Top 10 Stupid Six Sigma Trick #8

Optimizing Processes That Lose Money

So far, we’ve discussed Stupid Six Sigma Tricks #10: Conflating systems, methods and tools and #9: Confusing breakthrough with continuous improvement. This month, I’ll spend some time on a more subtle, and no less costly mistake that, in its extreme form, we’ll call Stupid Six Sigma Trick #8: Optimizing processes that lose money. “Well,” you might say, “Isn’t that what Six Sigma is all about? Taking unprofitable processes and fixing them or making profitable ones more profitable?” And you would be right. For the explanation, we need to dip briefly into the world of accounting. As your dentist says, “Don’t worry, it won’t hurt much and will be over soon."

Standard costs—Accounting 101

Why am I talking about cost accounting? Our perception of costs is one of the inputs we use in identifying, prioritizing and selecting projects for Black Belts to work on. If we get a distorted picture of costs, we end up selecting projects that don’t have the expected benefit to the bottom line (Hint: this is another reason Black Belts can claim millions of dollars of savings that aren’t reflected on the bottom line.).

Back in the day, when spreadsheets were big pieces of paper and computers were people who computed, a method was created for tracking the cost to make something.

Accountants had costs that varied with the items produced, such as raw materials, and costs that didn’t vary with what was produced and had to be paid whether I made one or 1,000 items, such as rent. The first are variable costs and the latter are cleverly referred to as fixed costs.

So the question arose: “How much does it cost to make this item? I want to make enough money selling them to stay in business as well as buy toys.” If you’re making identical widgets, then it seems to make sense to allocate the fixed costs across all the widgets you made, and when this is done it is called “overhead.” So, if I have $1,000 of fixed costs per month, and I make 1,000 widgets, each one bears the burden of $1 of my fixed costs, and this is called the "standard cost." If I add to that my variable cost associated with each unit, then I know the breakeven selling price and can set it accordingly.

The mismeasure of costs

Standard costs aren’t really an accurate description of the cost of manufacture, but if you manufacture commodity widgets at a constant rate, it’s probably not a terrible measure. The problem comes in when you aren’t making widgets, but making a number of different products (or providing a number of different services) at a number of different lines or locations and selling to a number of different customers. What happens if you take “overhead” costs, such as sales and marketing; the various support services, such as engineering and maintenance; and management, and allocate those equally across all the units you made that month?

Consider this scenario: One of your products or services (call it product A) may, by its nature, require a lot more work from your sales force to get the sale. Maybe it’s a high-price item (e.g., $100 per unit), so that revenue looks good, and because your sales force is rewarded on generated revenue, they keep at it to get the sale and the "Salesman of the Year" award. They work their tails off to sell as many of these things as possible. Using standard costs, this product may seem to be really profitable. When we look at the standard cost and figure out our profit, we expect product A to make $1.50 per unit profit; while product B, our old standby, makes $0.25 per unit.

This profit calculation is misleading because the cost of the sales force is allocated as overhead across all products equally. Because this product required more than the average amount of time to sell and yet has the sales costs allocated to it equally, it looks more profitable than it really is.

You’re thinking, "He’s going to advocate for activity-based costing (ABC), which allocates a cost to the product or service for every single action you do to make it." ABC generally takes a lot of work to set up and track, which dilutes or eliminates the benefit of doing it. I recommend allocating costs to maybe three levels: by customer, product, and location or line. This probably gets you 80 percent of the understanding ABC would get you with a lot less work. In fact, you can even use statistical sampling to estimate costs every quarter and not even directly track all the costs.

Let’s say that this product also requires a lot of support from your engineers—maybe they have to go on-site to work with the customer to set it up. Let’s also say that shipping has to use special packaging on this item, which takes them more time, and maybe running the product strains your equipment, so maintenance costs are somewhat higher. Again, these support functions are usually allocated equally across all products, so each time a product uses up more than its share of overhead it raises the standard costs of the other products, while its apparent overhead is lower than it should be. In other words, product A eats most of the pie, but splits the bill equally with product B.

The reason we had standard costs was so that we could make decisions about pricing and what to have our Black Belts work on. So, in our scenario, we have a product that’s consuming far more than its share of the overhead, but only gets charged for an equal share. If we subtract the standard cost from the revenue, we have what looks to be a very profitable product, which our sales force is selling as hard as they can. If we actually allocated all those costs, we might find that it’s a lot less profitable than we thought. Even this is only part of the story.

Total asset utilization

The cost allocation we discussed above looks at the resources used to make the product or service. It doesn’t account for the comparative opportunity costs as far as the process is concerned.

Let’s presume we have a single processing line that makes our products. (The case is stronger, but more complex, if you have multiple lines.) Remember that product A is high revenue, and is eating up more than its share of the resources. Let’s say we allocate the costs above and find that product A is significantly less attractive in terms of profitability than it was before when we were using standard costs. Maybe product A actually makes us $1 per unit, and product B makes us $0.75 per unit (our real profit margins).

Whew! For a second there I thought we might find out that product A is a money loser; management would have gone ballistic! Although it isn’t as profitable as we thought it was under standard costing, it’s still our most profitable product per unit, so we should keep on selling and making it. Maybe we should get a Black Belt to work on the process to increase the production rate of the line so we can make more product A and raise our revenues.

Or should we?

Well, what if that extra stress that product A causes to our machines is actually associated with more unexpected downtime? What if when switching our line over to a different product it takes longer to set it up to run product A than product B? What if we have to run the line more slowly when making product A in order to maintain quality? What if we end up making more scrap for product A than product B? What I am really asking is, “What opportunities do we lose when we run product A?”

To answer this, I need a measure that can tell me on some relative scale what I’m missing by running product A.

One method is to quantify the effectiveness of a process against an absolute, and then use relative effects of different choices to understand the opportunity cost. I have found that total asset utilization (TAU) is straightforward and helpful for this. TAU separates the effective use of time into two components—availability and duty cycle, and the effective use of the process into two components—efficiency and yield. These four factors are dimensionless ratios, so by multiplying them together you end up with a measure of how well you use the process to make something you can sell. Note that total time is 365 days a year, 24 hours a day.

Figure 1 - Example components of total asset utilization

Let’s make it simple and say we find the following:

It turns out upon investigation that everything is the same, except I have to run the process more slowly to make product A to maintain quality. Note that the profit per unit includes the cost of the extra time to run it on the line. (You can see how if product A had higher scrap, more changeover time or higher downtime, the TAU would be lower. The profit per unit would also be lower, but let’s keep it simple for now.)

Now (finally!) I have a way to compare the tradeoff between products A and B. If I choose to run the line for an hour, I can make and pass 100 units of product A or I could make and pass 200 units of product B (the TAU for A is half that of B).

Here is where it gets interesting: Which product really makes me more profit if I run it on the line for an hour?

Take a look at that for a moment.

Am I trying to tell you that you can reduce your revenue and increase your profit? Yep.

Am I seriously suggesting that rewarding your sales force based on revenue might actually decrease your profit? Yep.

Here is the key concept: all revenue isn’t good revenue. Some revenue is more profit-making than others. Some revenue even costs you money for every dollar you take.

Yet we have a “Feed the Beast” mentality: Running anything in the plant spreads my overhead across more units, and thus must be better than not running. This is driven directly by how standard costs are calculated.

What does this mean to Six Sigma?

So what would you have had the Black Belt work on in this situation? Before understanding the true cost and the TAU, we said it made sense to have them figure out how to increase the rate of high-revenue product A that you could produce on the line, so you could produce more units per hour.

Let’s take a look at that. Let’s further assume that you can increase the production rate by 10 percent with a Black Belt project and keep everything else constant, and you can choose to run either product A or product B full-time. What’s the incremental improvement to profit for the two products?

Thus, after working really hard to make a big improvement, my Black Belt could have made an improvement that increases the revenue off that line by $1,000/hour and the profit by $10/hour.

Is this a success story? In our ignorance, we would think so. We’d have a party and reward the participants with plaques and team jackets and some CEO face-time.

As my 7th-grade teacher used to say, “Your work is good, but not delicious.”

Had I understood the costs and profits, the same effort on product B could have resulted in an increase to revenue of only $500/hour, but an increase in profit of $15/hour by working on product B instead: $5 more per hour than working on our “most profitable” (at least per unit) product. For the availability and duty cycle above, that is 6,307.2 hours per year of production, and so that means if we could have run only product B, we’d have lost out on making $31,536 more by working on the wrong product.

Remember that as far as we knew at the beginning, product A was a high-revenue cash cow, and so we would prioritize it for work by the Black Belts. Now I set this up so that product A would still be profitable, because it shows how subtle, yet powerful, this effect can be, and because you would have thought I was cheating if I made it unprofitable from the start. But in the real world, we find that if you properly allocate the costs, and that if you really understand the effect to TAU, some products that are high-revenue cash cows lose you money for each unit you sell. Management isn’t stupid—they just don’t have the real information, and the true cost is hidden by using standard costs.

What would happen if product A were actually a money-loser? Let’s say that after we allocate the costs, product A actually loses us $0.05 for each unit we sell for $100.

We would use our Black Belts on the process, improve the production rate by 10 percent, increase our revenue by $1,000/hour (that’s an increase of $6,307,200 per year), lose money faster than ever and have a party to celebrate the team’s success. This doesn’t count the cost of the time the project took, the experiments, etc. That’s just frosting on this rancid cake.

So there we are, expecting lots o’ bucks to flow down to the bottom line as our sales force goes out and fills up the plant with product A running at an all-time high rate. Even with the highest revenue on record ($69,379,200), we are actually losing $3,153.60 more than last year.

And the muttering begins. “See, I told you Six Sigma was just a fad. It didn’t work here.”

The effect of SSST #8

This stupid Six Sigma trick occurs because standard cost accounting hides true costs. That and the lack of a way to measure the opportunity costs lead us to choose the wrong project. This has implications beyond project selection—from how we reward our sales force to how we price our product and choose our customers.

The only way out of this one is to allocate costs properly and to understand the tradeoffs. Sorry, I wish it were easier. This is also something that needs to come from upper management because it involves how the company measures itself. If we use standard costs, we may be hurting the company by deploying Black Belts on projects we think will result in big savings.

I made up the case study above. Want some real numbers?

On one line that ran 27 different products, we found 12 products that lost money for each unit made and together accounted for losing 23.2 percent of the profit. They lost anything from $4.02 to $76.46 per unit. On the same line, we found that seven products by themselves contributed 82.4 percent of the profit on the line, ranging from $117.81 to $242.11 per unit. The rest of the products were minimal contributors to profit on that line.

In another case, a facility ran 1,836 different product/label combinations. Fifty percent of the items generated 63 percent of the revenues and 256 percent of the profits. Guess what the other 50 percent generated: 37 percent of the revenues and a loss of 156 percent of the profits.

This works for customers as well. At the latter facility, 50 percent of the customers provided 91 percent of the revenue and 123 percent of the profits. The remaining 50 percent only added 8 percent to the revenue, but cost the company 23 percent of its profits.

This still staggers me. Maybe it will you, too.

Not all these profits can be captured—you might be stuck with some product you need to sell at a loss so you can get the other business, but for the first time you can tell if these “loss-leaders” actually are worth it. Taking these loss-leaders and making them profitable would be a good Black Belt project.

Now imagine that your competitors spend some time understanding their costs. Maybe they price themselves out of the market for their profit-losing products, focusing on their profit makers, and your hard-working salespeople pick those other products up.

Is your first thought still, “Hurray! Our revenues are up!”?

I think the next wave in business improvement will include this better understanding of the true cost of making a product or providing a service. And the ones that get there first, win.

But I could be wrong.