AWS Public Sector Blog

Fighting fraud and improper payments in real-time at the scale of federal expenditures

Since 2003, the US federal government has made approximately $1.7 trillion in improper payments, with an estimated $206 billion made in FY 2020 alone. Improper payments are now anticipated to increase proportionally to new levels of federal spending, from the $1 trillion infrastructure bill, to the anticipated $3.5 trillion budget reconciliation plan. How can agencies fight improper payments at this scale? And what tools can agencies use to address fraud, erroneous data submission and other causes of this problem? Agencies can use Amazon Web Services (AWS) to solve the multi-sided issues of payment integrity.

Figure 1. Agencies face multiple challenges in eliminating improper payments. AWS can help agencies with delivering federal resources at unprecedented scale, updating technologies for a seamless citizen experience, keeping pace with sophisticated fraudulent activity, maintaining complete and accurate data to validate identities, and taking real-time action to eliminate pay and chase.

Figure 1. AWS can help agencies facing multiple challenges in eliminating improper payments.

Balancing payment integrity with the citizen experience

Let’s explore an improper payment scenario caused by fraud. Your agency distributes federal funds to citizens. A fraudster submits an application based on falsified records. After entering the agency’s queue, their application awaits validation using standard “Know Your Customer” (KYC) approaches, such as an algorithm that compares the applicant’s identity against those of fraudsters who have previously applied. Unknown to your agency, however, is that this fraudster has defrauded multiple other agencies under aliases and exhibited characteristics of fraud not evaluated by your algorithms. Because the agency lacks the complete picture of the fraudster’s behavior and relationship, their application passes through the system without a flag.

Here, if you delay the payment to manually investigate the application, you may delay an existing backlog of applications or falsely flag a genuine application for fraud. These “false positives” are a major cause of customer complaints. On the other hand, if you issue the payment and later investigate the application for fraud, you may find yourself in a “pay and chase” scenario, increasing administrative cost burden and decreasing the likelihood of recovery.

How does AWS help agencies eliminate improper payments?

This scenario illustrates several of the root causes behind improper payments. Without a way to look at a broader set of risks, incorporate more data sources, and move to real-time or near real-time analysis, agencies struggle to deliver frictionless customer transactions, prevent errors and fraudulent activity, and reduce administrative burdens while doing so.

The AWS approach to eliminating improper payments draws on 20 years of experience preventing fraud and transforming the customer experience on On Amazon’s website and mobile application, customers make hundreds of purchases per second, each of which must be screened for over 2,000 potential elements of fraud and deliver a superior customer experience. This experience taught us that data scientists, engineers, and investigators all need access to a robust set of capabilities to address the multiple root causes of improper payments.

On the AWS Cloud, teams in your agency can access a shared set of fraud and error prevention capabilities through what AWS calls a “real-time orchestration platform.” Illustrated in Figure 2, real-time orchestration provides a centralized location in which data science, engineering, and investigative teams can use “orchestrators” to address suspect activity.

Figure 2. Real-time orchestration on AWS empowers agencies to improve the citizen experience.

Figure 2. Real-time orchestration on AWS empowers agencies to improve the citizen experience.

By using these orchestrators, agencies can view, share, analyze, and act on data in real-time, supporting them to achieve fraud prevention and citizen experience objectives. Key orchestrators that enable these outcomes include:

  1. Master data management (MDM), virtualization, and encrypted data access: On AWS, you have access to data storage, cataloging, and governance tools to implement MDM and create centralized access to your data through data virtualization or ingestion. Tools such as Amazon Comprehend allow you to detect and obfuscate PII data to meet statutory requirements, while MDM helps you make sure that data in payment integrity systems are uniform, authoritative, and consistent.
  2. Flexible data ingestion, enrichment, and sharing: You can bring data onto AWS from diverse sources, including partner organizations, data consortiums, and AWS Data Exchange. AWS provides flexibility to gather the types of data you need to develop customer profiles of both fraudulent and genuine customer behavior. You can also implement data fabrics to improve data access, quality and organization through enrichment, automated metadata tagging, unstructured data management, and data catalogs.
  3. Real-time artificial intelligence (AI) and analyses: The platform empowers you to verify and validate users, detect anomalies and threats, and determine fraud and risk scores in real-time through tools like Amazon Fraud Detector and Amazon SageMaker. For example, Amazon SageMaker can support an end-to-end solution for fraud detection of claims. You can also use Amazon Neptune, a graph database, to create new insights into customer profiles through relationships between known and new fraudsters.
  4. Feedback loops for continuously improving fraud and error detection models: The platform allows you to establish real-time feedback loops in your models, helping you continuously improve fraud/error prediction, reduce false positives, and automate operations. On AWS, you can implement feedback loops and operationalize microservices using Amazon SageMaker within your payment integrity systems. You can also support investigator intervention of low confidence machine learning predictions using Amazon Augmented AI (Amazon A2I). This technique allowed DealNet Capital to process their financial forms and reduce the amount of time spent manually reviewing documents by up to 80%.
  5. Capabilities for improving investigator and citizen experiences: The platform integrates with case management tools, such as ServiceNow; multi-channel communication services, such as AWS Pinpoint; and intelligent chatbot services like Amazon Lex. These integrations help you to reduce investigator response time and administrative burden while improving the user experience. You can also integrate with AI, natural language processing (NLP), and image analysis services, such as Amazon Recognition, to adopt digital onboarding, KYC approaches, identification, and verification.

Customers find success using AWS to fight fraud and improper payments

Innovators in both public and private sectors are adopting real-time orchestration approaches built on the AWS Cloud. These approaches have helped the following customers reduce administrative burden, fight fraud, and move toward a frictionless customer experience.’s Buyer Fraud Service uses the cloud to collect more than 2,000 real-time and historical data points for each order and screens thousands of transactions per second for possible fraud. Real-time tools help prevent millions of dollars in fraudulent transactions a year.

The Maryland Department of Human Services (DHS) developed a first-in-the-nation human services platform using AWS. This platform now supports multiple state agencies to share health data across organizations, verify data integrity, and reduce the risk for abuse of state service.

Capital One’s fraud detection system automatically alerts customers to fraud. The system walks customers through fraud reporting steps, helps them lock their card and order a new one, and then unlocks a temporary card so there is no interruption in their ability to spend their money.

“With the AWS cloud… we can optimize on that sweet spot of offering sufficient protection, but not overdo it with too many false positives.”

– Nitzan Mekel-Bobrov, Ph.D., managing vice president of machine learning at Capital One

The Financial Industry Regulatory Authority (FINRA) uses the AWS Cloud to scale to meet both the volatility and size of the US financial markets. FINRA’s big data platform now holds more than 30 petabytes of data, and analyzes 135 billion market events per day for fraudulent activity.

Learn more  

A real-time orchestration platform provides access to capabilities that empower agencies to address the multiple root causes of improper payments. For citizens awaiting federal assistance, these capabilities mean less time waiting, less false positives, and a better customer experience.

To learn more about how AWS can help your agency with improper payment, reach out to your AWS account team to engage with us on a proof of concept of capabilities in the real-time orchestration platform, or contact us for more information.

Subscribe to the AWS Public Sector Blog newsletter to get the latest in AWS tools, solutions, and innovations from the public sector delivered to your inbox, or contact us.

The AWS Public Sector Blog needs your help. Please take a few minutes to share insights regarding your experience with the AWS Public Sector Blog in this survey, and we’ll use feedback from the survey to create more content aligned with the preferences of our readers.