Social Projects: Prototype New Ideas Early

Prototyping is integral to our work at Social Projects. We believe that the best way to understand and explore new ideas, is to test them as quickly as possible. It doesn’t have to be an expensive or hi-fidelity prototype, built from scratch. It can be a simple paper prototype, or hacked together with existing technologies, placed in front of a user as a conversation piece.

We have seen that time and resources are saved, when an idea or concept is taken out of the design studio, and tested, rather than focusing on polishing concepts in a closed environment. At Social Projects, we are taking this approach from the technology and design sector, and applying it with partners and collaborators in the humanitarian domain.

By identifying key service touchpoints — when a user and a service interact with each other — clever and simple prototypes can capture how the idea would work, and show you how it can be improved upon. The users will always know better than us.

“First digital touchpoint” — prototype
Before the concept of Refugee Text — an automated information service — came about, significant research and prototyping was conducted around the following question:

“How might we create a positive and dignified experience for refugees and asylum seekers in Denmark?”

The design challenge back then, focused on how to improve the initial meeting between refugees and the Danish state at the border. Based on observations and findings, most refugees wanted the first person they met, to be a friendly local, rather than a border agent.So, the first prototypes we made, specifically explored how to connect citizens with refugees.

“Talk to a Loal” — poster

With posters and flyers at key hotspots in Germany and Denmark, we tested “Talk to a Local”. This service provided contact information to personal Skype accounts — where we sat and took calls and texts, fielding questions about everything from where to get medical assistance, to what the ‘chances’ were for asylum in various European countries.

Doing this prototype early in the process, allowed us to understand how a digital information service could reach people, with crucial information, early. It showed us how the service could be improved upon, and things to avoid. We learned that it was important not to raise the hopes of people, with the language we used, and for people to understand that the answers to their questions were tailored to them, their nationality and their specific situations. We also learned that just having the on-boarding poster visible at the border, was not enough. Users understood the service more clearly, when it was recommended to them by a friend of volunteer. Finally, we realised that refugees in these hotspots, had not yet reached their destination, and the more basic needs of safety, food and other essentials have to be met, before any more detailed information on legal requirements could be given.

On-boarding — prototpye
We’ve conducted a few different experiments with how to get users to interact with a chatbot. One prototype focused only on the on-boarding process, and whether refugees’ informational needs could be met purely with automated information, or, at what point, human intervention would be required.

After sourcing legal information from legal advisors in Europe, we structured it into a simple chatbot, and created simple on-boarding materials. With posters, flyers and prepaid SIM cards with the service advertised, we went to Flensburg in Germany, to test with a local volunteering initiative. We wanted to simulate how we could use volunteering networks, to distribute the materials and on-board users remotely. The onboarding materials were given to volunteers, with a simple introduction. Then we observed how they got refugees to use the service.

“We need more cards! It’s awesome!”
Observation of volunteers distributing on-boarding materials

The volunteers hung up posters, which were quickly translated to other languages in handwriting, around the station, handed out flyers, and gave out the SIM-cards to new arrivals. One volunteer that we hadn’t met before, took on the task of helping refugees sign up to the service, helping them send an SMS to the service number, and explained how the service worked, by following the instructions on the poster. We found that using the position and work the volunteers were already doing, our service was distributed and trusted. This meant that the service later be scaled, so that humanitarian staff, translators and volunteers, could all become on-boarding agents for a more refined, chatbot information service.

Facebook — prototype
After several elements of what today is Refugee Text — a chatbot for refugees, was tested and iterated upon, we continued exploring different channels and ways of reaching our target audience. We ran workshops with experts, asylum seekers and volunteers to understand where and how on a refugees journey, information should be made available to them. One of the main findings was that different information, at different levels of detail, was appropriate at very different geographical areas. For example, it was more useful to learn about the required documents for asylum applications early, so that people had time to source them. Similarly, learning the number for emergencies in the Mediterennian, once you are already in Europe, wouldn’t be as useful, as if we can target that information to users outside of Europe.

We also learned that the main channels of information for refugees on the move, was social media and private WhatsApp groups etc. Coincidentally, Facebook launched developer tools for their Messenger platform around the time we were looking at launching Refugee Text.

“Refugee Text” Facebook - prototype

At this point, we decided to launch a service ourselves, together with asylum experts in Denmark, Germany and Sweden, who provided the information. This prototype allowed us to meet refugees where they already were, without the need for a physical presence. We targeted networks and groups where information between refugees was already being shared, and managed to gain a considerate amount users. This prototype showed us how cost-effective digital distribution can be, even with hard-to-reach populations.

We continue using prototyping throughout our process. Currently, we are nearing the end of designing and developing a standalone CMS for chatbots. The need for this CMS came from launching the Facebook Messenger service. We discovered that we needed a better system to effectively update and translate information. In order for us to replicate the pilot with credible humanitarian partners, we would need to provide them with an easy to use interface to maintain and monitor their chatbot.

We first prototyped this CMS by combining Google Sheets “databases” with webhook calls from our chatbot running on We then used Google Forms to allow organisations and experts update theinformation held in the database, with the latest updates. This prototype worked so well, that our first pilot and prototype of Refugee Text, is still up-to-date and speaks with 10–20 users each day, delivering information and advice on applying for asylum in Germany, Denmark or Sweden.

Prototyping ideas early doesn’t come naturally in this sector, but it can be an extremely effective way for organisations to deploy new ideas, that will save them time and money. We feel a moral obligation to ensure that all of our prototypes, whether paper or digital, bring value to users, who may be in extremely unfortunate circumstances. We’ve seen how just being present, volunteering, and working alongside refugees and humanitarian workers, can give us the inspiration and understandings we need to turn concepts into reality.

Preview of our chatbot CMS

Other posts by Social Projects:
Social Projects: Helped UNHCR Deliver Information Online