Public organisations work in their silos. Every organisation tries to optimise their own services; all too often without realising that user journeys often span between multiple organisations.
When an immigrant wants to come to Finland to work she needs not only immigration services from Migri, but also a tax card from Vero (Finnish Tax Administration), a social security card from Kela ( Finnish Social Insurance Institution) and many other public and private services. Nevertheless, when Migri thinks about its services it tries to optimise only their small part of the customer journey, without paying attention to the interconnectedness.
The idea for a chatbot network has been around since 2017: Connecting services from different public authorities in Finland to build comprehensive services around users’ events-in-life (e.g. immigration). In Inland’s concept idea we utilise independent chatbots that are interconnected through an added network layer. Each organisation remains in charge of their own content, while at the same time providing a common service that centers around user needs in their specific situation.
Migri, PRH and Vero connect via the chatbot network
In March 2018 we started experimenting this service together with the Finnish Tax Administration Vero and built a demo aimed to help foreign workers navigate their immigration & taxation journey to Finland. The special focus was to be on entrepreneurs. This phase of the project ended with a public demo, which was received very well. Participants encouraged us to pilot this service as soon as possible.
In August 2018 the Finnish Patent & Registration Office (PRH), in charge of company registration as well as patents in Finland, joined the project and the second phase started. Together with the three organisations onboard (Migri, Vero, PRH) we built a common service to help foreign entrepreneurs start their business in Finland. This service is piloting from December 2018 to June 2019.
The collaboration and network building started in March 2018.
Double diamond-design process
In this project we used a classic double diamond-design process already twice: In both phases of the project (March-June 2018 and August 2018-June 2019) we started with user research, conducting interviews & surveys and then analysed the results to define the focus group: foreign entrepreneurs who have decided to start their business in Finland but don’t know where exactly to start. Based on the research phase we then created content for the three chatbots, that served the user needs. This content was constantly tested in usability tests with the end users. Based on the usability tests we adjusted the contents before we finally put it live in early December 2018. Lots of work was put into designing, testing and adjusting the transfers between the three organisations. While the service is in pilot phase the team continues to build additional content based on needs we recognise from conversations logs as well as further user tests.
The main evaluation of this project will take place during summer 2019.
During the design process we understood when transferring from one organisation to another there are three types of transfers: Reactive transfers, when the user asks a question about a topic from another organisation, proactive transfers, when the bots give additional information and refer the user to another organisation, manual transfers, when users specifically ask for another organisation.
Nevertheless in the process of building the “Starting up smoothly”-service we learned a lot of new things: We see the importance of teams having a room together and working in close collaboration, even if each organisation creates their own chatbot. Our view is that technology comes only after understanding the users and structuring the content for their needs. This practice has been adapted by the whole team of around 25 people involved within the three organisations, as well as our technology providers.
Inland has had a big role in shaping this project: Our way of developing the chatbot network is implementing pilots with different public agencies, rather than building a huge infrastructure and hoping that everybody would join. In every phase of the collaboration we learn more about the questions: How to collaborate with other organisation’s teams? Which content should we develop next based on customer needs?
Starting up smoothly – Connecting government agency information through chatbots
See also related project: