Is a manual process hurting your bottom line?
An independent hydrocarbon exploration company engaged in the acquisition, development, and exploration of oil and natural gas properties had identified the need to transition a manual process for detecting anomalies to a Machine Learning solution and integrate several previously siloed data streams.
Security analysts have estimated an industry over-billing rate of between 10 and 20 percent in service sectors where this organization spends about $500 million annually. The largest segment of that annual spend, have an estimated over-billing/fraud rate of 20 to 25 percent. While the company’s financial security team was able to manually audit 4 percent of that segment, it hoped to increase audit coverage to 100 percent through automation.
A serverless approach for tracking anomalies
Working in partnership with the client, Pariveda developed a solution that examines 100 percent of electronic tickets and invoices, analyzes the data, and runs defined and industry-specific rules to detect anomalies across the vendors. The team also developed a case management system to allow the client to prioritize, investigate, and track the anomalies in one centralized location. A client-facing product manager worked with the development team to refine the anomaly detection rules and drive greater accuracy in the results of analyses for each use case.
Pariveda took a serverless approach to reduce the client’s maintenance workload as much as possible. Using Lambda, API gateway, S3, SQS, and step functions allowed Pariveda to concentrate efforts on the business logic and saved the client from worrying about the infrastructure in the future. Going serverless also kept costs in check. The solution isn’t used on a continual basis throughout the day, and the client pays only for what they use.
Automation yields deeper insight, reduces costs and boosts productivity
Pariveda delivered in Agile fashion an ML solution that will have paid for itself based on the amount the client expects to recover from flagged tickets and invoices. This solution effectively automated the process of auditing and detecting anomalies, freed up the client’s security team to concentrate their efforts on recovering money and preventing waste, and laid the foundation for additional data sources and analysis.
More than $5 million in anomalous work tickets and invoices have been identified to date. The company is now able to identify vendors with problematic work tickets and invoices and manage the resolution of those anomalies through the case management component of this solution. Additionally, the client is now aware of anomalies with vendors not historically suspected of being problematic, so there are opportunities to correct vendor processes to improve work ticket and invoice accuracy and volume, which will reduce its invoice processing burden over time. This capability is being delivered on the AWS platform at a daily cost of $13 thanks to the serverless nature of the solution.