The Origins of Category Management

I was fortunately enough to have a fascinating discussion with Steve Zimmer, who currently leads a team at US Car.  For those who may have been around for a few years, Steve was the architect of the Extended Enterprise concept pioneered by Chrysler during the years of 1989-1998.  The origins of the Extended Enterprise went before this period, however, and began with the early work that Steve did while he was a senior purchasing buyer at the Ford Motor Company.  (I was a young buck assistant professor at Michigan State University’s Eli Broad Graduate School of Management at the time, working with Bob Monczka….Steve spoke several times at the Executive Program held at MSU, and did a great job as I recall…)

Steve described his experience this afternoon over pizza and sodas in Detroit.  “When I first started at Ford in the early 80s I was working with Jack Hughes on plastic parts.  I was involved in inventory control area on what they used to call “residual parts”, which were leftover parts that you ended up trying to sell into the service parts organization.  They had all these computer systems to help analysts try to figure out what to do with it.  This was my first exposure to the part number mentality with a bunch of analysts.  Ford had a lot of rejects and I worked with a guy who had programmed it for years, and I realized how crazy the part number system was.  So  I flowcharted the whole thing out and found that there were loops in the system that didn’t work and made changes.  Unfortunately I eliminated several people’s jobs in the process!  But they didn’t know what to do about it at Ford.  So I started to do programs on how to handle it, and in 3 months I was laid off.

Later I was rehired and brought back into purchasing.  Jack Hughes sought me out because during this period in the 1970’s, price controls were coming in off of the extreme shift in the cost of oil America was experiencing.  Suddenly, overnight Ford Motor was exposed to over $400M of economic exposure for pricing.  This impacted transportation costs, resins, chemical feedstocks, and a whole bunch of other areas.  Buyers were being told to just make this cost increase go away, but of course didn’t have to tools to do so.  So I started to try to project the economics around these cost increases, using some of the ideas I had developed working in the part number space.

The first thing I did was I interviewed buyers and tried to get an idea of which parts were made of what material, and the relative level of exposure they had to oil prices.  I learned quickly that  it wasn’t all that accurate, and the information was highly dependent on the buyer.  So then I got the forecast of material projections and labor and did some projections.  I looked at exposure vs. budget, and most of it was on an annual basis.  I figured out you could meet exposure by either resisting or deferring it, and was able to come up with a way to put a value on deferment.  This allowed buyers to use some combination of deferral and resistance that would equal the same outcome.  For instance, you could do a 50 day resistance combined with a 90 day delay, and it would get you there.  I took this and then developed a “must pay list” for every buyer of material.  The thought I was crazy, but I assured them that this would work!

But understanding exposure wasn’t enough, as my boss came to me and said we are under attack from petrochemicals, and we have to deal with it.  What is the leverage available?  So  I talked to a few people, and tried to understand where the raw material was coming from in the third or fourth tier of suppliers.  I needed to find out the types of plastics we were buying, and I couldn’t get that information – but we needed it!  So I sat down with all of the Ford parts and ran programs on them.  This was awful – you had to use punch cards to find out what the parts were made out of and where we were buying stuff.   I had boxes of punch cards!  But I was able to map out the value chain back to petrochemicals, and was able to estimate weights which were then pushed back onto part numbers. I recombined the data and understood where they were getting the parts.  My boss  Jack said we need to get contractual control over our supply chain, so we started to leverage out buys at the petrochemical level, and started writing contracts with the big petrochemical companies.  Eventually we did the same thing with zinc and steel.  Later when I worked at Chrysler, we were the first automotive company to leverage the steel buy.  Chrysler didn’t have the leverage that Ford or GM did –and so they leveraged it and got them to drop ship steel directly into the suppliers.

What happened then was that this whole exercise got me thinking about commodities, NOT parts.  And Ford was spinning off a new heavy truck department, with a new 9000 truck.  For the buyers who bought parts this new low volume truck would have more exposure than other products out there.  But how could I get all the parts and consolidate them in a way that made sense to measure exposure?  The only way I knew how to do this was to dive down into the Bill of Materials – and do cross-tabs.  Fortunately, Ford was one of the few car companies at this time that had a decent commodity coding system for each part number with a prefix that described the car the part was attached to, the function of the item and the suffix which described engineering levels, color, and material to a certain degree.  This was the first intelligent database system I had seen in a long time, and it allowed me to do a few things.

What I would do is I would start with the commodity description that the part code was associated with.  Say it was transmission, followed by assembly, etc.  I blew out all the part codes out into an alphabetical listing and arbitrarily adopted a simple language to group the parts.  If the part code was trans-assy – that meant it was a transmission assembly, and belonged in the transmissions category.  Although   there were thousands of part codes, I was able to work it down to 200 specific groupings of parts.  And then I compared this grouping to a list that should the buyers for each part.   This allowed me to see if they were assigned to the right people in a way that made sense, and in some cases, I was able to re-categorize the buys into a logical grouping by buyer.  I had to have a conversation with the buyers to let them know that “I’ve mis-categorized your buys, and I’m going to add some parts to your list of managed parts”, and was able to get by with it.

So once I did this, I was able to take a look at a “deck” of the current buy for a commodity group for Ford’s production buy.  I started by looking at no more than 10 commodity groups, and look at the production buy across these commodity groups.  With the data organized this way, I was able to see that I could consolidate the number of suppliers by commodity, the value of the buy for the group, using data that no one had ever had before.  Then I created a matrix of around 35 existing buyers,  pulled all the heavy truck stuff out of it, and ran summaries  of the data.  I could then start to seehow many commodities I was dealing with, and was able to reload the commodities to a smaller group of buyers with broader responsibilities that might cover more than one commodity.  This was tricky, as I had to consider the workload.  For example, just the two groups of engines and transmissions might be too much for one buyer, so it had to be measured in terms fo the complexity of the commodity group.

The other big advantage to working with commodity groups was how to transfer ownership of one buyer’s activities when he left for a new role to a new buyer.  Historically if a buyer gives responsibility of a part to another, we had to fill out forms and transfer all the parts over one by one.  With the new system, transferring to a new buyer was easier because we knew what commodity group all the parts were attached to.

Later at Chrysler we were able to do the same thing, to drive strategic alignment by commodity code by buyer, and identify had the highest value of business.  But this was a lot more complicated, because the data cleansing at Chrysler was much more laborious.   We were getting MDM codes where a name of the guy was the part descriptor!  At one point with an SAP implementation we had 12,000 non-production categories many of which were duplicates of other MDM numbers.

What this story shows was that prior to launching any type of SRM initiative, there is a need to really understand the discipline of what you are doing.  The discipline involves beginning to understand and build better data on what suppliers are doing, so that you can establish the right type of business driver you are asking them to deliver to!  The other important lesson here was that conducting category analysis studies helps a team to discover the cost drivers, and begin to establish ways to taking out cost that cross multiple part numbers and purchases.  For stampings, Ford used to get quotes on one part at a time, with no concept of “press loading”.  They would bid and source only part at a time!  In such cases, the material cost doesn’t change, so the only advantage one can derive is to load the presses at a given supplier efficiently to drive the best productivity and reduce setup costs.  This analysis also helps you to understand the level of supplier risk, and if you were giving too much business to one supplier given their capacity load.  These types of thoughts on cost drivers and risk didn’t come from just the buying activity, differentiated the strategy and established the optimal approach to buying. .

An important moment came when I asked Steve about the importance of trust in supplier relationships.  He thought a moment, then responded:  “ I don’t like to use the word trust – I prefer the word “Expectation”.  Trust comes and goes, and there is a lot of variation in the way trust evolves across cultures.  In the US people will trust you early but be easily disappointed and not come back.  In Europe they may not trust you at all, and only much later begin to approach something like trust.  In other cultures in Asia, trust may never really occur.  By really understanding what it is you are buying, you are able to set the expectation of the supplier that drives the relationship.  Trust is a function of people meeting expectations which in turn builds the relationship, not the other way around.  You can look back at supplier relationships and you trust those that met the expectation and the commitment.