Dear Trevor, Please Get Well

Dear Trevor

Please get well. If writing an open blog to me makes you better quicker, please accept my reply.

We have known each other for three decades. We have fought an endless and thankless battle to improve supply chain planning. I am unsure what drives the fight within us, but we fight on. Your thoughts have helped me a lot. I am thankful.

I, too, miss Roddy Martin. Roddy and I fought like brother and sister, but the arguments helped me gain clarity. Now, I struggle with the vacuous gulf between thought leadership and social media presence. We have many experts writing opinions but too little genuine research. The number of event companies marketing themselves as research firms is an issue. There is no Hippocratic oath in our industry to do no harm. The aimless podcasts and meaningless panels drive me crazy because we have not effectively improved decision-making in supply chain processes by implementing processes and technology. Yet, we espouse it.

Yes, we have improved transactional efficiency, but we remain unclear on how to best use data to drive insights to align the enterprise with ever-changing markets quickly to drive growth, improve margin, and deliver reliable customer service.

I appreciate your experience and intellectual curiosity and have gained much from our open debates. Here is my answer to your questions:

My root issue with Artificial Intelligence is the lack of clarity on what drives value in a value chain. If we invest in quicker and deeper answers against the wrong outcome--say cost or OEE--we sub-optimize shareholder value. Should public companies care about shareholder value? I think so, but not at the expense of the customer experience. I embrace AI if we are clear on the outcomes. Today, we are not.

The problem is that we have not defined value. Let's take an example. For the past three decades, companies have invested in technology and process automation in demand planning, but in nine out of ten companies I worked with last year, the result was a negative Forecast Value Added result. When we degrade the forecast, we increase risk and bullwhip while throwing the supply chain out of balance. Yet, we do not measure it. I think that we need to align on what is value and the drivers/discipline to get there before we apply shiny objects to the outcomes.

You and my friends at Lokad are pushing probabilistic forecasting. This is a concept that I think makes a lot of sense for inventory outcomes. But, the question that I cannot get answered is what is the role of the forecast in a probabilistic model for driving planned orders in manufacturing, aggregate buying in procurement and making a feasible plan for logistics. Doesn't there need to be a visibility translation layer? And, in probabilistic forecasting, there is no forecast for the supply chain planning system of record. The focus is on outcomes. My question is, "Doesn't the supply chain planning system as a system of record need to record a forecast to align the organization?" I don't have a good answer. I want to focus on the redefinition of work. I am tired of focusing on technology for technology's sake.

I spoke to Mike Landry of KetteQ today about the interoperability of pipeline forecasting in SalesForce.com with a ship-from model in demand planning. Doesn't there need to be a unified data model and a translation layer? Ninety-nine percent of my observations are that supply chain planning is a ship-from model. Companies struggle to put channel data in a model that focuses on which items to ship from a DC because there is no logical representation of the channel in the model. I chided Mike for labeling this autonomous planning without defining the process flow. However, I am glad to have new market entrants to challenge the status quo.

This is an example when I asked the students in my current outside-in planning class to map data sets to improved models in the Cynefin model. The problem is that today, functions own data, and traditional Advanced planning only helps us with the known/knowns. Over the past three decades, organizations have become more supply-centric, making a supply chain a function to compete with other functions.

Gone are the days when the term supply chain connotated the process of improving outcomes in the delivery of goods and services from the customer's customer while orchestrating make, source and deliver across a value chain. We have largely failed in the definition of decision support for the Global Multinational.

Consider the homework completed by one of the students. We cannot move forward unless the organization owns supply chain outcomes and we can challenge conventional norms. Companies need to embrace outside-in processes that embrace and use unstructured data, streaming data and structured data together to improve sensing. Today's supply chains do not sense. They often respond badly.

Even worse, in many companies we confuse responsiveness with reactivity. Reactivity, or knee-jerk reactions, are even worse.

In short, hurry back to health so that we can work on this together.

All the best in your recovery!

Lora