In 2020 one of the dominant “must-have technologies” to succeed in supply chain management (SCM) is “data science”. Initially, any new technology is given a pass on “close scrutiny” or the need to deliver value – but eventually, patience runs out and seasoning is needed to make the transition from “fun to do” to “technology that upsets the social order”. This blog captures lessons from two past gurus. My favorite from Gene Woolsey is: Common folklore holds that the only absolute requirement for system acceptance is “top management commitment”. The hard facts are if the people on the ground floor do not see the technology as a help, they will slowly, and with malice, ignore it to death.
In 2020 one of the dominant “must-have technologies” to succeed in supply chain management (SCM) is “data science”. This has been defined as: at the holiday parties among corporate executives each must have two or three of these projects in the works that will once be finished, enable the firm to “leap tall buildings in a single bound”. Initially, the new technology is given a pass on “close scrutiny” or the need to deliver value. In fairness, this “pass” is absolutely critical. The current best example is the COVID-19 vaccines which pull on research started 30 years ago. I have observed firsthand the benefit of patience with new technologies that eventually generate substantial SCM benefits.
However, patience alone is not sufficient. Firstly, patience is in limited supply; by the second-holiday party, the executives need to point to some successes. Secondly, all new decision technologies require some “seasoning” to make the transition from “fun to work on, but not practical or a complete failure” to technologies that upset the social order and generate improved performance. The data science community is now at the point of identifying “seasoning”. One of the outstanding leaders in this effort is Polly Mitchell-Guthrie from Kinaxis. In some exchanges on this, I have suggested looking to two past seasoning gurus: Dr. Peter Norden and Dr. Gene Woolsey. Both were presidents of INFORMS and key contributors to the past and key contributors to INFORMS Journal on Applied Analytics (IJAA, then called Interfaces). At age 50 this journal has proven itself as the place to learn about seasoning. In this blog, we will review a few highlights for each of them. This is not a substitute for reading their original work.
For many years Prof. Norden taught an applied modeling course (then called IE/OR industrial engineering and operations research) for seniors and graduate students at Columbia University. I was privileged to co-teach this class with him for four years. Below are some guidelines he provided students.
Core Modeling Process
- Structure the problem.
- Manipulation generates insight.
- Translate insight from a model into meaning options and recommendations for “manager” or audience – just a discussion has little value.
To be a smarter modeler:
- Learn about the manager, the working staff, and the organization.
- Learn to ask the right questions.
- Identify/recognize relevant elements, significant parameters, and significant relationships among elements and parameters.
- Speculate the right size and right content (bounding, include, exclude).
I first experienced a presentation by Prof. Woolsey in 1983 at an IBM Modeling conference in Boulder, Colorado with Gary Sullivan. I was early into my career, Gary was a “senior analytics professional” and manager of the advanced industrial engineering (AIE) department in Burlington, VT. Gary’s observation was, “I cannot believe IBM let Gene talk to an IBM group, but he is on the mark.” I left with many takeaways, three were:
- I am the invoice – to be successful the industrial engineer must know the core process and work well with the people who know the process.
- Where is the Data? At this time many groups were trying to build models first and later attempt to find the data required. It is critical to secure your data supply line, which Gary was doing in building LMS. I am sure Gene would update this for today’s world with “secure the right data”.
- Managers would rather live with a problem rather than implement a solution they do not understand.
Gene was a prolific writer of “short, but true tales” to demonstrate the importance of “seasoning” with technology to achieve success. Many of the papers were originally published in IJAA and a collection of these can be found in Real-World Operations Research: The Woolsey Papers. Two of my favorites are:
On System Acceptance: Abstract: Common folklore among members of our profession holds that the only absolute requirement for system acceptance is “top management commitment.” May I respectfully suggest that, in my experience, another requirement is equally absolute: user (bottom-level management) acceptance. I have now seen the spectacle, in a number of cultures, of splendid systems going down to defeat and taking the committed top management right along with them. The hard facts are these: if the people at the bottom of the managerial heap who will have the joy of facing the green screens every day don’t see the new system as a help, they will slowly, and with malice, ignore it to death.
Some Reflections on Surviving as an Internal Consultant: In this article, Gene covers hiccups internal consultants run into and how to work through them. These lessons apply to SCM initiates today. One issue identified is: Line Group X asserts that Y is true, we have a model that says they are wrong, how do I tell them this without being rude? Tactic: We modeled your situation, and it suggests a different policy. There are two reasons: our model is wrong, or our data is wrong. Could you take some time to show us where we messed up?
For a new technology to be successful it requires the correct seasoning which requires time in the trenches. Success is defined when the organization cannot imagine life before this technology. The new hot technology, data science, has reached the same crossroads as other technologies. The experienced and naive alike will find guidance from two past gurus helpful in implementing technologies that upset the social order. Although a manager will live with a problem rather than implement a solution they do not understand, they also know complexity exists where you ignore it or not. They are anxious for a quality that can handle the complexity but is still understandable or reasonable.