Blispay Improves Fraud Model Performance with Ekata APIs

Success Story

Ekata Saves Data Scientists Time

Data scientists are very, very busy. They often spend much of their time cleaning the data for their models, and not much time doing the core tasks expected of a data scientist. In fact, data preprocessing and feature engineering typically take about 60-80% of the time it takes to build a model. Fortunately, collecting and preparing model data is part of our process. Data scientists do not have to spend the bulk of their testing time cleaning Ekata data- the data has already been preprocessed.

We want data scientists to understand our data so that they can spend the bulk of their testing time and budget testing the model with our data. So, we created a machine learning guide to help data scientists use our identity data.

We also help many of our customers with their models. We help them by running exploratory analysis including correlations and information values. We offer advice on feature engineering to make sure that our customers get the greatest possible benefit from our data. Feature engineering is one of the main avenues of support that Ekata provides to customers. We provide insight that we have garnered through our model building process. This insight helps our customers decrease the time spent on and cost of data preprocessing and feature engineering. When it comes to feature engineering, we’ve already done the heavy lifting. We also provide guidance to customers regarding experiment selection and frameworks to test our data in models.

Why Test Fraud Models with Ekata?

One of the most significant reasons data scientists should test fraud models with our data is consistency. Ekata data is designed specifically for risk, so our customers do not have to worry that the features will someday change. No other identity data product has as many features as Ekata, and those features are consistent globally. Also, with our data, you do not have to integrate with numerous disparate sources to get new data points into your model.

Ekata delivers more than 50 market proven features that are instrumental in identifying fraudulent transactions in industries ranging from eCommerce to marketplace onboarding. Among these market proven features are address validation, phone carrier checks, email and IP cross checks. These features are easily available through a RESTful JSON API that makes ingestion of the data as easy as possible. And our data is consistent.

Take our Global Address Validation API for example. Quite a few address validation APIs are available today, but few of these APIs are built for risk. The vast majority of these address validation APIs are built for any use case under the sun. These APIs were not built with a specific model in mind. So, the companies who provide these APIs can change the outputs- and change them at any time. What happens if you are using an address validation API and the model has features A B C, but then the company changes those features to D E F? – What happens is your entire model blows up because the API model’s features were changed, and then you must retrain your model.

Because Ekata APIs are designed with a specific use case in mind- risk, you do not have to worry that the features will change. The data provided by our APIs is always consistent, and data consistency is incredibly valuable when it comes to machine learning models.

Ekata Identity Data APIs

Each Ekata API is designed to address different types of fraud. For this article, we are focusing on the Global Address Validation API, Global Phone Intelligence API, and the Global Identity Check API.

Global Address Validation API

The Ekata Global Address Validation API can be used to parse, normalize, validate, and geocode addresses from any country in the world. The API normalizes and verifies addresses in 170+ countries and gets a unique identifier (UUID) for every address. This API only speaks to the validity of the given address and tells you where the location is. The Address Validation API is often used to calculate address velocities that can be used in a fraud ML model.

Address validation is especially beneficial for merchants as they must know that an address is valid before shipping a product. Using our Address Validation API, merchants can validate the shipping address up front before getting payment authorization for an order. Merchants can prevent the system from accepting an order that can be shipped because of an invalid address.

Global Phone Intelligence API

The Ekata Global Phone Intelligence API can be used to validate a number and verify that the number is assigned to a carrier. The API identifies more than 5,000 carriers and provides lightweight fraud verification when only a phone number is provided.
The Phone Intelligence API is often used to detect risk during progressive signup flows using line type. Example use cases for this API are promo abuse fraud and detecting automated signups by bots that compromise platform integrity.

Global Identity Check API

The Ekata Global Identity Check API is a sophisticated and comprehensive API returning 70+ data signals and network insights, from a single query. The API provides businesses a clear picture of consumers based on real-time global data, network insights, and machine learning across the five core data attributes of email, phone, person, physical address, and IP. This API also features Confidence Score, a feature that provides a real-time, predictive risk score ranging from 0 to 500. Typically, the higher the score, the riskier the transaction.

The Global Identity Check API is an umbrella product, and it is used if you need an identity to assess the fraud risk. This API is a great choice for businesses that want to improve the performance of a fraud detection ML model. It can also be used to improve the rules of a rules-based fraud prevention system. Out of all our APIs, the Global Identity Check API is chosen most often for fraud use in machine learning (ML) models because it provides the best breadth and depth of data to impact your model positively.

Blispay Uses the Identity Check API

Blispay is one of the many Ekata customers that uses the Identity Check API to improve the performance of their fraud model. Blispay provides instant access to digital Visa cards. Consumers in the US can apply for a digital Visa card and get access to that card in a matter of minutes. And these digital cards can be used anywhere Visa is accepted.

An instant digital Visa card is a convenient way for consumers to make smaller purchases where they would not get financing, like a $1,000 bike for example. Consumers can use the digital Visa to finance a small purchase like a bike instead. While this instant Visa technology is great for consumers, it also opens the door for fraudsters. One of the ways the Identity Check API helps Blispay fight fraudsters is by providing breadth and depth of identity data for their model to learn from. For example, Ekata Identity Check API can identifying over five thousand global phone carriers. This is useful data because certain carriers and protocols are often riskier than others and can be indicators of potential fraud.

At Blispay we’re committed to a best in class mobile application process, and Ekata is a critical part of the fraud model that helps make that possible.

- Jared Young | Head of Risk and Analytics | Blispay

Ideal for Tree-Based Models

For companies building models for fraud prevention, our APIs are extremely fast and ideal for tree-based models. And for companies that have strict response time requirements, we can refine our APIs to suit your needs. Need help deciding which API is right for your fraud model? Our team of machine learning Solutions Architects are happy to help – contact them today.

Download this page as a .pdf


Learn more

One of our experts will be happy to give you a free guided demo of Ekata Identity Check API.