Nexus Helps Leading Energy Provider Save Millions of Dollars through an AI Sales/Use Tax Analysis Solution

About the Company 

A Fortune 275 company based in Irving, Texas, this leading energy provider operates across 20 states and the District of Columbia, serving about 4.3 million customers with a diverse portfolio including electricity, natural gas, and renewable energy plans. As the largest competitive power generator in the U.S., the company boasts a capacity of approximately 39,000 megawatts, utilizing sources like natural gas, nuclear, solar, and battery storage.

Goal

The energy provider aimed to enhance the accuracy of their Sales and Use tax payments by verifying invoice data before committing transactions to the General Ledger.

Problem 

• Invoices often contained inaccurate or missing data, leading to incorrect entries in the General Ledger.
• Tax Determination Engines (TDE) like Thomson Reuters ONESOURCE Indirect Tax operated without integrating relevant business rules or context, hindering effective error detection.
• Errors in tax reporting were typically more expensive to address after being committed to the General Ledger.
• Processing over 18,000 invoices monthly made manual verification impractical without intelligent automation.
• Difficulty identifying and correcting miscategorized invoices, often resulting in tax overpayments.
• Needed to maintain compliance with Sarbanes-Oxley and improve the efficiency and accuracy of tax reporting processes.

Solution 

• Tax Assistant was integrated withthe provider’s TDE and ERP systems to systematically flag invoices needing corrections, accommodating the high volume of invoices processed.
• Utilized advanced analytics to detect patterns leading to overpayments and other inconsistencies in tax data.
• Conducted real-time reverse audits on purchases before transactions were committed to the General Ledger, allowing for immediate rectification of discrepancies.
• The system adapted by learning new patterns from user interactions, improving its accuracy and efficiency.
• Directed questionable invoices to a review queue for assessment by tax experts, ensuring human oversight where necessary.
• Separated invoices into three categories—'The Good,' 'The Bad,' and 'The Maybe'—to streamline the review process and prioritize attention to potentially problematic transactions.

Results 

• Continuous improvement in the accuracy of aggregate tax payments as the system learned from interactions and corrections made by the team.
• Automated systems handled routine data processing ("grunt work"), allowing the tax team to focus on strategic tasks ("knowledge work").
• Insights and corrections became integrated into the system's processes, enhancing future operations.
• Enabled real-time assessment of new vendors and purchase patterns, improving decision-making.
• Errors were identified and corrected before being sent to the General Ledger, preventing future discrepancies.
• The proactive approach in correcting entries before finalization significantly reduced audit costs.
• Overall tax efficiency of the firm was optimized, leading to cost savings and more accurate tax compliance.

Conclusion

The Nexus Cognitive partnership has built an environment with iterative development to quickly create new reports with emerging data as the business needs change. The system continues to help the energy provider better manage operations and respond to an ever-changing environment.