The Peanut Butter Problem
Adapted from an actual manufacturing analysis. Identifying details, dates, and financial information have been changed or generalized to protect confidential information.
I was working on forecasts for a manufacturing company while completing my MBA. Forecasting wasn’t new to me. I had models for volume and product mix, and I could build a reasonable picture of what I thought the business was going to do. What frustrated me was that even when we predicted the mix reasonably well, the costs didn’t always behave the way I expected. I could see where the forecast and actual results differed, but I wasn’t always satisfied that I understood why.
Then, in an accounting class, I was introduced to activity-based costing.
It was one of those rare classroom moments when I immediately stopped thinking about the assignment and started thinking about work.
Our products competed in two very different markets. One group consisted of relatively high-volume fabricated products. We competed against large steel converters that happened to make products similar to ours. The other group consisted of more specialized, lower-volume equipment sold against companies dedicated to that particular market. Those competitors had different business models and, I suspected, different cost structures.
We sat somewhere in the middle.
I knew what competitors charged, and their prices told me something about the assumptions behind their businesses. A large steel converter and a specialized equipment manufacturer weren’t likely to have the same cost structure. That made me question whether our own two product families were really consuming our resources in the same way. Yet inside our company, we were essentially treating them that way when we allocated overhead.
Our costing system spread overhead broadly across products using direct labor. Years later, I still like the phrase we used for it: we were spreading the peanut butter.
Activity-based costing made me wonder whether that assumption was reasonable.
I went to our controller and asked if he was familiar with activity-based costing. He was somewhat familiar with it, so we sat down with my course material and started talking about how the textbook concept might apply to our business. We also involved our marketing manager, who had responsibility for product management.
We spent a lot of time discussing what should drive a particular cost. Should it follow revenue dollars, manpower, square footage, equipment, or something else? There wasn’t one formula we could substitute for the old formula. We measured what we reasonably could and estimated where necessary.
There wasn’t much disagreement that I remember. Mostly there was genuine interest in finding out whether we had stumbled onto something useful.
The work eventually became our MBA team’s field project. We interviewed managers, examined costs and tested different ways of assigning them.
Then we ran the numbers.
The first result was disappointing.
The analysis showed that our high-volume products had been carrying more manufacturing overhead than they probably should have. So the hypothesis was directionally right. But the difference wasn’t large enough to explain what I was seeing in the business.
I don’t remember exactly what caused us to take another pass. It may have been a conversation with the team, my professor, or simply the requirements of the assignment. I do remember being disappointed. We thought we might have found something, and the numbers hadn’t given us much.
So we looked again.
On the second pass, we widened the question beyond manufacturing overhead. We started asking how other resources the company was consuming should be assigned between the two product families.
That’s where the analysis became more interesting.
Sales, marketing and engineering effort weren’t distributed evenly between the two product families. Neither were the assets supporting them. Those resources were concentrated much more heavily on the lower-volume side of the business.
That made sense when we looked at the operation. Some of our high-volume products were mature. They didn’t require much engineering attention. They weren’t receiving the same marketing support. Sales resources weren’t being consumed in the same way. Other products demanded considerably more attention and investment.
The way we had been allocating costs didn’t reflect those differences very well.
This time I was excited. We had something we could try that I believed might move both the revenue and net-income lines.
I also had something I hadn’t had before. Competitor pricing had bothered me, and the product profitability numbers didn’t seem to fit everything I could see operationally, but those were still hunches. Now I had an analysis I could take into a discussion about what we should do.
There was also a more personal satisfaction. The company was paying for my MBA. I was glad to bring something back from school that might provide a return on that investment.
We became more aggressive on price with the high-volume products. We also raised prices on some of the specialized products.
The specialized side worked about the way I hoped. We increased margins without sacrificing meaningful volume.
The high-volume side didn’t.
We lowered prices and gained some volume. But our large steel-converting competitors’ prices came down too. We could never seem to drop the price far enough to create the advantage I had imagined. We sold somewhat more product at lower prices, and revenue ended up roughly flat.
At the time, I took some consolation in believing our competitors were probably making less money too. I couldn’t see their books, so I couldn’t know that. What I did know was that understanding our costs better didn’t give us the cost structure of our competitors.
I wanted to institutionalize what we’d learned so that the differences we’d discovered would continue to influence the questions we asked. That never happened. We didn’t formally change the way overhead was applied.
The analysis still heavily influenced my decisions for the remainder of my time with the company. But I eventually left. We also hired a new controller. I don’t know how much of that thinking survived.
What did survive was the way I looked at the numbers.
When someone puts a product margin in front of me now, I know there are assumptions underneath it. How was overhead allocated? Why was that driver chosen? Does this product actually consume resources that way? What changes if the assumption changes?
I’m not the controller or the accountant, and I never will be. But I know enough to ask about the assumptions behind the numbers.
← Essays