title. Pite Pite Financial Chatbot
date. October 2017 - September 2018
location. Yangon, Myanmar
Role: Design Product Manager at ThitsaWorks
Problem: How might we leverage digital solutions to provide financial education and access to financial products to Myanmar people in a more accessible way?
Process: We conducted a series of in-depth interviews with potential users, subject matter experts and financial providers, as well as user-tests and scoping of previous research on financial behavior of Myanmar citizens. These methods were employed throughout the product lifecycle as an iterative process.
Outcome: A minimum viable product of a chatbot launched on Facebook Messenger and Viber available for a community of users that includes a product directory, eligibility checklist, a search feature for the nearest financial institution, and a financial calculator. By the end of my tenure, a first client has signed on to using the bot as lead generation tool, reaching hundreds of clients that have completed an online loan application through the bot.
The lack of financial literacy in Myanmar remains a hurdle to true financial inclusion – a crucial driver of sustainable economic growth of the country. 80-90% do not have bank accounts and access to formal credit and financial services in Myanmar.
Yet, the country is a ripe environment for innovation and is experiencing a surge in the deployment of new technologies. Currently, it has a mobile penetration rate of between 90%-100%, with a reported 80% smart phone usage rate from mobile providers and growing a subscriber rate of almost 300% and three telecom operators. Additionally, internet usage is increasing at a rate of 97% or higher.
ThitsaWorks, a technology start-up in Myanmar that provides financial technology solutions wanted to increase financial inclusion through the use of technology. Through both formal and informal in-depth interviews with small-business owners, taxi and trishaw drivers, and others Myanmar-people that were unbanked, we found the lack of financial education to be a huge gap. Our conversations revealed insights on why consumers were risk-averse, a better understanding of the depth of the knowledge gap they faced along with the type of information they desired.
These insights allowed us to develop a proof of concept, or prototype that included a few core features to test in the field environment.
Our prototype featured a chatbot where users could ask questions pertaining to calculating loan interest, finding out how much they could borrow and finding their nearest MFI. We then took that concept and interviewed in the field again, this time with subject-matter experts or people that worked closely with our target population (the working poor) and understood some of their attitudes and behaviors. The interview covered topics from messaging of the product, digital literacy, content to include and financial behaviors. Simultaneously we also conducted some moderated user-tests to elicit initial responses from users.
From our interview with subject-matter experts and users, we drew an additional set of insights about the users' possible pain points and behaviors when interacting with the product. We mapped these insights on a User Journey Map to understand the touch points and behaviors that may happen in the course of this interaction, on the interface level, content level and feelings as well. The Journey Map especially proved to be helpful for continually directing our attention to the needs of the user and the parts of the product flow that we needed to pay careful attention to. With the synthesis of insights, we made corresponding adjustments to make the chatbot more user-friendly, especially with the awareness that the lack of digital literacy remained a constant challenge. These included facilitating the navigation of preferred languages, icons that were easily recognized, and tabled one feature that proved would be problematic.
With a more polished proof of concept, we were able to start presenting our product idea to financial providers who we thought would find the product helpful in educating their clients. Many providers were not only receptive but they were interested in the possibility of using the chatbot as a lead generation tool for potential customers. Our meetings triggered more brainstorming sessions on various workflows of the services or products that the chatbot could facilitate (such as loan applications, finding agents of financial providers etc) that would be aligned with the interest of customers.
This marked a turning point in our project where the chatbot would start serving the dual purpose of financial education and also providing financial access. Our continuous exploratory meetings with stakeholders and business partners led us to revise many iterations of product workflows and adjustments in information architecture so that the interface made sense for both the end-users (the consumers) and the businesses using the chatbot.
We then needed to take this list of potential products (that included savings, insurance, loans) and find out the level of interest from potential users as well as understanding what types of information they would prioritize gathering in order to consider such products. One method we used to gather users' input was a prioritization exercise that presented cards of each product or information module.
This method validated much of the content that we planned to include in our backlog of features. For example, users wanted to know how they could tell if they were eligible for certain products, ways they could prevent default of loans, and how to tell if institutions were credible.
ThitsaWorks is still developing the chatbot today using an iterative design research process. To see the latest, click here.
Eventually, after working with our business partners that would be using the chatbot for lead generation along with our developers, we were able to launch a minium viable product with some core features. We started an online digtial campaign on our Facebook page driving users to the chatbot. From there we were able to monitor conversations through the Facebook messenger to see where users were dropping off, their pain points and core questions.