New Belgium Practicum Experience

Invited to participate in the Supply Chain Practicum was an invaluable experience as it was challenging, outside of my comfort zone, and led to incredible outcomes. This experience can be segmented into three phases: the initial data discovery, the trial and error phase, and all hands on deck phase.

Reflecting Back

Looking back to the early weeks of January I was enthusiastic about this rare opportunity and what value I could bring to the table. I felt confident about my skills and ability to solve problems. The exact day before the course started my life was dramatically changed by the most traumatic event to date. At that point I did not want to take this course I was certain I would not have the mental ability to keep up, work on, or understand how to generate solutions. I was terrified and asked to drop the course to allow some other bright eyed human to rise to the occasion. Somehow I didn’t end up dropping the course and I showed up to the first meeting greeted by my teammates Reagan Travis and Tyler Shipstad. I had previously met Reagan and never met Tyler before. From the initial conversations I knew these two were top level talent and business to the core. They were ready to conquer this project. This was the first time I experienced the pit in my stomach and recognized I needed to hold on tight. Reagan was gone the second week where Tyler and I ventured on site to New Belgium to have our first discussions with Travis Burge the Production Planner, Mike Woodard the Demand Planner, and Brian Goodwin the Master Scheduler. We met in a small room in New Belgium passing multiple divisions to get to. I was intrigued by the decor and the people in the room. It was what you expected extremely eclectic with a variety of hip faces. I remember noticing the little details like crushed glass bottle walls and bottle caps inlaid to the cement counter tops. Glancing at the computer screens within the Branding and Design department my heart fluttered for the familiar interphase of Photoshop and Illustrator. What was I about to get myself into?

Sitting in the first meeting I was naturally quiet acting as the scribe, writing down everyone’s input on what the structure of the project should entail. I cannot lie I was slightly overwhelmed by the vast tangents and immediate scope creep we were walking into. I commend all of their excitement and passion, but there was information delivered that contradicted what we believed our task was. We discussed everything from aggregate orders between Fort Collins and Asheville, optimizing their canning line, and the impact of the highly variant variety packs. It was none-the-less overwhelming. Leaving the meeting Mike Woodard took Tyler and I through one of the brewing cellars to a giant cooler room with miscellaneous bottles and cans and decomposed six packs. He told us as “employees” we could take whatever we like. During this moment I am not sure what state of confusion I expressed, as I do not drink beer let alone alcohol. I graciously collected a six pack pretending I knew what I was picking out only to give to my big brother.

Leaving the meeting we both knew we had our work cut out just forming a solid project scope. We informed Reagan of the meeting and sent off the notes. The following week we scheduled a meeting to access their Data Cube and get a private tour of the facility. On the tour I was informed of how special and unique this establishment truly was. There was love and pride bursting from the floorboards, from metal bolts on the canning line, all the way to the tee shirts being sold in the gift gallery. During our meeting all three members were present. With the help of Andrew from IT we all had our own unique logins and were ready to roll. We then discussed New Belgium’s old forecasting methodologies and how they created a system called Chimple (so simple a chimp could use it) illustrating the starting point of where their forecast numbers should fall (and can be adjusted from there). They then went on to discuss the laundry list of external conditions that impact the sale of their products, the craft brew industry as a whole, and the market. There were many other things we had to be aware of beyond the seasonality of the product.

Leaving this second meeting the voice in my head questioned whether this project was really a production problem versus the inefficiencies of understanding their consumer’s demand for their product? The forecast was clearly a problem, but what was the root cause of these levels of variance?

The Data Discovery

Charging into the computer lab the next day the CSU team prioritized the role of managing our clients to the best of our abilities. We were to tell them what the project scope was to be and what our final deliverables entailed. Mapping out our Project Plan we decided on a “Bottoms Up Approach” to analyzing the behavior and variance of the Variety Pack product family. Getting into the Data Cube was a task within itself going through multiple portals with a variety of logins to activate our excel worksheets. At first glance, we felt pleased with the amount data and it’s organization, little did we know every product type, package type, and flavor had coding we were unaware of. This was our first wall having to decode and understand what we were even looking at. Mike provided us with a data dictionary that labeled the top products and points of information, etc. We also inquired about what areas of the data he used the most when creating a forecast for the Variety Packs, but aside from that it was up to us on how to approach the problem.

Prior to the Practicum I had taken Marketing Analytics taught by the wonder Dr. Gina Mohr. I fell in love with the course and the ability to organize, manipulate, and extract meaning from Big Data. This course contradicted most of the material we were provided with by New Belgium. Everything was imported through pivot tables in Excel and finding the raw data was a process of digging behind this interphace. I was immediately not a fan of trying to navigate their system, as I felt it directly pointed us to the bias that was already affecting their forecasts. Coincidently we lost access to the Data Cube and Mike and IT sent us off a static Data Dump of the raw data for Shipments and Depletions of their Variety Pack products. This was the information I was interested in seeing. The data in its most raw form.

The three of us all began varying versions of analysis to see what information we could derive from it. All taking different approaches we looked at sales by Market, Distributor, and SKU number. There was a lot of information in these workbooks that clearly coincided with the market trends shown in the Nielsen Data.

The team feeling content about having this data lead us to questioning how we were going to effectively create meaning and give New Belgium a forecast solution?

The Trial & Error Phase

Data, data, and more data. Living and breathing New Belgium Variety Pack data we explored a variety of ways to produce forecasts. Weeding out ineffective practices. My initial thought was to use a Holt Winter Multiplicative Model which drives statistical meaning to deseasonalizing and forecasting highly perishable products. This model operates on a 2X12 moving average and was taught during my analytics course last semester. This complicated forecasting model is conducted through Excel and is complicated to follow let alone build. Tyler and I were able to team up in creating this model for Variety Pack Shipments by Marketplace. Understanding the complexity and statistical explanation was going to be the hardest part to sell for using this model. And after conducting a few error assessments we settled on looking for more forecasting approaches. We were proud for our efforts to include this method and the upper level thinking that went along with it.

The next week we scheduled a check-in meeting with New Belgium where we met in the Bamboo Room. There was something eerie about walking on site that day. Things were shut off and the office was full of people standing around not working. My hypersensitive self could tell something was not right and when we were escorted to the meeting room by Mike and he successfully avoided answering what was going on around there. We went over our general progress in the meeting and asked clarifying questions. At the end of the long conference table sat a gentleman clearly distraught, crossing his arms and looking out the window. During the meeting no one acknowledged his presence and they simply let him be. Shortly after leaving we were informed of New Belgium’s layoff and my heart hurt for that gentleman. I could not imagine his anxiety and discomfort. Not only did this layoff impact many individuals it also impacted how carefully and tactifully we would now have to handle our clients. These people were going through big changes. This work environment certainly impacted our overall style of management throughout the remainder of the project.

All Hands on Deck

We had one last check-in meeting over on site where we discussed how things were going and inquired questions about an economic denominator we could use for a cost analysis. Gauging Mike’s response he was not intrigued about the cost questions. After this I finally spoke up and asked if there was potential for a member of the Marketing Sales team that could join our meeting. Enter Clay Hoffman. Two months ago Clay came to New Belgium from Starbucks. He spoke of all of the external anomalies we had identified, but was equally aware of every justification for brand and marketing changes. He could speak on behalf of the selection of products for the quarterly Variety Packs fully explaining how the processes undergo. Again the voice inside my head chimed in, “ding ding.” The data we had been pouring ourselves into was equally utilized by both Sales and Planning, yet there was little if any communication between the two, aside from the broad roll out discussions. If there was anyway to have Mike and Clay’s positions communicate more frequently there could be a huge potential towards driving the forecast variance down producing to meet the demand more accurately. In my (marketing biased) opinion, I fe