Clinics were spending so much time collecting information from their patients, whether it was on the phone before their appointment, or on paper when the patient walked through the doors. Too much time spent when they could be doing other important tasks around the clinic. 

The current features in the app only gave information about a patient’s appointment: reminders and prep via pre-visit instructions. 

The idea to collect more of this rich information was on our minds, so we created a chatbot to do so!

The beginning process of my research started with looking at different ways others have a chat and what we liked and didn’t like about the process.

Some things that came out of this initial research:
- Conversational Vs. more direct Q&A

- Quick responses Vs. Timely response

- Less but long responses Vs More but short responses 

- Types of questions we can ask and how we want to limit, or not limit the answers from the user.
Current clinic user flow of the process of gathering information from patients.
Current clinic user flow of the process of gathering information from patients.
Initial Post-it notes on the information we need to collect
Initial Post-it notes on the information we need to collect
Wireframes and planning
Quick 'wireframes' were created to really get a sense of the conversation. These message templates were just made on a long sheet to express the sense of what the chat would look like in it's entirety. ​​​​​​​
Once observation I noticed when using other chatbots, I found to be more conversational, we had to add some breaks where breaks would naturally be in a conversation. The hardest part of understanding where those natural breaks live. To do this, I used many strips of paper for each message. I laid them out on a sheet of paper to display the entirety of the chat from start to finish. The team and I moved around questions, moved around the breaks, and decided on a flow that made the most sense both logically and conversationally.
We want people to feel like they are talking to a health ally, not a robot and in the end, the script displayed that sense and people began to truly thing someone was on the other end typing in their reply
UI Design
The UI of the chat bot design was fairly simple. We used some precedence for inspiration and played around on paper and whiteboards to show initially where buttons might live, spacing of elements, etc.  

Once we had an idea, the design of the chat itself was a little more straight forward, especially because we decided to not allow for free back and forth conversation. We wanted to limit the chat to have the responses only the clinic needs so they are able to get the most direct responses from their patients. By doing so, we needed to design all the different elements that a chat might have. Some of these elements included a single and multiselect; text input;  date input; number input.

We started in whimsical and moved onto sketch for high fidelity wireframes. These were then passed off to the developer to work their magic!

As the feature progressed, more types of responses were designed and added such as image select and uploads!
Once information was collected from the patient, what do we do with the information?
A simplistic report was designed for quick scanning and understanding for the clinics to scan and see the responses easily. 
As time progressed, this simple structure has been used as the structure to build more formatted reports with headings and sections making it that much easier for the clinic to view their patient’s responses.

Prototyping / User test 
Being in a startup, things move fast. So as the developers were coding our designs, I made a prototype through Invision and gave it to our internal team. The goal was to get a good sense on what initially is and isn’t working on the chat and where we can improve before testing it out with outside users. 
Once the chat was up on our staging environment that was send over to the clinics we were working with to see what they thought.

Overall general feedback:
Speed of replies seemed to be a little fast
Length of messages seemed to be too short

A second round of edits were made to the chat bot and after a couple more rounds of testing on staging, we were ready to go for launching this to the clinics! 
Prototyping of initial chat
Prototyping of initial chat
Example of the initial chat
Example of the initial chat
The chat today!
The chat today!
Launching the chat
Launch day was an exciting day, as this was the start to many chats in the future of the company!
By adding the chat to the list of features, our clinics can now gather so much more information from just a single chat!

This project has been a catalyst for so many different features in the app and it has grown to have so many new features and additions. It was so much fun to be a part of something that is truly one of the back bones of the company. 

You may also like

Back to Top