Across emerging markets in Asia, the Middle East and Africa, many remain outside formal credit systems. Without credit cards, salary slips or bank loans, they leave no trace in conventional data ecosystems. This exclusion limits access to financial products and stifles both personal advancement and broader economic growth. Yet, these same individuals generate extensive digital footprints that can inform credit decisions, provided institutions update their interpretation of trust.
Christo Georgiev, chief executive officer and founder of LenderLink, said the problem stems from how institutions utilise data, not from its scarcity. “We need to move from data to information, then to intelligence,” he explained. Traditional systems record data effectively without applying intelligence to understand behaviour or intent. The shift moves from exclusion by absence to inclusion by presence. Consumers once invisible to lenders are now visible through digital and behavioural activity. Creditworthiness extends beyond prior loans to encompass how individuals consistently interact with digital and economic systems.
Georgiev pointed out that the current infrastructure fails to reflect how people in developing markets operate economically. “Traditional data alone does not sufficiently explain the behaviours of consumers in developing markets,” he observed. Institutions must stop assuming that a lack of formal data means financial invisibility. Limited consumer awareness of credit systems widens this gap. Many people do not understand how their scores affect access to loans, jobs, or rentals. This can weaken their ability to build a strong financial reputation, especially in areas where formal borrowing is rare.
Outdated models create a structural blind spot
Most banks still assess credit using internal records and bureau data. These models, built for developed economies, rely on prior borrowing. In countries like the Philippines, where under five per cent of adults have credit cards and fewer than 15% have borrowed from a bank, this creates a structural blind spot. “Banks end up having to make risky decisions with limited information, or more often, they choose not to lend at all,” Georgiev noted. Institutional rigidity, rooted in a fixed mindset, drives the issue, preventing the redefinition of trust for a digital-first economy.
Public and private credit bureaus struggle to build reliable profiles in these settings. The result is exclusion and disproportionately high borrowing costs. As conventional data falls short, institutions must adopt new ways to assess risk, building trust through behavioural evidence.
Building creditworthiness from alternative signals
The digital footprints people leave behind create alternative data. This data encompasses behavioural signals such as mobile usage, digital transactions, and identity verification, which reflects how people manage resources and interact with technology daily.
Mobile behaviour provides highly predictive insights. Subscriber identity module (SIM) card age, top-up frequency, app installs and charging habits all indicate financial consistency. “Someone topping up their phone with $10 every two weeks behaves very differently from someone doing one-dollar top-ups daily,” Georgiev said. These patterns help differentiate applicants who seem identical under traditional scoring.
Previously, Georgiev built a telco-based credit model in the Philippines, using mobile metadata and top-up history. This allowed lenders to identify reliable borrowers that traditional tools missed. Open finance data, such as digital wallet and bank account activity, provides additional visibility. With consent, this shows income flows and spending patterns, even without formal credit.
Digital identity helps with verification. A phone number or email linked across platforms supports continuity and stability. “In today’s world, someone using their primary contact details will almost always appear on at least one social platform,” Georgiev remarked. Psychometric tools, including artificial intelligence (AI)-enabled surveys and voice or video inputs, reveal traits linked to financial reliability. “A wealth of third-party data exists that can supplement and improve decisions and focus on the consumer’s personality,” he added.
These signals form multidimensional profiles that reflect liquidity, intent, and behavioural stability. Someone with consistent e-wallet inflows, shared SIM credentials and strong psychometric scores may prove more creditworthy than bureau data suggests.
From static rules to adaptive intelligence
Such signals need more than rigid checklists. Traditional banks often use frameworks that reject applicants based on single variables, such as age or income. These models are often easy to audit but simultaneously miss the complexity of real-life behaviour.
Machine learning enables a different model. These systems process thousands of variables to generate probability-based risk scores. “This involves moving from static checklists to adaptive intelligence,” Georgiev said. He implemented such a system at a Philippine neobank. Using alternative data and machine learning, the bank reduced default rates by 10 to 15 percentage points, cut costs and expanded the eligible borrower pool. The system cleaned raw data, engineered inputs and trained models on real-world patterns.
Ensuring AI accountability and explainability
As machine learning becomes more powerful, it also becomes harder to interpret. Lenders may not understand why a model rejected an applicant, and consumers may struggle to contest the decision. Advanced systems, including large language models, add complexity by handling unstructured inputs and opaque weightings.
“It is essential that in the triangle of scalability, explainability and predictive power, all three facets are equally addressed,” said Georgiev. High-performing models that lack clarity risk violating regulations and losing trust. Many regulators now require clear reasons when denying credit. Georgiev said models must meet fintech demands for speed, regulatory needs for transparency and banks’ need to manage long-term risk. Building systems that address all three enables sustainable use.
Rewiring inclusion as a mindset shift
Modern credit systems no longer penalise a lack of formal history. They reward digital presence, behaviour and consistency. When interpreted through intelligent systems, alternative data enables access for those previously excluded.
Georgiev said the institutional mindset is still the most significant barrier. “The biggest obstacle is not data or regulation. It is a mindset,” he emphasised. Institutions that cling to narrow views of creditworthiness will fall behind as the market evolves. Inclusion now depends on systems that learn, adapt and explain. Success lies in using the signals consumers already generate to build trust. As digital banks and fintech firms scale these tools, traditional banks must adapt or lose relevance.
Bureau scores and credit cards will no longer define access to credit. Identity, behaviour and intent will take precedence. This shift is not incremental. It represents a structural transformation in financial infrastructure. Institutions that invest in complexity and transparency will shape the next era of inclusion. Those that fail to evolve will be left behind. Where absence once defined exclusion, presence now defines possibility. Rebuilding trust requires seeing people as their present actions show them to be.
The Future Banking Working Group (FBWG) invites you to our upcoming session, “Elements of Consumer Credit Risk Analysis,” a strategic and tactical workshop exploring how financial institutions in developing markets can apply practical approaches to credit risk analysis, improve model transparency, and leverage alternative data to assess creditworthiness more effectively.
In this interactive session, we’ll discuss:
- How do rule-based and statistical credit models differ, and why do many lenders in developing markets still rely on scorecards rather than modelling?
- What are the key steps in developing a credit scoring system, including data collection, cleaning, transformation, and feature engineering?
- How can financial institutions utilise alternative data sources, such as telco, device, psychometric, and open finance data, to score underbanked consumers?
- What are the main challenges in using alternative data, including concerns around privacy, standardisation, and compliance?
- How are AI and large language models making credit scoring faster, more scalable and more accessible, and what does this mean for traditional banks?
Agenda (SGT)
- 4:00 PM – 4:10 PM: Introduction of Elements of Consumer Credit Risk Analysis workshop by Urs Bolt, Chairman, The Banking Academy
- 4:10 PM – 4:40 PM: Presentation by Christo Georgiev, CEO and Founder, LenderLink
- 4:40 PM – 5:00 PM: Q&A
Google 
LinkedIn
