# Improving Customer Transaction Decisions with Machine Learning
source: https://developer.mastercard.com/blog/improving-customer-transaction-decisions-with-machine-learning/index.md

Authors: Dr. Sarah Strano. Modern consumers expect instant gratification with their online transactions, and businesses competing for those customers are stuck between being too strict in catching fraud---often turning away good customers---or too permissive, allowing fraudulent transactions to occur on their platform. So how can a company verify that their customer is genuine? There are many identity verification solutions on the market, ranging from free (like social media) to static identity elements (like social security number) to dynamic identity elements (like name, phone number, and email address). The efficacy of these data sources relies not only on their accuracy---is the information up to date and correct?---but on their reliability in being truly connected to the potential customer in question, without adding too much customer friction. Who wants to give their social security number when they're ordering food delivery? So how can businesses find the balance between low friction transactions for good customers while stopping fraudsters? By using a smart solution that examines the validity of customer identity elements, the validity of the connections between the individual elements, and the patterns of their use online. This is exactly what Ekata does. In June, Mastercard completed its acquisition of Ekata, a global leader in digital identity verification solutions that provide businesses worldwide the ability to link any digital interaction to the human behind it. At Ekata, we use data-derived insights to deliver unique scores, data attributes and risk indicators that businesses then use to make more informed decisions. We help businesses around the globe identify good customers and bad actors in real-time during online account opening, payments, and other digital transactions. Ekata's identity verification data, machine learning technology and global experience, combined with Mastercard's fraud prevention and digital identity programs, will help businesses confidently know who their customers are and, in turn, help those customers safely interact online. Mastercard and Ekata's integrated services will build on both companies' commitments to ensure trust and the responsible use of data. Here's a closer look at how our technology empowers businesses with unique insights to drive better decisions Reducing Merchant Risk with Machine Learning Let's explore how our technology works with a real-life example. A merchant receives a high value transaction made by a new customer. The address for the billing and delivery are different, which raises a red flag. Do they automatically approve the potentially fraudulent transaction? Automatically deny the potential good customer? Or use human investigation to decide? Depending on the fraud and risk rules the merchant has in place, this new customer with a high value transaction could very well end up in what's called a manual review queue, where a human investigator uses data to decide the validity of the transaction. A LinkedIn search shows this customer is a recent college graduate who just started a new job in a new city. So, a large transaction could make sense---perhaps they're furnishing a new apartment. But who is paying for this bill? The recipient doesn't have much credit history, so perhaps the transaction should be denied as risky. Maybe this is a case of stolen credit card or synthetic identity? If the investigator is relying solely on free or static solutions to learn more about this transaction and the person behind it, they may not get the full picture. Ekata's Pro Insight helps connect the dots here, validating the connections between the customer-provided identity elements and their usage patterns in online transactions, providing confidence to the manual review agent that this is a valid customer and the transaction should be approved. Making a Better Decision The Ekata Identity Engine empowers businesses to make smarter decisions by combining sourced and owned data assets with machine learning capabilities to understand when to trust a customer or when a transaction looks risky. Using our solutions, businesses can feel confident that they can provide a low-friction pathway for good customers---and build that loyalty---while still stopping fraud. Behind the scenes, the Ekata Identity Engine has two parts, Identity Graph and Identity Network, that work in concert to empower the business with insights to make a decision. The Identity Graph stores and links a person's connections based on global authoritative identity elements between name, phone, email, physical and IP address. From the Identity Graph, we can look at names associated with a specific phone number and if that name matches the names provided as a part of a transaction. The stronger the links between identity elements, the lower the risk. Ekata has been sourcing global data for the Identity Graph for over 20 years and surfaces data from a broader range of sources beyond what credit organizations or free tools typically provide. The Identity Network is a collection of hashed and pseudonymized customer queries from a variety of sources that allow us to understand what behaviors are normal or should set off alarms. For instance, we can take a hashed version of a phone number and email address and look at how many times we have seen them used during a specific time range, when we first saw them together and how many other phones we have seen with that specific phone and email address together. Each signal can be useful in and of itself but when combined in a machine learning model, our Identity Network can tell the business if the behavioral patterns of the identity elements provided look like a good transaction. Ekata's award-winning global product suite includes high-scale and low-latency APIs used in transaction, monitoring, and customer onboarding, along with our Pro Insight SaaS solution for manual fraud review. These solutions empower global businesses to make smarter decisions to reduce fraud, provide good customer experiences, and build ongoing loyalty with their genuine customers.