Why Now? The Power of Root Cause Analysis in Times of Change
With much of the world focusing on the present disruption to their operations or, at best, looking forward, we are seeing many of our customers struggle with prioritizing root cause analysis.
Isn’t the cause of service failures force majeure?
Why analyze issues that are (hopefully) once in a lifetime?
Why Now?
As I have written before, there is never a bad time to be thinking about root cause analysis. Case in point, the U.S. Army uses after-action reviews to reflect on even the highest-pressure combat scenarios to create learning opportunities. Using a similar methodology, organizations can glean valuable knowledge about how to successfully operate in the future during times when processes are stretched to their limits.
Beyond the philosophical support for in-the-moment root cause analysis are the practical necessities specific to supply chain planning. Shifts to e-commerce fulfillment are raising customer expectations on delivery precision. Brick-and-mortar vendors and retailers must also collaborate better to ensure products are in stock during demand shocks. Chargeback penalty programs have not gone away and may continue to tighten as retailers fight for reliability and reduced costs. In the midst of all this change, the root causes of service disruptions are not the same as what you have researched before or what your system’s default settings are reporting.
Better and Faster
Manual research into root cause analysis can take up a significant portion of a planner’s workweek. At a time when the focus is on adapting to day-to-day service issues, even less time exists to properly identify trending problems.
Additionally, the emergence of new root causes of disruption complicates this analysis. For instance, when an entire fulfillment center must close unexpectedly for sanitization following a positive COVID-19 case amongst its employees, the root cause of potential subsequent late deliveries must begin to consider the robustness of the distribution network in this new normal of supply chain disruption. Companies stand at a turning point where historical methods for determining root cause will lead to biased assessments of what is going wrong and misguided investments into correcting issues unrelated to the disruption that won’t improve customer outcomes.
What can be done to address this quagmire? Fortunately, there’s an answer.
AI-driven root cause analysis fills the gap. First, synthesizing data across an enterprise leads to unbiased predictors of service issues instead of following a pre-defined research path that leads to the “usual suspects,” facilitating better scenario planning. (For example, a manual analysis of late deliveries often mistakenly lays blame on the last link in the chain: a transportation provider.)
Second, machine learning algorithms allow the analyst to go beyond the broad process steps contributing to failures. He or she can delve into the patterns of predictors that could not otherwise have been connected and identified without AI. Detecting root cause patterns also provides a path from the diagnostic into the predictive realm. When we know what characteristics or delays lead to late outcomes, we can intervene proactively through actions such as expediting orders and other ways of optimizing demand planning.
An extension of root cause analytics is the concept of perfect order flow, or how the process would operate with no exception handling. We see a growing interest in perfect order flow as delivery expectations become increasingly precise. An algorithmic approach capable of assessing not only failures but also successful orders — with minimal additional effort — can lead an organization to identify the conditions and requirements for successful deliveries.
Learning Domain Expertise
In working with several companies in consumer-packaged goods, we have recognized that at the heart of the vendor-retailer relationship stand analysts who aim to keep a constant eye on service for their customers. However, analysts face challenges when quantifying the impacts on service that the many tangible problems that the operation can subjectively describe. They find that these anecdotes do not cleanly map to sets of business rules that can be applied to analyze the data. This prevents them from painting a convincing picture of the state of fulfillment between vendor and customer.
One of my former clients described these patterns as the “data signatures” for root causes — unique and nuanced, but immediately recognizable when seen in the data.
Root cause analytics techniques based on AI enable this automated generation of domain expertise. For example, in a customer deployment of Coupa supply chain analytics solution, specifically the supplier chargeback analytics capability, an algorithm calculates the importance of business process and timestamp variables. It then uses these to predict late deliveries produced by delays in the picking operation. It can even drill down to particular product categories in a specific distribution center that will experience an unusually high impact.
When the analyst reviewed this diagnosis, she was able to connect it to a machine where a certain type of packaging would frequently get snagged. This depth in root cause analysis allowed the organization to weigh the cost of a corrective action against the benefit of improved on-time delivery and reduced penalties.
Arming analysts with rich supporting insight for delivery issues that point to data-driven improvement opportunities elevates the vendor-retailer and/or retailer-customer relationship to new levels of consistent service. We witnessed this with a major CPG manufacturer and one of their largest retail customers. With a clear picture of where breakdowns were more and less likely to occur, the two organizations partnered together to determine where to invest in their network to meet increasing consumer demand for speed and precision. This type of collaboration in demand forecasting is especially meaningful amidst the volatility of a COVID and post-COVID environment.
A Time to Test and Learn
While many aspects of how organizations serve their customers continue to be in flux, a system under pressure can also provide a great testing ground for continuous learning.
AI-driven root cause analysis facilitates fast, balanced, and actionable insight that complements the domain expertise of people who know your business best. Now is the time to embrace the future of root cause analysis using AI.
