Starting With An Unfamiliar Product
One of the things I've enjoyed most about working in Credit Strategy is the opportunity to move between different products and problem areas. The challenge, of course, is that every move comes with a learning curve.
When I joined Zopa's Point of Sale (POS) lending team, I was stepping into a product with its own commercial dynamics, risk drivers and ways of working. There was no shortage of information available, but it was spread across documentation, data models, pricing frameworks, historical analyses and conversations with colleagues who had built and developed the product over a number of years.
Before I could confidently recommend changes or investigate issues, I needed to understand not only how the product worked today, but also how it had evolved and why previous decisions had been made.
The Problem Wasn't Data, It Was Context
Having worked across a few different lending products, I've found that access to data is rarely the thing that slows you down. More often, the challenge is building enough context to use that data effectively.
Understanding which analyses are still relevant, where key assumptions came from, how different datasets fit together and whether somebody has already explored a similar question can take far longer than running the analysis itself. Most analysts will recognise the experience of starting with what appears to be a straightforward question and spending the first few days reading old work, tracing logic through data models and piecing together how different parts of a product connect together.
Having encountered that challenge before, I wanted to see whether there was a better way of approaching it when I moved into POS.
Capturing Years Of Knowledge
Rather than using AI primarily as a tool for generating code, I focused on using it to help navigate and build context.
I brought together product documentation, data models, pricing model code, historical analyses and information about key datasets into a single workspace. Alongside that, I created a set of instructions that helped the AI understand the product, the relationships between different data sources and the analytical approaches commonly used within the team.
The interesting part was that this wasn't a one-off exercise. As I worked on new projects and developed a deeper understanding of the product, those instructions evolved as well. Over time, this created a feedback loop where each investigation made the environment more useful, making it easier to build on previous work, understand historical decisions and quickly find relevant information when a new question arose.
The biggest benefit had very little to do with generating SQL or Python. What changed was the amount of time spent rebuilding context whenever a new piece of work landed on my desk, which meant more time could be spent understanding problems and developing recommendations.
More Time For Investigation
The impact became clear across a range of projects.
One example involved reviewing the performance of a strategic lending partner where outcomes were beginning to diverge from expectations. Rather than spending days reconstructing the background, I could move more quickly into understanding the drivers behind performance, analysing customer cohorts and assessing the potential implications for the portfolio.
I saw similar benefits when reviewing affordability assessments and policy changes. Existing frameworks, previous investigations and modelling approaches were readily available, allowing more time to be spent evaluating options and understanding likely customer outcomes.
The quality of the analysis still depended on the same level of rigour and challenge. What changed was the speed at which I could move from a question to a well-informed investigation.
Tools Change, Judgement Doesn't
There's sometimes an assumption that tools like this reduce the need for expertise. My experience has been the opposite.
Credit Strategy still relies on people who can understand risk, challenge assumptions, evaluate trade-offs and make sensible decisions when faced with imperfect information. If anything, good judgement becomes more important. Analysts need to understand where conclusions come from, identify when something doesn't look right and ensure recommendations are grounded in evidence rather than convenience.
AI can help surface information and accelerate analysis. It can't make decisions for you.
Why This Makes Credit Strategy More Interesting
For me, the technology is only part of the story. The more interesting development is that analysts can spend less time looking for information and more time engaging with the problems themselves.
Building context will always be an important part of Credit Strategy, particularly when working across multiple products and increasingly complex datasets. The ability to capture knowledge, make it easier to access and continuously build on previous work simply means more time can be spent on understanding problems, challenging assumptions and improving outcomes for customers.
That's one of the reasons I think it's such an interesting time to be working in Credit Strategy. And if you're someone who enjoys combining analytical thinking, technical skills and business problem solving, we're always keen to hear from curious people who enjoy tackling similar challenges.
Click here to apply.
