Trevor: Rollover and Think Harder

Greetings Trevor. I know that you cannot talk as you recover from cancer, and it must be awful laying in a bed reflecting. I am sorry.

I smiled when I found another LinkedIn note penned to me in my feed this morning. Here, I will answer it.

Hopefully, your intellectual pursuits--and this active debate-- will help you get well faster.

Embracing Our Geekiness

In this picture, I show you in healthier days reflecting during the 2014 Supply Chain Insights Global Summit over a glass of red wine. We go way back, don't we? Hopefully, age makes us better.

I look forward to enjoying a glass of red wine in the future, but I warn you, I might ask you to buy. I think somewhere back you owe me a bottle.

As a chemical engineer, I nodded when I read your blog. Your post listed your studies in systems theory, flow, and system dynamics. I get it. I agree. The world would be far better if supply chain leaders understood these concepts. It would shift our focus from transactional efficiency to better understanding and managing variability.

I'm curious what Jay Forrester would say if he could talk to us about using intelligent agents, large language models, and narrow AI. Our technologies allow us to solve problems that were once only theory. Yet, I am struck by how much, in the past, most experts are focused only on optimizers.

My Itch

Your reply makes it seem that I am not communicating. I'm so sorry about this. Let me try harder.

My issue is less with your choice to pursue probabilistic optimization and more with the opportunity cost to the industry if you only focus on optimizers. I want you to focus on improving work processes. Your talents are too deep not to drive more significant innovation in planning processes.

Bear with me as I construct my argument:

1) Why don't companies use their planning systems? Let's start by considering why companies don't use their planning systems to drive insights. As you and I know, there are many reasons. Today's systems are batch: a barrier to what-if optimization and simulation.

In the industry, talent turnover is high, and the knowledge level of business leaders today is less than what I experienced ten years ago. (Maybe I am getting older and crankier, but few teams can have the conversations I experienced a decade ago.)

As you and I know, a spreadsheet is the most used planning system. In short, modeling in a spreadsheet is inadequate on many levels, but it is done because our systems are challenging to use. You and I are part of a system that has largely failed the large multinational with hundreds and thousands of planners that turn over frequently. I would love to have you think deeply about the redesign of work with intelligent agents and self-service planning. I envision using Large Language Models to guide exception workflow and connect supply chain master data management to role-based exception management, enabling new forms of discovery. I envision expert models that train at the speed of business based on process-specific data.

2. Can A Faster Horse or a Better Mousetrap Help? I don't think so.

I am sure you remember Henry Ford's quote, "If I had asked, customers would have requested faster horses." (Forgive me for paraphrasing a statement with many renditions and questionable legitimacy.) Teams cannot conceive the possible. I want to push you to break the bonds and paradigms that limit our current thinking and redefine work.

Companies do not question forecastability before they slap in a technology bought from a Magic Quadrant. They are surprised when there is no magic. Most discrete configure-to-order companies have a finished goods Coefficient of Variation (COV) greater than 1.0. With this high variation, can we agree that the answer is not a better optimization engine for forecasting?

Yet, companies implement forecasting technologies in undisciplined ways without backcasting, data cleansing, and testing forecastability. I envision an AI engine that ingests, cleans and aligns market data, evaluates forecastability, and shares insights on demand latency/market drivers. The goal is to align the demand flows to the right models and supply strategies. I am proud of the progress of companies like BSH and Nestle that adopt better techniques from taking the class. Figure 1 is the BSH slide from the Youtube video of the outside-in case studies.

Figure 1. BSH Case Study from Outside-in Planning

I feel that we should not match demand and supply volumes in S&OP. Instead, I think that we should align demand and supply cycles and strategies based on flows.

Conventional demand planners do not focus on cycles and flows. Instead, they try to get precise on imprecise time-phased data. Unfortunately, too few companies have the discipline to align models to forecastability. The views of time-phased data do not enable the visibility of patterns. Companies lack discipline: few backcast to test models/optimizers or measure Forecast Value Added (FVA).

Instead, competent technologists and professors drivel on about optimizers. Enough, I say. In the words of Freddie Mercury in Bohemian Rhapsody, "We need to think bigger."

Figure 2. Thinking Holistically about Demand Management

3. The Role of the Planner. Being a planner is a thankless job. We have trapped bright people into working with archaic technologies trying to explain the output of planning to executives. In our research, we find that the planner has the lowest level of job satisfaction in supply chain.

If I had a magic wand, the planning role would become an orchestrator of intelligent agents, and we would implement self-service planning for the business leader. I think that we can do this with AI. Get well so we can get to work!

Get Well Soon

I look forward to seeing you and having a glass of wine.