity that will receive the donation?

To find out, in our experiments we compared the downstream effects of two distinct kinds of giving-by-proxy experiences. Some participants got to choose the charity that we donated to in their behalf (giving-by-proxy with autonomy), while others had the charity selected for them (giving-by-proxy without autonomy).

Interestingly, we consistently found that giving-by-proxy programs did just as much to inspire downstream charitable giving, even when participants had no choice over where their charitable donations went. Furthermore, participants reported feeling just as charitable (and expected others to see them as just as charitable), regardless of whether their giving-by-proxy experience involved autonomy.

In this way, our research suggests that Amazon could have implemented a simple modification to the AmazonSmile program, to address their concerns about spreading its impact too thin. In particular, the company could have modified AmazonSmile to give consumers less autonomy over the charity their purchase would support.

As noted above, Amazon decided to shut down the program because it felt its charitable donations were spread too thin. This could have been avoided — if, for example, Amazon had selected just one high-impact charity for AmazonSmile customers to donate to each year. Recent calculations reveal that the impact of donating to the most effective charities (such as those recommended by GiveWell) is approximately 100 times as large as the impact of supporting the average charity. Designating one highly effective charity for AmazonSmile to support could therefore have solved the “spread too thin” problem and — according to our research — preserved the indirect impact of the program, by continuing to inspire consumers to make downstream charitable donations.

The Way Forward

Our research suggests that companies hoping to improve their societal impact should consider giving-by-proxy programs, despite Amazon’s decision to cancel AmazonSmile, because of the many potential direct and indirect benefits that they can provide. This holds true even when consumers and workers are opted in by default and have little or no choice over the nature of these program. Our research also suggests that in developing giving-by-proxy programs, companies should make sure to do two important things: 1) They should incorporate the programs into their incentive structures, product offerings, and marketing programs, and 2) in designing the programs, they themselves should pick a single high-impact target charity, or at most just a few.

Giving-by-proxy programs inspire consumers and workers to give back in meaningful ways. We hope that our research will encourage more organizations to use the programs to contribute to impactful and valued causes — and to spur their customers and employees to do the same.