Project
Payment Data Analysis
Product & CX Consultant, Chillimint
I analysed millions of rows of card transaction data, declined payment logs, and customer support records across multiple financial institutions to surface hidden losses, friction patterns, and misconfigured risk rules.
- Using SQL and Excel, I built a clear model connecting user-level frustration to product-level drop-off to financial loss — giving banks and Visa a concrete picture of where to intervene and what changes would have the most impact.
- Findings directly informed product, UX, and risk policy changes that reduced friction in verification and risk flows, lowered attrition, and improved card activation and usage.
Overview
Banks were experiencing unexplained drop-offs and losses in their payments pipelines. Complaints and operational costs were rising, but the root causes weren't visible in high-level reporting. Understanding what was actually going wrong required getting into the data.
The problem
Most banks had incomplete visibility into where and why things were breaking down. They could see aggregate complaint volumes and attrition rates, but couldn't reliably identify whether users were dropping at specific verification steps, whether risk rules were blocking legitimate transactions, or how friction in one part of the journey created knock-on effects downstream.
Decisions were being made with bad or missing data.
What I did
I worked across large datasets of card transactions, declined payments, and customer support logs, using SQL and Excel to clean, transform, and interrogate multi-institution data. The volume ran into the millions of rows.
The analysis surfaced several categories of problem: misconfigured risk rules that were blocking legitimate users, confusing verification prompts generating unnecessary failures, and unnecessary friction steps that had no material fraud prevention value but significant abandonment impact.
I mapped these findings onto user journeys and operational processes — building a narrative that connected what users experienced to what appeared in the data. The key output was a clear model showing how specific friction points translated into financial loss: user frustration leading to drop-off, leading to failed transactions, leading to attrition and complaint cost.
I triangulated this quantitative analysis with the mystery shopping programme running in parallel, so client recommendations were grounded in both what the data showed and what customers reported experiencing.
Impact
Findings informed UX, product, and risk policy changes across multiple institutions. Friction in verification and risk flows reduced, card activation and usage improved, and attrition came down. Banks and Visa had a clearer, evidence-based model for where to focus future improvements.