Ant Group’s credit scoring system, Zhima Credit, evaluates over 3,000 variables per user to determine creditworthiness. The system scores more than a billion people in China, most of whom have no traditional credit history with a credit bureau. Before Zhima Credit, these individuals could not access formal lending products. After its deployment, Ant Group’s consumer lending platform, Huabei, extended credit to hundreds of millions of first-time borrowers. The system works because it replaces the traditional approach to credit (reviewing a borrower’s history of past borrowing) with a data-driven approach (analysing thousands of behavioural signals to predict future repayment). That shift, from history-based to data-driven decision-making, is rewriting the rules of finance globally.
The financial services industry is moving toward data-driven operations at a measurable pace. According to MarketsandMarkets, the global AI in finance market, which depends entirely on data infrastructure, reached $38.36 billion in 2024 and is projected to grow to $190.33 billion by 2030 at a 30.6% CAGR. The institutions and fintech companies that have built the strongest data capabilities are pulling ahead of competitors in every product category. The gap is widening.
What Data-Driven Finance Actually Means
Data-driven finance is not simply “using data.” Every financial institution has used data since the invention of the ledger. The distinction is in how data is used: as an input to human judgment or as the primary basis for automated decisions.
The Boston Consulting Group projects fintech revenues will reach $1.5 trillion by 2030, with embedded finance and digital lending accounting for the largest share of projected growth.
According to CB Insights’ 2024 fintech report, global fintech funding declined 40 percent between 2022 and 2024, pushing the sector toward consolidation and a sharper focus on profitability over growth at all costs.
In traditional finance, data supports human decision-makers. A credit analyst reviews a borrower’s financial statements, considers the data, applies judgment, and makes a lending decision. A portfolio manager reviews market data, reads research reports, applies experience, and makes an investment decision. The human is the decision engine. Data is an input.
In data-driven finance, statistical models and algorithms make the decisions. A machine learning credit model evaluates thousands of data points and produces a lending decision. An algorithmic trading system analyses market data and executes trades. A fraud detection model evaluates transaction patterns and blocks suspicious activity. The model is the decision engine. Human judgment is reserved for cases the model cannot handle and for overseeing the model’s performance.
This distinction matters because data-driven decisions have three properties that human decisions do not: consistency (the model applies the same criteria to every case), scalability (the model can process millions of decisions per day), and improvability (the model can be retrained on new data to become more accurate over time). These properties give data-driven financial institutions structural advantages that compound with scale.
The Data Advantage in Lending
Lending is where data-driven finance has had its clearest impact. Traditional lending evaluates borrowers using a small number of variables: credit score, debt-to-income ratio, employment status, collateral value. These variables are available for most borrowers in developed markets and have proven useful over decades. But they are limited.
Data-driven lenders evaluate hundreds or thousands of variables. Upstart’s model uses over 1,500 variables including education, employment history, income trajectory, and bank transaction patterns. The model approves 27% more borrowers at the same default rate as traditional models because it identifies creditworthy borrowers that a FICO score alone would reject. A borrower with a thin credit file but a strong income trajectory and consistent savings behaviour is low-risk, but a traditional model cannot see the income trajectory or savings behaviour because those variables are not in the credit report.
Square’s lending product illustrates a different data advantage. Square Loans uses the transaction data flowing through each merchant’s point-of-sale terminal to assess lending risk. The system sees daily revenue, transaction frequency, customer return rates, and seasonal patterns in real time. This data is more current and more granular than the quarterly financial statements a traditional bank would require. Square has originated over $17 billion in loans to small businesses using this approach.
In emerging markets, data-driven lending is opening access to credit for populations that traditional finance has never served. Tala and Branch in Africa and South Asia use mobile phone data (call patterns, app usage, device characteristics) to evaluate borrowers with no formal credit history. The data is unconventional, but the statistical relationships between mobile behaviour and repayment probability are strong enough to support profitable lending at scale. These companies could not exist without a data-driven approach to creditworthiness.
Data-Driven Risk Management
Grand View Research reports that risk management held 27.9% of the generative AI in financial services market in 2024, the largest single application category. The concentration of investment in risk reflects a basic reality: financial institutions that manage risk more accurately earn more money and survive crises that destroy less capable competitors.
Traditional risk management relies on models built from historical data and stress-tested against a limited number of predefined scenarios (recession, interest rate shock, currency crisis). Data-driven risk management uses machine learning to process real-time data from multiple sources and identify risks that predefined scenarios would not capture.
BlackRock’s Aladdin platform manages risk analytics for over $21 trillion in assets by processing market data, portfolio positions, and risk metrics continuously. The system simulates millions of market scenarios, far more than any human team could evaluate, to identify portfolio exposures that might be invisible when looking at individual positions in isolation.
After Silicon Valley Bank’s collapse in March 2023, several institutions accelerated data-driven approaches to liquidity risk. Traditional liquidity monitoring tracked aggregate deposit levels on a daily or weekly basis. Data-driven systems now monitor individual account-level deposit flows in real time, combined with social media sentiment analysis and peer institution comparison data. The goal is to detect early signals of deposit instability before it becomes a crisis. SVB’s failure demonstrated what happens when risk monitoring operates on yesterday’s data in a world where depositors can move money in minutes.
The Infrastructure Layer
Data-driven finance requires infrastructure that most traditional financial institutions were not built to support. Three infrastructure components determine whether a financial institution can operate in a data-driven manner.
Data integration. Financial institutions typically operate dozens of separate systems: core banking, card processing, loan origination, CRM, compliance, treasury management. Each system generates data in different formats with different update frequencies. Data-driven decision-making requires integrating data from these systems into a unified platform where models can access all relevant information. Plaid built a business around this integration challenge for consumer data, normalising bank transaction data from thousands of financial institutions into a consistent format. MX provides similar data integration services for banks and credit unions.
Real-time processing. Many data-driven financial decisions must happen in milliseconds. Fraud detection cannot wait for an overnight batch process. Credit decisions that take seconds rather than days require real-time model inference. Personalised product recommendations must reflect the customer’s most recent behaviour. The infrastructure to support real-time data processing at the scale of a major financial institution is expensive and technically complex. Fintech companies built on modern cloud architecture have an inherent advantage because their systems were designed for real-time processing from the start.
Data governance. Financial data is among the most regulated categories of personal information. The EU’s GDPR, California’s CCPA, and similar regulations in other jurisdictions impose strict requirements on how financial data is collected, stored, used, and shared. Data-driven financial institutions must maintain governance frameworks that ensure compliance while allowing their models to access the data they need. Getting this balance wrong in either direction is costly: too restrictive, and models lack the data to be accurate; too permissive, and the institution faces regulatory penalties and customer trust erosion.
The Competitive Divide
The financial services industry is splitting into two groups: institutions that have achieved data-driven operations and those that have not. The divide is visible in their economics.
Nubank, with its data-driven operations, serves 100 million customers at a cost-to-serve roughly one-fifth of traditional Brazilian banks. Revolut processes 500 million transactions per month across 38 markets using automated, data-driven operations. Stripe serves millions of merchants globally, using data from its entire network to improve fraud detection, pricing, and product recommendations for each individual merchant.
Traditional institutions are investing heavily to close the gap. JPMorgan Chase employs over 2,000 data scientists. Goldman Sachs has integrated data-driven models across its trading and wealth management operations. But these investments take time to produce results because integrating data-driven systems with legacy infrastructure is a multi-year undertaking.
For fintech companies, the data-driven advantage is existential. They compete against institutions with far more capital and far larger customer bases. Their ability to win customers depends on offering better products, faster service, and lower prices, all of which require data-driven operational efficiency. A fintech company that cannot build a data infrastructure capable of supporting real-time, personalised, AI-driven financial products will not survive against competitors that can.
The future of finance is not a prediction. It is already visible in the performance gap between data-driven institutions and traditional ones. The institutions on the right side of that gap, whether digital-native fintechs or modernised incumbents, are building compounding advantages that late movers will find increasingly difficult to close. Data is not becoming important to finance. It has already become the foundation on which every competitive financial product is built.
