Pregnancy Chatbot
Problem
The Nivi chatbot was created to help people access reproductive healthcare in India, Kenya, and Nigeria. This case study specifically looks at the flows I created for the pregnancy and baby guides. There were 2 main goals of the bot: 1) To make sure people went to get their health check-ups at the correct intervals, and 2) To ensure that clinics were providing a high level of care, dispensing needed health information and immunizations when appropriate. We also used the messages to educate users with the most current guidance from the World Health Organization (WHO).
Solution
I came on board to the Nivi team to develop the Nivi bot persona, design the customer experience and conversational flows, and integrate the best practices and medical knowledge from the WHO and HelloMamma, a digital health platform we modeled our medical interventions and outreach from.
Collaborators
Product Manager
Public Health Experts
Localization Specialists
Content Strategist
Flow charts
Next, the Product Manager and I created flow charts for the messages. I designed the interactions in a Google template based on the messaging using the bot persona and voice and tone I’d created. We then moved the designs into Miro to share with the engineering team.
Bot persona
I started by creating a bot persona to inform the voice and tone of each interaction. The team gave me the direction that Nivi’s voice should be like your “cool aunt,” someone non-judgmental, trustworthy, caring, and intelligent.
Localization
Finally, the Nivi messages were translated into Hindi and Swahili. The plan was to launch the bot in Kenya, Nigeria, and India.
Data showed that many people still chose to have the messages sent in English. Because so many people who were using the bot had English as their 2nd language (if not 3rd, or 4th!) we made sure that our messages were at a reading level of grade 5 or below.
We also made sure to use British English, and created a word list and content guide to document these decisions.
Journey map
Then, my content partner and I created a journey map of the Hello Mamma messages we were using as a guide. We took the 32 week antenatal plan and the 16 week postnatal baby guide, and color coded key health information and interventions to ensure they weren’t repeated too closely together, and were sent in the correct weeks. We then put a plan together using the journey map to lay out the 3 messages we’d send per week and in which order. This visual was also an excellent way to communicate to the rest of the team, and easy for the public health experts to review.
Antenatal care flows
Using the Google templated created by Hillary Black, I wrote messages to be sent out 2x a week modeled on the HelloMamma texts.
There were 32 weeks total, plus 8 visit check-ins / reminders to make sure users were seeing their health care practitioners. In those reminders, we’d ask to make sure the clinics were providing good care, and that patients were getting the required information and immunizations at the right time.
Postnatal care flows
After the 32-week pregnancy guide, users could opt in to continue messages post-birth. Our baby guide was 16 weeks long, plus 5 appointment check-ins / reminders to make sure they were getting quality care. As clinic satisfaction was one of the main goals of this chatbot, it was really important to gather this feedback directly from users.