As households reduce discretionary spending in response to the COVID-19 pandemic, concerns are high that a resulting fall in aggregate demand can lead to a lasting recession post-COVID-19. Consequently, policies aimed at stimulating consumer spending are of key interest to governments working to mitigate the economic costs of the pandemic. While some governments have relied on direct household cash transfers, others have used consumption vouchers, which are cash discounts for purchases of goods. Compared to cash, which can be saved, consumption vouchers expire if not utilized within a deadline; are available only if utilized for targeted goods; and as such, have the potential to be more efficacious at stimulating spending. We evaluate consumption vouchers leveraging randomized controlled experiments implemented at JD.com, a large e-commerce platform based in China. The scale of the experimentation - comprising about 110 million users - allow us to causally evaluate a range of vouchers on their digital distribution, targetability, money value and contractual restrictions. We find that the design of the vouchers is key to their efficacy, and that best designed vouchers stimulate incremental spending of about 2$X$ the expenditure incurred on the voucher. These results can help drive public policy and also underscore the value of distribution, transaction and testing of vouchers via digital platforms such as e-commerce, making them private sector facilitators of public policy goals.