Machine learning can help SMB banks outmaneuver upstart fintechs 🏃

Banks that serve small businesses play a critical role in the US economy, by providing a range of financial services that empower small businesses to grow, prosper and better serve their communities. Most large banks serve both consumers and businesses, however smaller banks often focus exclusively on the commercial and small business segment – making it a very important component of the overall banking system.

Small business banks are often driven by long-term relationships between bankers and their customers. As a result, there has historically been less focus on “automation” and “data science” in this segment when compared to retail banks. However, increased competition from a growing number of digital-first fintech startups is forcing commercial banks to rethink their approach and find ways to embrace innovative concepts without sacrificing their relationships.

Machine learning – the ability for computer programs to learn from historical data without needing to be explicitly programmed – is playing a leading role in this transformation. Fueled by the significant data sets that small business banks have available, machine learning can be used to train powerful predictive models that benefit both the bank (better decisions, reduced costs) and their small business customers (better experiences and less waiting).

Here are a few of the ways that small business banks are benefiting from machine learning. For a primer on machine learning generally, see this post.

1. SMB Loan Underwriting

Lending is a very important function of banks that serve small businesses because these companies often need to access funds to deal with seasonal fluctuations and working capital requirements. Whether it’s an unsecured loan, asset-backed loan or line of credit, machine learning can help banks improve their underwriting and credit assessment practices.

Credit underwriting is the process of determining which applicants to lend to and at which rates and terms. Machine learning is used to automatically analyze historical loan data and train predictive models that forecast the probability that a borrower will default on a loan. This prediction can be used to determine eligibility and to set risk-based pricing.

In their competition for borrowers with fintech lenders, banks have a major advantage: years of historical data. This data is valuable and forms the basis for effective machine learning – the more data that’s available, the better the model will be. Banks can use this data to build very accurate models, especially for assessing their existing customers, creating significant competitive advantage against upstart fintechs.

2. Bank Account Opening

One of the first things a small business owner needs to do when starting their company is open a bank account, and they’re typically looking to move quickly! This gives commercial banks an opportunity for a great first impression, and a fast and efficient process typically leads to a new customer and all the benefits that follow. However, banks typically struggle with account opening due to fraud risks and increasingly complex regulatory frameworks.

Machine learning can greatly enhance a bank’s account opening processes. Models are trained by analyzing historical data to “learn” which applications can be automatically approved and which require manual review. These predictions can be used to make decisions, or they can be used as a decision-support tool that assists bank personnel in making a final decision.

Another way that machine learning can assist with bank account opening is during document intake and analysis. Machine learning models can be trained to extract handwritten or digital text from documents (known as “optical character recognition”), which can streamline the account opening process by eliminating manual data entry and reducing errors.

3. Detecting Anomalous Activities (AML/BSA)

Machine learning models can also be used to monitor small business banking activity and scan for anomalies. This is typically done with a machine learning method called “unsupervised learning”, which is extremely effective at identifying outliers within data sets.

When anomalies are detected, a signal is sent that the customer’s activity requires review. This can help prevent fraud, money laundering and other suspicious activity, and it’s also an effective way to identify specific customers that are behaving very differently from the rest of the population. This can create opportunities for internal improvement and process optimization, and it can also identify key risk areas.

Since there is typically a lot of data available about customer accounts, the machine learning models become increasingly effective over time. Their goal is to identify every occurrence of fraud, while also limiting the number of flags that were incorrectly raised. Advances in machine learning technology, including “deep learning” algorithms, are also contributing to increasingly successful fraud prevention.

4. Product Recommendations

Commercial banks that serve small businesses typically offer a wide range of deposit and lending products. For example, different companies need different types of banks accounts, credit cards, loans, lines of credit, etc. to meet their individual needs. Machine learning programs can analyze huge volumes of data to make recommendations regarding which banking product would be best suited to each customer. These recommendations are typically based on a variety of factors, including the customer’s industry, size, transaction levels and cash flow, and the recommendations can be very beneficial in growing the relationship with a customer.


Small business banking is an industry that is defined by relationships, however banks face increasing competition and are being pushed to innovate and better serve their customers. Fueled by large historical data sets, predictive machine learning models provide opportunities for banks to create sustainable competitive advantages that grow over time.

About DigiFi

DigiFi is a technology company that helps businesses make better automated decisions.

Our platform lets businesses easily use automated machine learning and rules management to optimize critical decisions with no coding or technical expertise required. Repetitive work that used to take hours can now be completed in minutes, letting your team focus on what matters most.

Learn more at