Credit score and credit scoring: definition, methods and challenges

In short: A credit score is a figure that summarises the risk associated with a borrower, in other words their likelihood of repaying. Credit scoring is the method that produces this figure, based on financial and behavioural data. Unlike some markets, the UK has an established infrastructure: three credit reference agencies, Experian, Equifax and TransUnion, compile credit reports and scores that lenders rely on. This data is increasingly complemented by real bank data through open banking. This shift directly supports the responsible lending expected under the Consumer Credit Act and the FCA's creditworthiness and affordability rules. Two principles then become essential: the explainability of the decision and its auditability.

‍

What is a credit score?

‍

A credit score is a numerical value that reflects a borrower's level of risk at the time of a finance application. The more favourable the score, the higher the assessed likelihood of repayment under the agreed terms. Conversely, a weaker score signals a greater risk of default.

‍

This score is not a subjective opinion. It results from a statistical calculation that compares an applicant's profile with repayment behaviour observed across a large number of cases. The aim is twofold: to help the lender decide (approve, decline, adjust terms) and to standardise how applications are handled so as to avoid arbitrary treatment.

‍

It is important to distinguish an individual's credit score from a company rating. A consumer's credit score measures their ability and willingness to repay a consumer loan, a mortgage or a Buy Now Pay Later arrangement. Rating a company follows a different logic. This article deals with credit scoring applied to onboarding and to the granting of credit to individuals and businesses.

‍

Credit scoring: how it works

‍

Credit scoring is the method that turns data into a risk figure. The general principle unfolds in several steps.

‍

First, relevant variables are collected: income, outgoings, length of employment, payment history, debt-to-income ratio, regularity of financial flows. Next, a model assigns a weight to each variable. Historically, these models were simple, transparent statistical regressions. Today, machine learning approaches can capture more subtle relationships between variables.

‍

The model then produces a score, often converted into a probability of default. The lender sets a threshold: above it, the application can be approved; below it, it is declined or referred for further analysis. This threshold reflects the institution's risk appetite and its regulatory framework.

‍

A good credit scoring system does not simply output a number. It must indicate the factors that weighed on the decision, remain stable over time and be capable of recalibration when behaviours change. This is where the quality of input data becomes decisive.

‍

What data: declared information vs open banking

‍

In the UK, credit scoring has traditionally drawn on two main sources. The first is the data held by credit reference agencies, Experian, Equifax and TransUnion, which build credit reports from a person's borrowing history, repayments, defaults, County Court Judgments and details from the electoral roll. The second is declared information: the details supplied by the applicant (stated income, outgoings, employment status), supported by documents such as payslips or bank statements. These approaches have limits. Credit reference data is a backward-looking record of past behaviour, and declared information relies on documents that can be out of date or falsified, offering a snapshot rather than a dynamic view of a financial situation.

‍

Open banking changes the picture. Framed by the second Payment Services Directive (PSD2) and now overseen in the UK by the Financial Conduct Authority and the Open Banking framework, it allows read access to a person's bank account data through secure interfaces, with their explicit consent. The lender then observes real flows: income actually received, regularity of incoming money, recurring outgoings, any incidents, capacity to save. This gives a faithful and up-to-date picture, far harder to manipulate than a scanned document, and it complements rather than replaces credit reference agency data.

‍

This shift towards real data strengthens the relevance of creditworthiness analysis. It also reduces the risk of document fraud, since the information comes directly from the banking source. A good practice is to favour collecting data at source, that is, obtaining information from a trusted third party rather than relying on declared information alone. Declared information does not disappear: it remains useful for putting a situation in context, but it is now cross-checked against bank data to make the score more reliable.

‍

Credit score and responsible lending (FCA and the Consumer Credit Act)

‍

Credit scoring is not just a commercial tool. It sits within an increasingly demanding regulatory framework. In the UK, consumer credit is governed by the Consumer Credit Act and by the FCA, whose rules in the Consumer Credit sourcebook (CONC) require a creditworthiness and affordability assessment before credit is granted. The FCA's Consumer Duty reinforces the obligation to act in the customer's interest and to avoid foreseeable harm.

‍

In practice, the lender must check that the borrower can reasonably afford to repay, without being pushed into difficulty. This assessment must rely on relevant, accurate and sufficient information about income and expenditure. The rules also cover newer products such as Buy Now Pay Later, with the UK moving to bring this sector within the FCA's remit. For reference, the equivalent EU framework is the second Consumer Credit Directive (Directive (EU) 2023/2225, known as CCD2), which does not apply in the UK but pursues similar aims of stronger creditworthiness assessment.

‍

In this context, a credit score built on real bank data becomes a compliance asset. It makes it possible to demonstrate that the lending decision rests on a genuine analysis of repayment capacity, rather than on an unverified declaration. The FCA also stresses clear communication with the consumer and the ability to explain outcomes, particularly where a decision is based on automated processing. Scoring must therefore be designed to be explained, not only to decide.

‍

Limits and explainability (AI, bias, UK GDPR)

‍

A credit score remains a probabilistic estimate, never a certainty. Three points call for vigilance.

‍

The first concerns the quality and representativeness of the data. A model trained on biased histories can reproduce, or even amplify, discrimination. Responsible scoring therefore requires regular review of the variables used, the exclusion of prohibited criteria and monitoring of differences in treatment between population groups.

‍

The second concerns explainability. The most powerful artificial intelligence models can become opaque, to the point where even their designers struggle to justify an individual decision. Yet UK GDPR governs solely automated decisions producing significant legal effects, such as a credit refusal. It gives the individual the right to obtain human intervention, to express their point of view and to challenge the decision, and the Information Commissioner's Office (ICO) provides guidance on how this must be applied. Explainable AI, able to set out the determining factors behind a score, is therefore not a luxury: it is a condition of compliance and of trust.

‍

The third concerns auditability. Every decision must be capable of being reconstructed after the event: which data was used, which model, which threshold, which result. This audit trail protects both the consumer, who can challenge, and the institution, which must prove the compliance of its practices to the regulator.

‍

This is precisely Meelo's approach. The platform does not issue a universal bureau-style credit score. It performs an explainable creditworthiness analysis on real data drawn from open banking, with a decision delivered in 2 to 5 seconds, documented and auditable. Each assessment comes with the factors that drove it, which allows the lender to justify its decision and to meet its responsible lending obligations.

‍

In conclusion

‍

A credit score summarises a borrower's risk, and credit scoring is the method that produces it. In the UK, lenders draw on credit reference agency data, increasingly combined with real bank data made accessible through open banking. This evolution improves accuracy, reduces document fraud and aligns with the responsible lending expectations of the FCA and the Consumer Credit Act. But performance is not enough: a credit score must be explainable, non-discriminatory and auditable to be compliant and trustworthy. Only on that condition does scoring become a genuine tool for responsible decision-making, serving lender and borrower alike.

‍

FAQ

‍

Is there a credit score in the UK?

‍

Yes. The UK has three credit reference agencies, Experian, Equifax and TransUnion, which compile credit reports and produce credit scores from a person's borrowing and repayment history. There is no single universal number, however: each agency uses its own scale, and each lender builds its own assessment from credit reference data, declared information and, increasingly, real bank data obtained via open banking with the person's consent. So while credit scores exist, the decision to lend rests on each lender's own analysis rather than on one shared score.

‍

How is a credit score calculated?

‍

A credit score is calculated from financial and behavioural variables: income, outgoings, debt-to-income ratio, regularity of flows, payment history, along with credit reference data such as past repayments and defaults. A statistical or machine learning model assigns a weight to each variable and produces a score, often expressed as a probability of default. The lender then compares this result against a decision threshold that reflects its risk appetite and its regulatory framework.

‍

What do open banking data add to credit scoring?

‍

Open banking allows access, with the person's consent, to their real bank data through secure interfaces framed by PSD2 and the UK Open Banking framework. The score no longer rests only on credit reference records and declared documents, but on the flows actually observed: income received, recurring outgoings, capacity to save, any incidents. The benefit is twofold: a more faithful and up-to-date picture of creditworthiness, and a strong reduction in the risk of document fraud.

‍

Must a credit score be explainable?

‍

Yes. UK GDPR governs solely automated decisions with significant effects, such as a credit refusal, and provides a right to human intervention and to challenge the outcome, with guidance from the ICO. The FCA also expects lenders to communicate clearly and to explain decisions where credit is declined. A credit score must therefore rest on explainable AI, able to set out the determining factors, and come with an audit trail that allows each decision to be reconstructed.

‍

Sources:

‍

• UK credit reference agencies (Experian, Equifax, TransUnion), credit reports and credit scoring.
• Financial Conduct Authority (FCA), Consumer Credit sourcebook (CONC) creditworthiness and affordability rules, and Consumer Duty.
• Consumer Credit Act, statutory framework for consumer credit in the UK.
• UK GDPR and the Information Commissioner's Office (ICO), automated decision-making and the right to an explanation.
• Directive (EU) 2023/2225 (CCD2), EU equivalent on consumer credit and creditworthiness assessment, provided for reference.