Check your health: do the STD SexCheck
In the summer of 2014 we received an invitation from TNO eHealth, an independent research agency, to pitch a solution for a project. The question for Ecreation, part of Watermelon: create an application or tool to reach two target groups with an increased risk of getting an STD. These groups are poorly educated youth between 16 and 24 years old, and youth in the same age category but who classify as the second or third generation of immigrants from countries including Suriname, the Dutch Antilles, Turkey, Morocco and a number of eastern European countries. A second goal of the project: find an easier way to make an appointment at the GGD clinic in the instance of a suspected STD. It was very important the application would reach a large number of people.
Working for SOA Aids Nederland
With the information above, we focused on two key aspects:
- Be present on the channels where the target group already spends time
- Don’t try to act like the target group, but show that we take them seriously
We won the pitch, in part due to our experience with how this target group prefers to communicate online with automated chat solutions. Our proposal was to create a chatbot that asked the target group all the questions relevant to creating a personal STD risk profile. Using this profile and the assessed degree of risk, the individual would be given personal advice.
Before we started this project, we held a number of sessions with a focus group. We found it important to truly listen to the people who would eventually be using the platform. We asked them about our design for the application, asked them about how they search for information on more ‘private’ subjects, if they speak with other people about it and if so, how. All these answers were combined into a word web in order to analyze the most important terms. These terms were the input for the application’s first prototype. The focus group participants also created moodboards with images, text, colors and their thoughts about the subject, which also proved to be very useful.
We used the information gleaned from the focus group to choose a tone of voice, design a look and feel and create campaign assets. We continuously held meetings with stakeholders including TNO, SOA Aids Netherlands and GGD Rotterdam. These groups provided us with their thoughts about the application and additional information ‘approved for use by the government’ such as information about STDs based on triage, how to prevent them, how to recognize symptoms of an STD and what to do if you have an STD.
From idea to prototype
The first prototype chatbot was called SafetyCheck and it was made available as a web app. The chatbot continued to ask questions after giving or receiving answers in order to simulate a real conversation. When necessary, the chatbot instructed the individual how to plan an appointment in the online agenda. We presented this prototype to 14 boys and girls who initiated various chats with the chatbot. Based on these chats and a separate survey we analysed the results. We met with the project stakeholders and concluded that the ‘conversation structure’ approach wasn’t the most effective way to lead them through the risk analysis process. The most important conclusion was that the users didn’t know what they needed to ask, and when they did ask the right questions, the chatbot didn’t lead them to making an appointment online, as most users left the conversation too early as they felt the pace of the conversation was too slow.
Ask your users how to make a better product
Based on these findings, we created a second prototype that asked closed questions using a triage approach. This way of asking questions made it possible to also ask questions important for the researchers on the project: gender, level of education and sexual preferences. After changing the conversation structure, we tested the chatbot with new focus group participants. They were asked the same survey questions as the first focus group, but this version of the chatbot was spot on. It did what it needed to, checked if there was a risk of having an STD, and if so, lead the individual to make an appointment online. The focus group also concluded that the name SafetyCheck didn’t meet their expectations, so we had them think about a new name. The name STD SexCheck was chosen from a list they came up with and voted on.
Let’s go live!
After a number of co-creation sessions with the users and stakeholders, test situations with the focus groups and changing the conversation structure of the automated chat – it was time to go live and start reaching out to the target group. The official pilot started in Rotterdam, as the GGD was the clinic in Rotterdam where appointments could be made. The pilot ran from May 2015 to May 2016. We used a Facebook campaign targeted to 16-24 year olds, living in Rotterdam +15km. Within the first six months we reached our target group and nearly all available appointment slots were filled.
After these first six months we weren’t as effective in reaching the target group, so in addition to the Facebook campaign we also started an Instagram campaign. We realised we also needed to add platforms, as just hosting the chatbot on the website wasn’t attractive enough. We successfully ran a test on WhatsApp and investigated other channels including Snapchat.
We learned from this project that building an automated chat solution is a dynamic process and end users and stakeholders are of great influence on the outcome. The technology and needs of users change quickly, making it important to anticipate these changes as early on as possible.
Our chatbot technology goes live in August 2017. Create a free Watermelon Messenger account to discover all the advantages of using our tool for customer service! 😄