A refreshing approach to sealing revenue leaks for a soft drink giant

It's hard to retrieve what you've lost if what you're looking for is hidden in a disorganized data jumble. That was the dilemma facing this name-brand global enterprise. The firm knew it was losing revenue from unfairly claimed invoice deductions, discounts, and debt write-offs that had reached $14 million by the beginning of 2019. Yet it had no systematic way of determining the root cause of these issues. Trying to retrieve any of that money seemed like a daunting, needle-in-a-haystack-type task, so the company needed confidence that the effort was worth the outcome. It had long regarded these losses as just part of the cost of doing business.

Our ongoing relationship with the company meant our operations teams saw first-hand the challenge of invalid deductions and the opportunity to address a problem that was only going to keep growing. What if we could set up an automated system that would allow the company to retrieve some of the money it considered long gone? And what if a clear picture of the source of the problem could stop it from happening in the future?

To help it answer these questions, the beverage giant brought us into its North American offices in January 2019. Our first step was to deploy a value share team – domain experts who measure their success by the sole criterion of achieving clients' desired business outcomes.

This hybrid project team performed in-depth analytics, working in close collaboration with company stakeholders and their operations team to share learning for ensuring corrective procedures. It confirmed what we suspected: not only could we retrieve about $1 million in just one year in formerly written-off monies, but as soon as the solution was up and running, we could also make sure such revenue losses didn't escalate in the future. We committed to stemming these future losses as part of our value share arrangement with the client.

With the go-ahead from leadership, we built customized automation enhancements for specific client challenges, such as invoice and contract extractors for effective data search and backup pull automation to pull backups from their existing system and convert them into Excel for further analysis. Using cognitive analytics, which quickly draws inferences from its existing knowledge base to make sense of the massive volume of unstructured information residing in deducted, discounted, and unpaid invoices. It then reuses these learnings for all future inferences, creating a self-learning loop that increases its accuracy every time around.

The system divided invoices into three categories:

• The first covered invoices with legitimately claimed discounts and deductions
• The second covered valid but preventable discounts resulting from delivery delays, processing errors, or incorrect shipments, among other causes
• The third covered unauthorized deductions, such as unearned cash discounts and refused shipments

The analytics revealed multiple pain points. For example, it found scores of instances in which overdue invoices still got an early payment discount. It unearthed multiple occurrences for which the same retailer received different payment terms, apparently at random. And it found vendors with a high level of invalid deductions who could be billed back to the tune of about $3.3 million.

Within a year of our deployment, we had processed roughly $10 million of invoice data and pinpointed more than $3 million of invalid discounts and deductions that the company is now billing back to its vendors. And we made good on our commitment to retrieve $1.7 million in a 12-month period.

The company is expecting an additional $3 million in recoveries by the end of 2020 and anticipates preventing another $1.5 million in leakage by the same date.

Relationships with vendors have also improved because the new and more accurate standardized system ensures fairer treatment across the board.