U-Report is a program intended to amplify the voices of young people in countries where UNICEF works. It was originally implemented using SMS to allow youth to register as “U-Reporters” and then receive and respond to questions about their lives. Responses are aggregated and shared back to U-Reporters, government and NGOs, and are published publicly. In emergencies, such as the 2014-2016 West Africa Ebola outbreak, U-Report was also used to deliver lifesaving information. In 2016, UNICEF started experimenting with allowing youth to interact with U-Report using internet messaging applications.
Responding to trends showing the increased use of messaging applications among youth around the world, U-Report decided to expand its channels. Telegram was chosen first in early 2016 because it was one of the first messaging applications to provide an open API that allowed organizations to develop and integrate chatbots with other platforms. Telegram, however, had low penetration in the countries where U-Report operates, so it was never used at scale. Later in 2016, U-Report tested the Facebook Messenger API for delivering U-Report surveys. By late 2017, the integration was still at an early stage, but the U-Report team viewed its partnership with Facebook as a critical investment that would result in cost savings to U-Report and U-Reporters because, unlike SMS, there is no per-message cost to use Facebook Messenger.
U-Report’s use of messaging applications relies on structured decision trees. U-Reporters send messages and are asked specific questions based on the content. U-Report sometimes supplements this with appointment-based activities, where incoming questions are answered live by UNICEF staff and answers are streamed live on Facebook. These sessions allow U-Reporters to receive a more in-depth human response that is then amplified for a broader audience. UNICEF is experimenting with machine learning and artificial intelligence for future U-Report chatbot conversations, but human involvement is still considered essential to provide critical learning that will later optimize the machine learning structure.
While U-Report has since developed about 20 different conversation streams using Facebook Messenger, their deployment remains in the early stage and their effectiveness relative to toll-free SMS is yet to be thoroughly assessed. As of late 2017, U-Report had only deployed a chatbot at scale in Myanmar, where 15,000 U-Reporters have been engaged through Facebook Messenger following difficulties integrating with the mobile network operators for SMS communications. U-Report representatives have resisted drawing early conclusions about the Facebook Messenger integration, noting that the U-Report platform’s architecture, originally built only for SMS, is still being adjusted to optimize the integration. One clear benefit is that Facebook Messenger is more cost effective than SMS for, has great UNICEF and, therefore scaling potential.