The unfortunate truth is that the core characteristics of the manufacturing industry – unmonitored supply chains, a host of third parties, underlying assets in the form of inventory, and multiple and frequent transactions – make it a sweet spot for procurement and inventory fraud schemes. Common vulnerabilities and fraud risks in the manufacturing sector include product counterfeiting, bid rigging, conflict of interest, warranty claims fraud, IP infringement, theft or misuse of inventory, and more. And it’s taking a financial toll. The Association of Certified Fraud Examiners (AFCE) estimates fraud costs $200,000 per incident.
Fraud is impossible to eliminate entirely, but there are common ways to reduce its probability and more quickly identify red flags:
- Assess and actively monitor internal controls. Existing controls, thresholds and procedures should be regularly reviewed and assessed for relevance, adequacy, and effectiveness.
- Develop a robust, well-communicated fraud response plan.
- Know your supplier. Performing background checks and integrity due diligence can ensure that the manufacturers or suppliers are of reputable standing, and it can highlight the manufacturers or suppliers’ interests, associations, related parties, and possible conflicts of interest.
- Conduct regular checks on quality such as routine checks for non-deliveries, repeat deliveries for the same order, and discrepancies between purchase orders and delivery.
- Optimize the power of data. Big data is not only useful to provide insights, but organizations can extract some real value in the form of opportunities they were not previously aware of.
Putting Data to Work Against Fraud
Utilizing Artificial Intelligence (AI) and Machine Learning (ML) technologies to automate the detection and triage processes allows for faster resolution (with the interconnectivity and interdependence of data growing like never before, simple univariate anomaly detection techniques frankly do not cut it anymore). The most advanced companies are leveraging sophisticated anomaly detection techniques to pinpoint the oddities in their data instead of trying to manually find them buried in dashboards and reports. Using AI to automate fraud detection allows a manufacturer to instantly pick up on red flags such as excessive shrinkage in inventory, an abnormal rise in invoice volumes, split purchase orders, multiple payments made to vendors without any corresponding services rendered, unusually low or high bid price, and a sudden and unexplainable rise in customer complaints.
Companies need algorithms that look across a variety of data sources, metrics, and segments to uncover trends and relationships in order to more confidently assess where the true anomalies lie. But where do you start? Here are five common AI/ML approaches to fraud prevention and protection to consider for your organization:
- Anomaly Detection is an unsupervised learning technique to automate detection of anomalies and make it more effective, especially utilizing large data sets.
- Logistic Regression is a supervised learning technique that is used when the decision is categorical. It means that the result will be either ‘fraud’ or ‘non-fraud’ if a transaction occurs.
- Decision Tree algorithms in fraud detection are used where there is a need for the classification of unusual activities in a transaction from an authorized user. These algorithms consist of constraints that are trained on the dataset for classifying fraud transactions.
- Random Forest uses a combination of decision trees to improve the results. Each decision tree checks for different conditions. They are trained on random datasets and each tree gives the probability of the transaction being ‘fraud’ and ‘non-fraud.’ Then, the model predicts the result accordingly.
- Neural Networks is a concept inspired by the working of a human brain, using cognitive computing that helps in building machines capable of using self-learning algorithms that involve the use of data mining, pattern recognition, and natural language processing.
Real World Examples
The following are examples of common procurement fraud issues our customers are currently facing.
- Supplier Collusion – track trends in supplier spend to detect changes in price and volume above expected norms.
- Employee Spend – analyze employee expenses (e.g. travel, materials, business tools) to identify unexpected trends in duplicate, mischaracterized, or fraudulent activities.
- Purchase Order Fraud - Assess PO details including invoice frequency, vendor occurrences, and price to detect risk.
The corporate use cases for anomaly detection are practically endless, from spotting fraud to revenue leakage to system outages, you can quickly identify outliers that impact profit. As data has grown along with unpredictability, more attention has been devoted to predicting anomalies as a proactive measure, as opposed to a reactive approach. The technology is here today to help you take control and minimize profit-cutting fraud.
The Anomaly Detection Engine is a new offering from Kaizen that leverages proprietary machine learning algorithms to identify and quantify business' anomalies in a condensed, easy-to-use interface. More information can be found here.
ABOUT THE AUTHORS
Andy Williamson – Founder and Chief Product Officer
Bobby Falconer – Industry Expert and Business Consultant Lead
Originally published on 9/1/2022 by SmartIndustry
Mastering the Data Lake
A New Approach to Anomaly Detection
3 Anomaly Detection Techniques to Kickstart Your Analytics Journey in the Chemical Industry
Gartner Competitive Landscape For Data & Analytics Service Providers