A few years ago our Mint customers were cumulatively losing millions as overdraft fee due to insufficient funds in their bank accounts. And these loses were triggered by smaller transactions under $50! Committed to powering prosperity around the world, we knew we had to do something for our customers who live paycheck-to-paycheck.
We needed to find a meaningful solution that could help Mint users avoid such exorbitant non-sufficient fund fees. So, what did we do? We brought Data Science and Behavioral Economics together and intersected them with design thinking. This combined power aided us to dissect data on the real-world financial transactions by Intuit customers and unearth the circumstances that lead to an overdraft. Our rapid experimentation also revealed that loss aversion is a strong motivator for action.
These impeccable insights helped us to build a machine learning model that could accurately predict overdrafts, and even trigger alerts allowing customers to take preventive actions. As of March 2019, Mint sent out over 650K alerts about possible overdrafts, and users acted on those saving significantly on fees!
Data science & product engineering – how do you connect the dots?
The Mint example demonstrates how data, engineering, design and deep customer understanding work in tandem to create world class products and customer experiences. But then bringing together these diverse sets of expertise – each with a distinct lingo and unique perspective – is far from easy.
Both operate on two separate premises. While data science is all about building predictive models to foresee the future, engineering is more about structure and the present state.
It is product management that can connect the two dots. However, delegating functional requirements and AI-powered specifications to engineering and data science teams require specialized skills. Pin-pointing the actual business problem and identifying the right solution can be daunting amidst the bevy of business problems and plethora of ML models to choose from. After all, how do you decide which algorithm is more efficient to deal with a certain problem?
Product managers of the future, therefore, will need to develop a strong affinity towards statistics and trends, and an innate business acumen to be able to make data driven decisions. Unfortunately for most part, today, the alignment between data scientists and product managers does not exist. Both reside in mutually exclusive zones.
D4D is how we decode the impasse
Our key to this mystery maze is Design for Delight (D4D), Intuit’s very own version of design thinking developed inhouse and the single source of truth for all our innovation efforts. Based on three core principles: Deep customer empathy, go broad to go narrow, and rapid experimentation – D4D has been instrumental to Intuit’s success in developing the AI/ML-driven software and services that help our consumers make better financial decisions.
This principle acts as a fastener bringing together the two spectrums – data science and product management. By virtue of being in business for over three decades, Intuit is a treasure trove of customer data in the small business and personal finance spaces with QuickBooks alone crossing 4 million users globally.
Each data set represents a distinct opportunity for solving a problem. We leverage this by combining the domain expertise of both product managers and the AI team through joint D4D workshops. Participants identify addressable problems and spot automation opportunities to ensure that the end customer gets the smartest product and a great user experience (UX).
Through this amalgamation, we leverage smart use cases which have already solved customer pain points. All of these learning go into defining Intuit’s roadmap for smart products, and UX visualization.
Ultimately, it’s all about making lives easy for our customers by breaking the human-machine barrier. We believe a combined power of data science and design thinking can facilitate access to information, and maximize user experience. It is this overarching intent that led us to augment our offerings such as the QuickBooks Assistant which today is helping freelancers access information and save time by asking questions in natural-language. This desire is guiding all our business decisions including the recent buy out of the data analytics startup Origami Logic – a move expected to accelerate funding, payments and better access capital for our small business customers. During a press interview, our CEO Sasan Goodarzi articulated this very well, as he stated that revolutionizing our customer engagement is the sole reason why Intuit is gravitating towards being “an AI-driven expert platform.”