Why an Open Banking credit score?
Recently we launched Signal, our highly accurate and explainable Open Banking credit score. Signal enables lenders to move beyond the limitations of traditional credit data, lend responsibly to previously underserved customers and reduce defaults. It uses machine learning and transaction data to accurately predict an individual’s likelihood of repayment, meaning you can use highly relevant, up-to-date financial behaviour data to accurately assess anybody - not just those with previous credit history.
Signal is ready-to-use, trained on transactional data and loan outcomes - amassing millions of data points - that we’ve collected over six years. This means you can make effective decisions from day one - with no need to use capital or take risks to collect data for training a model.
Why Open Banking data?
Open Banking offers an incredible opportunity to predict credit risk more accurately and for more people. Understood properly, transactional data contains thousands of rich, highly relevant and up-to-date insights into an individual’s financial behaviour
Signal leverages our entire insights library to make sense of transaction data and translate it into 500 meaningful, explainable financial behaviour features. These include characteristics that describe and predict income stability, liquidity, gambling behaviour, cash usage, affordability, stopped payments and many more. These features and their interactions are combined in a type of machine learning model (Gradient Boosting Machine) chosen for performance and explainability. Each score includes pointers for lenders to understand and evidence risk decisions.
How does it help me increase acceptances?
Using Signal, lenders are able to accurately assess applicants they previously could not with traditional credit data. Those declined using traditional data can be offered a new path, and invited to complete an Open Banking journey to provide more detail to the lender - including good management of income/expenditure, payments to savings accounts, financial commitments, stable spending habits, well-projected balances, good affordability, risky behaviours (or absence of).
Signal brings all these details into a single score that allows you to select the most creditworthy section of the population and responsibly increase acceptance rates. This includes applicants with thin files, with adverse credit history (but who are now stable), and those new to the country.
Case Study - increasing acceptances:
Using Signal Open Banking credit score, our partner unlocked 37.8% more accepts whilst maintaining their default rate.
How does it help me reduce defaults?
Signal enables you to reduce defaults and implement lending strategies with precision. It is highly accurate across the lending spectrum and ranks effectively for all populations. The score is detailed to ensure creditworthiness is predicted precisely. For example, two individuals with similar traditional credit scores may have completely different Signal scores. Despite their similar credit histories, differences in their financial situations may mean a significant difference in their propensity to pay (proven by data), for example risky spending behaviour or affordability. Signal - by leveraging our entire insights pipeline - enables lenders to split risk more effectively, and make more accurate and informed decisions.
Case Study - reducing defaults:
Using Signal in place of their existing provider’s score, our partner found they could reduce their overall default rate from 11.7% to 9.7%, whilst increasing acceptances from 17.5% to 29.8%.
Signal was built to enable lenders to understand and evidence risk decisions. It’s the best of both worlds: high accuracy, driven by machine learning, and clear explainability.
To understand and evidence a decision with Signal, each individual’s score includes five ‘pointers’, providing feature combinations that most contributed to the person’s score.
For example, this applicant’s Signal score of 924 was most influenced by the below features:
- Liquidity Score over the next 30 days
- Overall Income Stability score for Longevity (length of time they have been receiving that income)
- Average month cash withdrawal amount over the last 3 months
- Volatility in their monthly loans repayment amounts
- Income Stability score for Amount Consistency (how consistent the income source amount is - how often it varies and by how much)
These pointers are standardised to aid quantitative analysis - meaning lenders can both evidence decisions for regulators and also analyse and understand their populations at scale. They can identify the features that are most predictive of risk for their customers, and analyse how this varies across their population. These pointers are human-readable and can also help underwriters or reviewers make sense of an individual’s score and situation.