Inventory Management Review: Supply Chain Management

In my last article, I introduced the topic of postponement. This article will cover the subject in more detail and with less regard to software.

Your three most major buyers, Target, Office Max, and Staples account for over 40% of your entire volume. Each of these buyers requires unique packaging. The unique packaging is crucial. Staples will not purchase a unit that has the Target packaging and vice-versa. You woke up this morning only to walk into work and find out that the North American DC (distribution center), located in Detroit, Michigan has a bit of a problem regarding this quarter's orders. It turns out that Staples requested 40,000 additional units, which is 3,500 more than your safety stock of Staples SKUs (stock keeping units). Also a problem, Office Max ordered 10,900 less than expected leaving you with 30,000 additional units when safety stock is included. Target ordered 5,000 more than expected leaving you with 2,000 safety stock units for Target.

Total, you have 32,000 units from Target and Office Max sitting in the DC that you can't sell, and you have 35,000 backordered from Staples that you want to sell. Oh yeah, and to make things a little worse, the lead time from your manufacturer in Thailand is 8 weeks; 1 week of which is assembly, 1 week of which is packaging, and 6 weeks of which is transit. At this point, it doesn't really look like things are going to go too well. Too bad too, I mean you have the right product, but the packaging is all wrong. You're swimming in leftovers from Office Max and Target and you can't fill the order for Staples. Staples is unhappy, and so are their customers. Not to mention you; all of this lost business and leftover inventory all because of a packaging problem.

The DC understands the problem, but what are they going to do about it? They have 1600 other SKUs to get to 3,200 different buyers. These three SKUs only account for three 1% of company revenue and the DC is busy worrying about other issues. Sure, it would be nice if they could change packaging and send out the rest of those 32,000 units cutting the backorder down drastically. Unfortunately, the DC just doesn't have the manpower to turn the distribution center into a packaging plant. Sure would be nice if they could.

What would be the effect if they could?
The manufacturing plant is still in Thailand, so lead time would for total units would be still be 92 days. However, if they were sent over with generic packaging, then the units could be customized at the latest effective moment. Consider the benefits:

Instead of customizing SKUs for each of the three buyers 8 weeks in advance, the generically packed parts could be sent to the DC and then customized only 1 week before they need to be shipped out of the distributor. This effectively cuts buyer-specific lead time down to 1 week. This effectively cuts the lead time from 8 weeks to 1 week for the customized SKUs.

What kind of an effect does this have on safety stock?
Here, safety stock is needed due to variable demand. The formula for this is:

z*SQRT(Lead time)*standard deviation of demand.
(For help on this equation, refer to my article on safety stock)

Really, the only term in that equation that is changing is the SQRT(LT), so we can isolate this term and use it to determine the change in safety stock resulting from postponing packaging until the DC.

LT0=old lead time=8 weeks
LT1=new lead time=1 week

SQRT(LT0)=SQRT(8)=2.83

SQRT(LT1)=SQRT(1)=1

Now we can use this to determine the % decrease in safety.

%Decrease = (LT1-LT0)/(LT0)=(1-2.83)/(2.83)=-65%

This means that by postponing packaging until the DC, there is 65% decrease in safety stock that can take place.

Thinking about what I've written about in previous articles, this should be fairly clear. Safety stock is held because of statistical deviations in lead time and demand. In this case, we're mainly concerned with deviations in demand. If all three SKUs were the same for each of the buyers, then the demand would tend to even out a bit. If Staples ordered more, which they did, and someone else ordered less, then it could be evened out. With different SKUs, you lose the ability to respond to these fluctuations.

Postponing allows you to take advantage of these fluctuations because it effectively turns all of your inventory into generic inventory than can be sold anywhere.

What about the DC?
The DC probably doesn't want to start packaging products. Depending on how much you can benefit from postponement, you might want to change their mind. A 65% reduction in finished goods safety stock for 3 SKUs is fairly high, but certainly not out of the question. Granted, these benefits will be partially offset by the increased demand for WIP inventory that will be held at the DC waiting to be turned into FG, but nevertheless, there will be a decrease in inventory. This is because the pooled risk of the 3 SKUs has a smaller standard deviation on demand, which will be discussed in a future article. Also, WIP has a smaller holding cost than FG.

Back to the DC, it might be a good idea to convince them to package. Localizing the final good might not even as difficult as you think. It could be as simple as slapping a sticker on the final product before it goes out.

Whatever the changes that need to be implented are, the effects of postponement cannot be ignored. By keeping materials generic for as long as possible, inventory can be cut and sometimes this cuts can be drastic.

For ideas on how your firm can postpone production in their supply, please leave a comment with your email address, or email the author.

I recently had the pleasure of attending a guest lecture delivered by Optiant director of research, Ph.D. John Neale on the topic of multi-echelon inventory management software implementation. Optiant is a supply chain software solutions provider for companies including Gillette and HP. His discussion described how the implementation of Optiant's PowerChain® software suite helped to dramatically decrease inventory and increase service for a well-known manufacturer of consumer adhesives. This post will give a history of the adhesive manufacturer's supply chain problems and the solution they chose. The results of that solution will be explained with a brief lesson on multi-echelon inventory.

Supply Chain
The manufacturer being discussed operates primarily in North America. Their supply chain has the following characteristics:

Raw Materials
First, they procure over 3000 raw materials from multiple sources and hold the raw materials at one of their two manufacturing sites until they are processed. This is the first stage (echelon) where inventory is held. After the raw materials are turned into finished goods, they are immediately shipped to a distribution center (DC).

Finished Goods
The firm has two DCs and over 1800 SKUs (stock keeping units. This is basically how many different products and package variations they have on those products). The DCs are the final stop for finished goods inventory before they are shipped to a myriad of retailers. Of those retailers, Wal-Mart, Staples, Home Depot, and other mass market stores constituted one third of their overall volume. The DCs also receive some finished goods which come from other manufacturers in the form of finished goods. The DCs are the second stage (second echelon, thus making more than one echelon, hence the name, multi-echelon) where safety stock is held.

Old Inventory Policy
Prior to Optiant's consulting work and software implementation, the manufacturer had no real method for determining their safety stock. Essentially, they used a trial and error method where they would set a level and if they were stocking out too often, they would increase inventory. When they stopped stocking out, they would scale back inventory. Their inventory was also high because of expansion in their product line and high service levels demanded by stores like Wal-Mart.

The manufacturer lacked the expertise regarding how to set a safety stock level that optimized each local inventory stage, and additionally, they were without the experience necessary regarding how to optimize the overall system. As John Neale put it, this is a problem because while safety stock formulas can be useful for local optimization, one point he made was,

"Don't just optimize things in isolation."

Upon realizing that there were better ways of doing things, they contacted Optiant.

Optiant
What Optiant did for them was more than just selling them a software package. Optiant spent many weeks learning about their supply chain constraints and gathering data, and then used PowerChain® to optimize the safety stocks for each of the 1500 SKUs and each of their 3000 raw materials.

Data Requirements
Much of this data is demand data. In order to get a feel for demand, Optiant uses historical demand projections. In order to figure out what kind of deviation there is on these projections, they look at historical projections and compare them with historical demand realizations. As you can imagine, a lot of companies don't keep accurate records regarding this data. The less a company has in the way of records, the less effective software initially is. Keep this in mind before you bring in any consultants: start collecting data before they get there, so you're ready to roll once they're on the clock.

Optiant also required supplier data, including lead times, costs, and a bill of materials (list of parts required for each SKU).

Software
Once they have the data, they can start to use their software model. I'm not clear on the math that runs the program other than that it uses algorithms to minimize holding costs while maintaining service level requirements. I didn't bother asking for more detail, because what I understand is this: the software they have works and is based on the kind of framework that you would expect to come from someone with a Ph.D. from MIT, which is precisely what Optiant co-founder, Sean Willems, has. The point is, the program is complex, but it is not baseless, and it is not a hoax. It is exactly the kind of complex software I was referring to when I wrote about the kind of advantage that professional software can offer that Excel can't even come close to providing.

Results
Before you refuse to believe the software works without understanding every detail behind it, consider that Optiant's solution allowed this adhesive manufacturer to raise their service level while lowering safety stock value by over 20%. First of all, 20% is a very large reduction in safety stock value on its own. In addition to this, they were able to raise their service level while lowering safety stock. At first glance, this seems too good to be true. Normally, the way to raise a service level is by raising safety stock, not by lowering it. Why is this case any different?

Multi-Echelon Inventory Management
This case is different because it is a multi-echelon inventory model. What this means in this case is that they had the opportunity to hold inventory at two stages: the raw materials stage and the finished goods stages.

Balancing Raw Materials and Finished Goods
Remember, holding costs are a function that involves the value of the inventory and at the raw materials stage the value is considerably less expensive. This means that if the adhesive manufacturer has short production lead times from raw materials to finished goods, which they do, then they can afford to hold large amounts of raw materials, small amounts of finished goods, and still be in a position to meet demand. Thus, by reducing finished goods inventory and increasing raw materials inventory, they can increase service level because of their ability to quickly turn raw materials into finished goods, and they can reduce inventory costs because they are holding less finished goods.

Risk Pooling
The other reason they are able to reduce safety stock value while increasing service is because they have so many SKUs that all use the same basic raw materials. The importance here is the inherent flexibility that raw materials when they can become a variety of different finished goods. This allows them to keep materials raw for as long as possible, which reduces their vulnerability to fluctuations in demand. The vulnerability to these fluctuations is limited because many of their glue product SKUs are essentially pooled as one product with an overall demand that is less likely to fluctuate as long as products are kept as raw materials that can be turned into any product once demand projections are closer to demand realizations. To further illustrate this is an example from the MITSloan Management Review about apparel manufacturer Bennetton Group SpA and how they delay final goods production by keeping raw materials in a position ready to be turned into finished goods:

An inventory of undyed sweaters gets stockpiled in one location; coloring takes place only after specific orders have been received. This pooling of demand across geographical areas, and across colors, helps Benneton greatly reduce inventory risk while more effectively meeting customer demand.1

Another example cited in the article is how the house paint industry holds only base paints which colors are added into instead of holding onto hundreds of different colors at each retail location.

The effects of this are incredible because for the adhesive manufacturer, paint companies, and Benneton, the risk of each individual product in the product line can be vastly reduced by simply keeping finished goods as raw materials for as long as possible.

Paint companies no longer have to worry about having too much blue paint and not enough red paint. Statistically, the variations in each type of paint will even out. So if yellow doesn't sell as much as expected and green sells twice as much as expected, paint companies are still ok as long as they have the right amount of base paint. Unfortunately for the consumer this makes it difficult to return paint.

Luckily for the adhesive manufacturer, risk pooling works. So does the software Optiant creates.

Final Notes on PowerChain®
Originally, the adhesive manufacturer only hired Optiant so that Optiant could use to use their software to tell provide a report detailing how to optimize each of their 1800 SKUs and 3000 raw materials. The CFO of the adhesive firm was so impressed with the forecasting abilities of the software that he eventually invested in a license of the software. I'm not sure whether or not they worked with Optiant to adapt the PowerChain® software to their other computer programs for automated entry of optimal safety stock into their other systems, although this is something Optiant does.

I'm not trying to suggest that you dive right into the investment of such software, although I can't imagine Optiant would mind, but hopefully this post has given you a better understanding of what inventory management software packages can do for you and what the implementation process entails.

1 Sunil Chopra & ManMohan Sodhi, "Avoiding Supply Chain Breakdown", MITSloan Management Review, Fall 2004, Vol. 46, No. 1