Digitisation in administrations: From problem to prototype in 12 weeks

The Nähwerk is a service offered by the social institutions and businesses of the City of Zurich and offers places for various target groups of work integration. Long-term unemployed adults and young people without a school connection learn new skills at the Nähwerk and prepare for a new or re-entry into the labour market.

The work-agogues of the Nähwerk have followed the appeal of the city council and contacted Digi+. They believe that their many manual processes can be improved through digital support. This much can be revealed here already, the solution worked out can probably save the sewing factory around 175 person-working days per year.

What can a sustainable and meaningful digital transformation in the administration of the city of Zurich look like? We are certain that technological progress is only social progress if people are at the centre of it. Accordingly, the first step in the cooperation with Digi+ was not to solve the posed problem, but to analyse what is the cause of it and whether a digital solution is the right one at all.

From problem to tested prototype in twelve weeks

Using the method of future experiments as a model, we set about the following task:

1. find out what the problem is,
2. how it could be solved,
3. validate ideas for solutions,
4. concentrate on the most promising idea.
5. prototype the idea and
6. perfect it through iterations.

Not how, but what

In projects, we often find that problems have not been identified precisely enough or that technological wishes and proposed solutions have already been incorporated into problem definitions. In our experiments, we therefore almost always start with a loop in the problem definition. What exactly is the problem and where does it come from? Based on the already complete service blueprint provided by Digi+ and interviews with Arbeitsagoginnen of the Nähwerk, we were able to sharpen the problem again. Subsequently, we defined our task by means of so-called «How-Might-We» questions and created framework conditions by which ideas are evaluated by means of learnings in the problem identification.

Create, filter, validate and focus ideas

In the first step, we freely generated over 70 ideas together with Digi+ and made them easy to understand through sketches. We then filtered them through the framework from the problem definition and evaluated them for the first time. We created hypotheses for solving the problem for three ideas and validated them with the help of low-fidelity prototypes with employees of the sewing plant. We made many decisions in a decentralised way within a short period of time. Our tool Agree has simplified these decision-making processes, and all decisions can be viewed and documented by everyone at any time. Based on this, we decided together which idea we wanted to develop into a prototype.

A prototype can take many forms: a website, an app, new processes or, for example, changes in working environments.

Prototyping, trying, iterating and iterating again

In the last step, it was decided in this concrete project to develop an app. What must the app be able to do in prototype status, what would be nice to have and what is unrealistic? We used user stories, i.e. "As I want to ", to record what the users' expectations of the app were. Based on these user stories, we implemented the first versions of the prototype. Already in 2000, before the first iPod or YouTube, Jakob Nielsen published that you only need five tests to detect 80% of all errors in a product.

To get as many insights as possible, it is better to test three iterations with five people each instead of one iteration with 15 people. We implemented and tested a total of four iterations of the prototype over two weeks in order to gather as many insights as possible.

Finally, we checked with the project stakeholders whether the final prototype solves the problem and the How-Might-We question.

Experimentation is worthwhile

Based on the findings from the prototype phase, Digi+ and the sewing unit carried out an impact assessment. Among other things, it was examined to what extent the prototype could solve the sewing unit's problem, how much time it would save in the future and also what the development would cost. Through the implementation of the prototype, both the reduction of errors and the simplification of processes could lead to a saving of about 175 working days. In addition, the employees in the sewing plant would gain valuable digital skills that would help them personally and in the labour market. This could be realised with a one-time expenditure of 57 working days. Accordingly, an introduction would be economically advantageous after just three months.

Even though we disagree with much of what Jeff Bezos says and does, we agree with him here. It is challenging to go into a project without knowing what the end result will be. It can be even harder to know that the outcome may not work. But just because a prototype doesn't produce the desired result doesn't mean you haven't learned anything - quite the opposite.

Experimenting before making a final decision can save a lot of money. It can make results more inclusive and draw attention to mistakes before they are made on a large scale. We therefore believe that this openness to results leads to much more valuable and sustainable outcomes than a decision made on paper.