FinTech Machine Learning Leaders

Thomas McCrohan and Leonard DeProspo
Thomas McCrohan and Leonard DeProspo
July 14, 2017



Today our colleague Abhey Lamba published a report called “Self-Driving Business on the Horizon? A Look at Enterprise AI". Out of our coverage, we believe Square has telegraphed the biggest commitment to machine learning (ML). Square applies ML to manage risk when underwriting a merchant, in marketing to target prospective clients, in support to upsell value-added services, in account management to reduce call center activity, and in its Square Capital lending unit to make better credit decisions. Visa and Mastercard are making investments in ML, particularly to curb transaction fraud.

Key Points

Square committed to, and already benefiting from, Machine Learning applications. Machine Learning (ML) has allowed Square to provide access to card acceptance to a broader base of merchants, often smaller merchants, that otherwise did not have access to the important payments rails of Visa and MasterCard. ML has also automated processes that historically required much more human intervention. For example, Square processes approximately 3 million transactions each day, and ML has allowed for fewer than 2,000 transactions to be manually reviewed. ML has also allowed Square to become very efficient in direct mailing efforts - Square was able to scale their direct mail by 3x without compromising their conversion rate, subsequently driving their total cost of new client acquisition by 50%. Square has extended the application of ML to improve sales efforts, including when it is an opportune time to up-sell an existing client new value-added services. Square’s internal models trigger staff to call or email a client when the data suggests the client would be most receptive to a new value-add service. Management views ML as so critical, that they have created a course for all of their engineers to learn how to start programming ML. The course is mandatory and is being extended to other areas of the company outside of engineering talent. Square’s objective is to approach every business challenge through the lens of ML.

Large amounts of data aid ML process. To leverage data science, large quantities of data are necessary and we note there have been 254 million unique cards used on Square during the last 12 months, there were ~1 billion transactions on Square in 2016, and the company has over 2 billion data points collected and updated daily on Square’s sellers. Square’s 10-K notes that the company’s machine learning efforts enables the company to capture and analyze over a billion transactions per year and automate risk assessment for more than 99.95% of all transactions.

View the full research report for important disclosure and analyst certification information. Ratings and/or Price Targets may change. Refer to the US Equity Research Portal library for the most recent company research.

Back to top