A FinTech pioneer founded in 2013, this banking unicorn rapidly emerged as a prominent player in the global banking and financial services landscape with over 8 million users. The company is at the forefront of the digital banking revolution, offering its customers a range of innovative financial products and services worldwide.
Since its founding, it has rapidly expanded its presence and customer base, with operations spanning Europe, the United States, and other international markets.
The company has raised close to US$ 1.8 billion from some of the world's most renowned investors.
We joined the Financial Empowerment domain as part of the Lending Foundations squad which focused on building the underlying platform to power our lending products & services. Our scope entailed connecting with internal and external data sources and training credit risk models for decision-making around lending products such as Overdrafts, Consumer Loans, and Installments.
The team was already functional without a Product Manager for 6 months and lacked the strategic direction to ensure its work was contributing to the company’s vision and objectives. Our scope involved collaborating closely with Credit Risk and Data Science teams.
We were tasked with building the foundational strategy to enable us to scale across our core markets in Europe, namely Spain 🇪🇸, France 🇫🇷 and Italy 🇮🇹 while improving our setup in Germany 🇩🇪.
We hit the ground running and onboarded very fast onto the existing environment. We met with stakeholders, used the product, and reviewed documentation around priorities, roadmap, and prior challenges.
We conducted several workshops with the team members to reflect on the ways of working, vision and scope, and upcoming initiatives to prioritize.
The result of our initial discovery was that we were up against several problems in the short-to-medium term.
We formed a strategy centered around Now, Next & Later which can be outlined below.
While executing our proposed strategy, we defined the following principles to abide by:
In the short term, we focused on improving processes, launching key initiatives that were close to delivery, and focusing our efforts on reducing legal and regulatory risk due to legacy issues in our setup.
Ways of Working
We worked with our development squad to rethink the ways of working. With a focus on transparent communication and time to delivery, we introduced new processes.
We moved away from 6-week Kanban cycles for delivery to 2-week sprints using Scrum, introducing rituals like Backlog Refinement, Spring Planning and Reviews, and Retrospectives.
Moreover, we prepared an outcome-driven roadmap based on our proposed strategy which we socialized actively with our stakeholders ensuring alignment on a common vision.
Finally, to bridge the gap between Product, Engineering, Data Science, and Credit Risk we introduced a “Credit Risk Modeling” trifecta - a dedicated taskforce with representatives from each domain who co-designed and executed on next steps.
Fixing Legacy Issues
An existing integration with a third-party bureau was in place where we reported activity on existing users and pulled credit scores for lending decisions. However, it was implemented with legacy systems and outdated requirements. Not enough attention had been given to the setup resulting in a high risk of non-compliance and thereby introducing legal fines and cases.
We focused on efforts on building a relationship with the credit bureau and kicking off a clean-up exercise. Over a million record entries were analyzed on both ends and outdated account statuses were updated.
In doing so, we introduced additional monitoring and alerts within the code to be able to narrow down where issues occurred. Thereby reducing bugs by a significant amount.
This activity was coordinated by us in collaboration with Business Operations, Legal, Data Analysts, Credit Risk, and members of the external third-party credit bureau.
Launching Key Initiatives
The squad had been working on several initiatives which needed a push towards the finish line. We jumped in to support and coordinate the go-to-market strategy for a revamped credit modeling platform and a micro-service that improved repayment capacity.
Revamped Credit Risk Modeling Infrastructure
The existing infrastructure had room for improvement in terms of automating workflows, providing easier audit capabilities, and being flexible to retrain models efficiently.
The squad had been working on implementing a new infrastructure that would serve as a key pillar to deliver our product strategy. We jumped in to ensure its launch by:
Improved Repayment Capacity Calculation
The repayment capacity micro-service was running in legacy codebases without clear ownership. Moreover, it was overestimating income calculation and thereby making inaccurate decisions.
⚠️ An internal audit had flagged these issues and requested their resolution by a defined period.
The squad had worked on rebuilding the micro-service with improved income assessment to reduce the probability of default when lending to users. We joined the effort and:
Now that we have an improved way of working in place along with the foundational technology to continue to make better lending decisions. We were ready to explore what’s next.
Explore Additional Credit Bureaus
A core part of our ability to expand to additional markets was the need to integrate with external credit bureaus in those respective markets to make informed lending decisions.
While it was feasible to deploy lending products in new markets with small ticket sizes, having a strong credit risk model to make better decisions was key to scale.
We benchmarked 6 credit bureaus with whom we conducted backtesting exercises to test their credit scores. The activity was done in collaboration with the Credit Risk team. The workflow can be summarized as shown in the following diagram.
Our criteria to decide were based on:
We narrowed down the providers to propose Germany and Spain, while the squad continued to explore options for the remaining markets.
Improved Credit Risk Model
As part of the Credit Risk Modeling Trifecta mentioned earlier, we proposed prototyping a product and market-agnostic credit risk model. The current model was trained on German users specifically using Overdraft as a product and was used on all our products and markets.
Performance was subpar and there was room to explore alternative approaches. We collaborated with Data Science and Credit Risk to prototype a model with new features that would perform well in all markets and products - one model to rule them all.
Early testing results showed a 40% improvement in our Gini coefficient. This provided us with confidence to proceed with investing resources to develop the model.
Our mission came to a close in about 9 months. We then handed over our work to a full-time Senior Product Manager who took over to continue to drive things forward.
💡 Delivered key initiatives within Lending that drove profitability for the domain by ~60%.
💡 Conducted discovery on the expansion to two additional core markets by shortlisting and liaising with external credit bureaus.
💡 Implemented effective processes and created comprehensive documentation for streamlined regulatory reporting.
💡 Extended Mission Duration: Initially planned for 6 months, the mission was extended for up to 9 months to provide continuity.
.svg){kind=small}
