I interviewed David McPhetrige who discussed Predictive Analytics for Inventory Optimization.

 

 

 

 

 

 

David, you have told me that you are passionate about predictive analytics for inventory optimization. Why?

 

In the supply-chain world, “predictive analytics” is already in use:

 

  • It enables optimizing supply-chain networking design, structure, and what-ifs
  • It’s fundamental to improved forecasting, such as demand sensing

 

There are already plenty of great solutions on the market for supply-chain network design and forecasting. These are important high-level pieces of the supply-chain-optimization puzzle.

 

Years ago, I began pursuing predictive analytics for a critical, but overlooked, piece of the puzzle.

 

It sounds like you have already listed all general supply-chain categories. What’s missing?

 

Understandably, supply-chain professionals focus on what a supply chain’s future state could be – “we can get this improvement if we re-structure our network like this,” or “we need a forecasting tool that will provide X improvement.” To use predictive-analytics in these areas requires changes, and changes always take time.

 

But in the meantime, their businesses are running on their existing supply chain. I wanted to develop a solution that can provide big improvements – simply by changing one or two values in each inventory item’s Item Master record. These are values that don’t require changes to the supply chain – no negotiations, no improvement projects, no brick and mortar. NOW, not in the future.

 

I knew that the key to this was to use the power of predictive analytics to determine and truly optimize these one or two values.

 

First, what are these values would you change?

 

The values that trigger and quantify replenishment: Safety stock if MRP, quantity per card and number of cards if multi-card Kanban, red zone buffer if DDMRP, min and max if min-max, etc.

 

How do you use predictive analytics to do this?

 

Here’s an example:

 

  • Let’s say an inventory item’s target fill rate is 98%
  • Last quarter, we achieved that fill-rate targetwith an average of 100 on hand
  • Does this mean that 100 is the item’soptimal level for next quarter?
  • Clearly, no – because next quarter’s demand and supply values will be different – even if nothing changes in the item’s supply chain
  • But if 100 isn’t optimal for this item, what is?

 

First, everyone knows that constantly changing an item’s one or two inventory policies that drive its inventory level doesn’t work – in fact, it usually makes things worse. And everyone knows that formulas don’t work.

 

Predictive analyticsdoes work:

·

  • Let’s say a business measures its actual service-level, or fill-rate, for a quarter – 3 months. Perhaps its target fill rate is 98%
  • First, predictive analytics correctly represents all relevant factors – about 12 – for any inventory item’s actual supply chain
  • Second, it performs thousands of independent, random simulations of an item’s actual demand-and-supply process for, in this example, a quarter
  • For each simulated quarter, it finds the optimal inventory level that achieves the item’s target service level – without expediting
  • Predictive analytics now has thousands of optimal inventory levels
  • Finally, from these thousands of optimal inventory levels, predictive analytics selects the level that provides the desired confidence level of achieving the target fill rate in any quarter

 

That inventory level may support, say, achieving a 98% targetservice level for, say, 7 quarters out of 8, or 11 months out of 12, or other confidence level. The target service level, service-level cycle, confidence level and many other inputs, of course, are variables

 

How did you ever figure out the formula for this?

 

Well – I realized that modern computing speeds now enable a probabilistic simulation – what some call Monte Carlo simulation. So over the course of many years, I developed, perfected, patented and marketed my predictive-analytics solution.

 

It optimizes any and every inventory item – in any and every location.

 

This seems like a very narrow scope for a solution. Why?

 

As I said earlier, I see plenty of great predictive-analytics solutions for high-level supply-chain optimization. Rather than competing with them, or with big ERP, my goal is to enhance any solution by providing a big business improvement that can be implemented immediately.

 

And finally, where might I learn more about is your predictive-analytics solution?

 

Yes. It’s software-as-a-service, and it’s w-w-w dot right-sized-inventory dot com.

 

 

 

About David McPhetrige

 

 

 

AAEAAQAAAAAAAARIAAAAJDFkMWI1ZmIzLWJlN2MtNGVmYS04ZGM1LTU5YWE5YzFkOGFiMw.jpg

 

David McPhetrige

 

 

Business Partner at Right Sized Inventory

 

LinkedIn Profile