Business Intelligence vs. Inventory Optimization: Are You Solving Symptoms Instead of the Real Problem?
In my decades of working in replenishment and inventory optimization, I’ve noticed that many companies seek solutions to symptoms rather than the underlying problem. I want to let you in on a secret: software companies are happy to sell you something you don’t need. Of course, you’ll never want to deal with that company again.
What I try to do with my clients is to dig a little deeper to find out if they’re solving the real problem. For example, I get many requests for business intelligence (BI) tools; mostly because they want to monitor symptoms. They want to have an aging report for inventory or a report on vendor performance. While these are fine reports to have and tools to leverage action, the problem is that these are reactive processes.
Let’s examine the first: inventory aging reports. It’s great to know what inventory is aging and what might need to be written off, marked down or transferred. Too often, inventory levels are a result of human emotion. Sales might have promised better sales, a buyer might have been stung in the past by an out-of-stock and over compensated, a miss-ship of the vendor might have happened and the warehouse didn’t notify the planner, or consumer behavior might have changed and wasn’t noticed until the days of supply threshold was crossed on a BI report. The real cause is that the inventory did not compensate for what could happen. This is what probability forecasting does.
Sales targets, vendor performance, human emotion and consumer behavior are all factors that a strong machine learning-driven system can understand and compensate for. A modern system can take the probability of sales targets and smooth the impact on future inventory. Machine learning can understand the shipment performance of vendors and compensate for variability. It can also understand the day-to-day buying patterns of customers and make dynamic changes to adjust, transfer and alert before things become a problem and the potential of inventory might be in jeopardy (under or overstock). Ultimately, a modern inventory optimization solution can handle most of the exceptions before they become a problem that a BI tool would detect after the fact. Consistency of results removes the human emotion from the inventory chain. A user that sees good results has less need or desire to interfere (the machine learning engine is also reporting on such interference). BI tools show the problem too late. Solve the problem first.
How about vendor performance BI? I get asked a lot for the ability to somehow know the impact vendors are having on service goals to customers. Tools to track late shipments, measure vendor fulfillment performance, out of stocks, and such. These are great metrics to hold a vendor accountable and ask for some compensation, but they require that something happens and that humans must track down and enact some remedy. In other words, the pain has happened and now you must spend on labor to recover something from a poorly performing vendor. What if the inventory optimization system understood the vendor performance and compensated? A system that understands the delivery patterns of a vendor calendar (order, ship, and delivery), that understands how frequently delivery dates are not met, that compensates for common shorts is going to buffer for what is predictable. Because this is automated, those safety stock levels have a more concrete dollar value. Instead of the BI measuring impact after the fact, a modern forecasting system should compensate for the deviations and provide a means to proactively measure the impact.
Most BI reports that I encounter alert users to something that happens beforehand that could have been reduced or eliminated. Before you seek some BI tool or a tool that identifies a pain point, ask yourself: can you solve the cause of the pain point?