Ways of Doing
When to start small and do it for real
There are some types of service design problem which are first and foremost about user experience. The assumption that must be tested is: can we design a form that someone can complete, or a website that someone can navigate.
This is the type of problem that the classic discovery->alpha->beta-live evolved to address. It prioritises user research based on mockups and delays use by real users until there is enough certainty about the design of the user experience.
There are other types of service design problems where the (often quite risky) assumptions that need to be tested are not about the design of a user experience. These include:
- understanding complex user behaviors enough to be able to know what to design (e.g. complex combinations of health conditions or businesses exporting to multiple products to multiple markets)
- if user behavior can be influenced (e.g. switching career, getting healthier)
- what parts of a service can be automated and which should not (either because they demand a human interaction, or the scale does not warrant automation)
- that a particular bit of technology can do what people hope it can in real use, not just a technical proof of concept (putting data land ownership in a blockchain, or creating a chatbot from a corpus or text doesn’t really prove anything)
- that the organisation developing the service willing to really commit to changing what it does
These were the types of problem that Universal Credit had to try and solve in 2013. The team used the language of alpha and beta, but those were just labels attached at convenient points of what was, and still is, continuous delivery of a real service.
Starting with a very small number of real users from a postcode area in South London, a minimal (but real) user interface and a lot of manual processes, the approach was to build something that was just real enough to get real users flowing through the system.
Sometimes starting small and doing it for real is the only thing that works.