Problem Overview

The Police Brandenburg wanted us to redesign the current digital service for criminal complaints. They already had a installed online platform, however there were many pain points that needed some rethinking to make the online service more attractive, as 90% of users would still prefer to report the complaints via phone.

The team had the task to not serve the clients needs from the perspective of an agency, but to rather found a startup, that could solve problems of the public sector while being economically viable without funding.

Solution Pitch

Our solution is dedicated to the use case of shoplifting. The mobile application „Robo Retail“ rethinks the way criminal reports for shoplifting can be processed from the side of the retail company. With the help of AI and automatisation applied to the scan of IDs and products, detectives and employees are enabled to report shoplifting in a legally correct and efficient manner. Consequently, the solution not only saves time and loops of communication, but also builds a dataset that informs the organisation of detectives and prevention of shoplifting.

About the Client

The Brandenburg Police Department consists of the main police headquarters, four regional police headquarters, the state office of criminal investigations, and the office for special services. The main Police Headquarters is responsible for the safety of 2.5 million residents in 14 districts and 4 cities. Its area of responsibility encompasses 29,476 km2, stretching from the northern-most Uckermark district to the southern-most Elbe-Elster district.

The all-encompassing police reforms from November 2011 have succeeded in building a modern police structure that can respond effectively to the changing living conditions in the state of Brandenburg. This included building a strong online presence and digital services. One of these services is the one of criminal complaints that is divided into four main areas: One for bikes, one for car-related issues, one for shopping on the internet, and one for others that include criminal violence and shoplifting.

Understanding the Problem

To get to know the client and their respective problem, the team met with the Police Brandenburg in a common meeting. The most important finding from that meeting was that so far only 10-15% of criminal complaints are reported online, in most cases the police would be called via phone to come on site. Furthermore, the presentation of the flow of the website's current online form for criminal complaints was shown.

In the next step the analysed the usability of this website more deeply and found several pain points:

- As the way the form was presented for four different types of complaints, the form so far has been standardised and lacks in logic sequencing.

- There was not distinction between required and optional fields. Often the applicant was asked to give information that wasn't needed in the end and had to click through all of the components.

- There was no way to attach any photos or documents of proof. That means the responsible police officer has to ask afterwards for details and material of proof, which means loops in communication and a prolonged time until the case gets solved.

Additionally, the team did extensive desktop research on the topic of criminal statistics and on the technicalities and design of website forms to inform ourselves about the topic.

Another method to download assumptions and gain an understanding of the users and stakeholders was the stakeholder charrette. It clearly showed dependencies and interrelations between internal police employees and the citizens related in the criminal case. Applying this method also made the team realize the importance of the role of the public prosecutor's office. Another assumption that came up was that the applicant might be scared to do something wrong when filling out the online form and therefore prefers to let it be done by an officer. That was one of many assumptions we wanted to know more about in user research.

User Research – 1st Round

In the first round of user research, we wanted to dig deep into the process of criminal complaints from beginning of a crime to the case being solved by the prosecutor's office.

Therefore we conducted 8 semi-structured interviews digitally.

Those included: 3 applicants of criminal complaints, 3 police officers and 2 lawyers.

The three main insights after synthesis included the following:

- There is a conflicting interest between a good reported complaint with options of free text that saves time in the process afterwards, but one online form that is concise and doesn't take too long.

- When an officer gets ordered on-site to do the reporting themselves, they often simply use pen and paper to document and later on enter the data into their software system. This not only results in mistakes but also means a lot of bias as there is an additional layer of reporting between the applicant and the decisive lawyer.

- The applicant is not informed about the loops it takes and the current status of their complaint. As the process can often times take up to several months, the applicant is left in the dark with a lot of uncertainty.

Service Design Concept – 1st Round

After a round of ideation, we came up with the concept of OSAAS. For shop owners affected by vandalism who have little time and need to know the status of their criminal complaint in order to deal with insurances, OSAAS is a digital platform that offers a faster form to fill out and regular status updates. Unlike the form by the police Brandenburg itself, OSAAS uses simple language, logical components and offers the feature to upload pictures of the crime.

Ultimately, the four key promises we wanted to make in our concept were the following:

To build our idea as a prototype, we were inspired by the website of the german post that also offers status updates on sent packages and a website called dickstinction that allows to fill out a super quick form within a few minutes to report any type of sexualized pictures sent via social media.

In order to test the solution-market-fit, we quickly build a landing page. Our goal was to test the users' openness to pay for that service and how much they would be open to be paying. We came up with different pricing modells and wanted to draw conclusions for our business modell based on the number of clicks on our website.

As a feedback, the experts of the course questioned our solution in terms of a specific user group. So far the use case was very broad, we did not do enough research on the most cases of criminal complaints nor did enough interviews with applicants sending in a report via the online form. Consequently, we iterated our concept.

Iteration

The question on how to make a sustainable and profitable business model led the team to a new direction of looking into B2B concepts. Therefore we did another round of desktop research. Especially striking was the „Kriminalitätsbarometer Berlin / Brandenburg“ that showed that shoplifting was a serious threat to businesses and the most common crime. Additionally, a study by the EHI Retail Institute showed the extreme number of 4.4 billion euros that get lost due to inventory difference and shoplifting each year in Germany. The Kriminalitätsbarometer explicitly stated, because retailers would often not report the crime, it is not taken seriously and the number of criminal complaints must increase in order to reduce the crime of shoplifting over all.

We found a standardised paper form on the Internet for small retail business to report criminal complaints. The three pages long paper form, again, included pain points we knew well from the digital form of the police, like missing logic and the need to put in the same personal information in regards to the shop etc. time and time again. We gained the insight that the time needed to fill out the form could be reduced by using digital means of automization and pre-filled components. Our analysis of the different components can be seen below.

User Research – 2nd Round

To build empathy with retail employees and detectives that are using this type of paper form daily, we conducted interviews. These were held in a semi-structured manner, partly digitally and partly directly in the field. We wanted to find out information about shoplifting (how often? what products? how can it be detected?), criminal complaint forms (who fills it out? what is the process behind?), and possible future business models (what are the calculated costs for one complaint? how can a digital solution be implemented into existing processes? what needs to be considered in legal terms?).

We interviewed police officers, employees, branch managers and detectives from the following companies: Edeka, Rewe, H&M, Alnatura, Nanunana etc.

The three main insights after synthesis were the following:

- As shoplifting crimes below 50 euros were often dropped by lawyers, employees didn't see the pressing need to report as the timely invest was too high for it to be worth it.

- From more legal weight is trespassing. That is the case when a former shoplifter that got prohibited to enter the shop again ignores the ban. However, since all the data is collected manually, shop owners often don't know if a person has committed the crime of shoplifting before and could be sewed because of ignoring the ban.

- Criminal reports are most of the time filled out manually by an employee or detective, but need to have a signature from the shop manager in order to be legally correct.

Explorations with Speculative Design

As the topic of this design sprint was data driven experiences, we were inspired to think beyond the possible, leaving ethical considerations behind and speculating about the future of shoplifting detection.

Three possible speculative scenarios were developed. The first one in which the use of cameras with face detection could spot criminals that have a current ban from the shop and automatically send criminal complaints of trespassing to the police.

in the second one, cameras would also be used, but in this case with thermal image in order to detect shoplifting and conspicuous behaviour beforehand.

In the third one cameras would detect shoplifting in self-service check-outs. Through payment with credit card, an automated system would thus identify personal details and send out a criminal complaint reports immediately.

All of the three exploration would need data as a basis to learn and develop their services. Especially the first one was explored in more detail through Google's teachable machines and one can expect an data-driven experience to be feasible in the foreseeable future.

Final Service Design Concept

Building on the findings from our desktop research and further insights from user research, the team developed Robo Retail. This solution reinvents criminal complaints in the case of shoplifting by digitalizing the process in an efficient way.

Before we build our interactive protoype with figma, we did some paper sketches individually in the team, then looked for commonalities and synthesized the ideas into one final solution.

The mobile application has three main functions:

1. To be able to scan ID cards and therefore fasten the process of personal data entry. With additional ML technologies the personal data of the ID can be checked.

2. Secondly, the stolen goods can be scanned via a QR code. Consequenlty the final financial amount of the shoplifting can be calculated automatically.

3. Lastly, the app allows to quickly describe the facts of the case. By typing digitally and giving suggestions, this process won't take as long as manually written documents before.

To respond to the pain points we identified, Robo Retail includes three main benefits:

- Our solution fastens the process. From 30 minutes of filling out the paper version, we promise to have a criminal complaint done within 5 minutes, so reducing the time around 5/6 of it.

- By building a dataset, the retail shops will be able to identify people that have committed a crime at their shop before and might be under a current ban and can therefore automatically sent the criminal report for trespassing.

- The dataset also allows to draw conclusions from statistics like what products were stolen most often or when would be the best time to hire detectives. This will enhance preventative measures.

User Testing and Evaluation

To measure wether the solution fits the market, we initially built a landing page to advertise our solution. But the team quickly learned, that the decision makers won't be looking on websites for new technological solutions but need to be contacted over our own network.

This is why we recruited retail employees and detectives over own contacts.

In several rounds of interactive usability testing with the figma prototype we got a lot more insights into the users behaviour as well as current problems with existing processes in the shops.

The prototype itself was received quite positively, users could quickly follow the logic of our solution. A big learning we had was that although having the initial feedback from experts to reduce the number of pages for minimal effort and inputs, users would be able to follow more easily if each step of the process would be on a single page, just like the one on paper.

One of the biggest question we had before implementing included how to design the technical touchpoint with the police. How could the data of the criminal complaint be delivered and processed in the software system of the police afterwards? An ideal version we imagined would be an API from the police's side, but this won't become reality due to data security doubts. After some testings we found out, the most pragmatic way to be able to launch the mobile application as soon as possible would be to export the data into a pdf document and send it via email to the police, as most users were already in contact with the police over that channel.

Discussion and Ethical Considerations

There are two key points to discuss in terms of our solution.

The first one being data security. As our app would be an external service provider but needed to process personal data, the question where to store that data is a crucial one. The team's suggestion was to save the data onto servers of the retail shops themselves. So far, the detectives were using computers and digital softwares for their video monitoring. So servers would be already installed.

In times of pen and paper, criminal complaints would be copied once and put into a big folder where detectives could have access afterwards. In most cases, the police would give one case a certain number and later on report on the outcome of a complaint. Detectives therefore had to have access after sending out the complaint in the first place. To rebuild that function, our mobile solution would also need to give access to cases later on and therefore store the data which would need to be encrypted. Interesting to see is that currently papers are printed and stored somewhere in the shop offices, there seems to be no problem with data security. As soon as solutions get digital, data privacy becomes the hottest topic to solve.

The second key point to discuss would be the ethics behind our solutions. It was clear to us, that our solution could negative impact on people that committed a crime. So far, occurrences would often fall under the table when there were not „heavy“ enough (in terms of value of the stolen products). Having a mobile solution that fastens the process, one can imagine that even small misbehaviour would have long term impact with housing ban etc. The team has been brainstorming and thinking about ethics all along - and how to keep a humanly perspective in it.

One of many thoughts went into the direction of a function for shop managers to manually decide how long a housing ban would need to last instead of pre-defining the amount of years no matter who and what the circumstances.

Outlook

The team is currently continuing to work on the solution. The next steps include the following:

Find more retail companies to test with and not only test digitally via a shared screen but using a real phone in the field.

Iterating the user journey and existing screens to enhance usability and easiness of use of our prototype.

A missing function of our prototype is the one to let the shop manager sign the form at the very end. This would need to be done via a signature function digitally that could be send, so far the detectives would print the form and give it to the manager.

Make the touchpoint with the police interactively testable. So the prototype needs to be able to at least export a pdf that can be send to the police and letting the police have a look on whether this pdf would be legally acceptable and enough to be processed internally.

Continuing to redefine our service blueprint in order to identify blind spots in the user journey.

Calculate the costs for software development and marketing in order to see how much money would be needed to launch the first MVP.

Rework the pitch deck of the solution and try to approach decision makers showing our mobile application and its benefits.

Learnings and Reflection

It was the first time the team of three designers got together to work on a service design challenge for the public sector to reinvent the police. There were several learnings within the group after taking on this challenge:

It helps to not be imbedded into a rigour system but firstly think of oneself as an outside party like a startup in order to design with high ambitions and less (financial and legal) restrictions. To make certain recommendations for actions can be done later on, but to have impact on the public sector it makes sense to look for solutions that can have a sustainable business model behind.

Although very confused and unsure at the beginning in which directions to head, two crucial approaches really helped: one was the openness (and maybe force) to iterate and rethink the idea all over again. The second one being detailed user research and the connectedness to the user's perspective all along.

The decision to go for B2B solution might be very profitable, but to find customers, in this case decision makers high up in companies, is even harder than usual end users.

Overall, the team was super happy with the outcome, the team constellation (we still had lots of fun during the most frustrating moments) and the impact the idea had on our own learnings and reflections but also the ones testing and evaluating it.