- Jim Gavigan
So, Why Manufacturing Intelligence?
This blog originally posted on Jim Gavigan's LinkedIn page here.
This blog gives some insight as to why that when I left OSIsoft in May of 2015, that I got into the delivery side of the business instead of just selling software licenses. Unfortunately, that opportunity didn't go as well as planned, but the seed for starting my own business was planted (maybe watered) during this time. Here are my reflections from back then:
So, why am I starting a manufacturing intelligence practice? Well, you could say my love for data and what it tells me as an engineer goes back almost 20 years. I remember using Modlink to pull some data over from a Modicon PLC to find out how much chemical usage we had for a modular pumping system we were building for a customer in 1996. I used the same techniques to help an engineer working for ABB better understand loss of life calculations in a large transformer a year or so later. He wanted the transformers to be more reliable and just needed the data to help him better understand how to make a transformer survive longer in the field. I was too inexperienced to really understand the financial impact of either application, but certainly for ABB, the impact would be huge.
As my career changed into a sales role, I still gravitated toward data and what it told me. I was an avid user of our COGNOS system when I worked as a sales engineer at Rockwell Automation. I used it to start to understand what we were and weren't selling to customers in my territory so I could better understand how to attack a well penetrated market. I even remember taking some good-natured ribbing over writing a macro that I could run against spreadsheets I downloaded from COGNOS that would tell me who the salesperson was from the distributor I worked with for each account across each analytic I ran. That way, I could send a list to each salesperson from some analytic I ran and say "Bob, here is what we aren't selling at your accounts - how do we change this? Is there a reason? Did you even know?" This approach helped us grow what many saw as a tapped out market by 25% over 5 years. We started to understand who was and wasn't buying certain products and that allowed us to delve into why. Sometimes it was a lack of education, sometimes competition, and sometimes the customer just didn't need that particular set of products from anyone. But again, we asked "why" and the financial impact was huge.
My first tour of duty with LSI taught me about the PI System, which I and so many others in this industry have held in such high esteem for so many years. We were doing some work for a large customer of ours producing cattle feed of all things, and PI showed us where we were struggling with batch accuracy. The customer came to us with a request for much better accuracy than we had been able to produce. We asked "why?" and found out that they were trying to land a new customer and accuracy would be critical in landing the customer. Once we saw the data in a way we could ingest, we solved a problem we had been working on for months in just a matter of days. PI didn't solve the problem, we did. But, what PI did was tell us where to look, and how to look at our controls problem. Again, we strived to understand what the data was telling us to solve a problem. What the data told us allowed us to make two companies millions of dollars each - one doubled their business and had to build a new plant; the other did a lot of the building of said plant.
I remember working with an ice cream manufacturer just before leaving LSI and my second day on the job at their plant, I watched what looked like perfectly good product falling off the conveyor belt going to the waste bin; not having been picked by the robot pickers at the end of the line. Everything looked perfectly good to me. I then inquired about why this was happening to the automation engineer I was working with. He looked at me and said "that is the million dollar question we would love to have answered." Again, the answer was in the data. I had a friend and supplier come in and train me on how to decipher the Cognex vision system and I was able to start understanding what the data from it was telling me. I was able to understand how many of each type of defect was happening per shift. It turned out that their vision tests were too stringent and they were rejecting perfectly good product. This was worth thousands of dollars per day per line. So, the data told me that the marketing department shouldn't tell manufacturing what is "good" and "bad" in all cases. Yes, the marketing group help set the guidelines for the vision system test was developed. Again, the vision system was rejecting thousands of dollars worth of product daily on each line - all we had to do was ask the data why ice cream bars were being rejected. Again, once I asked "why," the financial impact was huge.
Then, my career took me to a two plus year stint with OSIsoft, the makers of the PI System, the market and industry leader in real-time data infrastructure. Working day in and day out with the PI system showed me that so many customers from so many industries struggled with the same thing - how do they best understand what the data they are collecting is telling them and how do they get beyond just being reactionary and letting PI be a break/fix tool for their business? What I started to learn was that there is power in context. Context in terms of assets and events, and to notify people when patterns that are detrimental to the business are happening so that the impacts of these bad production or operational patterns are minimized. I also saw that there will be questions that one will want to ask of the data in the future, but one may not even know what that question is today. Organizing the real-time manufacturing data in terms of assets and events, and combining it with other sources of data from the web, maintenance systems, financial systems and the like allow one to ask these questions and see correlations that haven't even thought of until the data is centralized and organized. Now, the power of data can truly be unlocked, because context is key.
So, my passion for collecting and analyzing data really has been intertwined in my career. I was a controls engineer first, sales person second, and have blended back and forth between those two roles over the last 10 years. In every position I have had, using data to help me or my customer get better has always been critical and the financial impacts have been huge. So, I guess it should come as no surprise that I am starting a manufacturing intelligence practice. I have been a believer in data for my entire career. I remember listening to an exchange between one of our industry principals at OSIsoft and a CIO friend of mine. We ended up with this quote: "you have to turn data into information, information into knowledge, knowledge into action, and action into dollars." That is my passionate pursuit, and has been for almost 20 years whether I realized it or not - turning data into dollars. So, starting a manufacturing intelligence practice only seems natural to me, and going back to LSI was certainly the right fit to do so.