Zingerman’s Mail Order ships gift baskets full of gourmet foods all over the United States. As a very seasonal business, ZMO experiences extremely high variability in demand. Tom Root, Managing Partner, tells Roberto Priolo of how standard work helps this small organisation respond to big challenges.
Roberto Priolo: What were the factors that convinced ZMO to go down the lean route?
Tom Root: There were three contributing factors. First of all, we were growing 25-30% year on year, which had caused us to move four times in eight years. Because we were batch building gift boxes, we would spend the night shift making the assortments, storing them and picking them during the day to ship. We were running out of space to build and store the kits before they got picked the next day. We literally had boxes stacked everywhere, including above the bathrooms.
Secondly, Zingerman’s Mail Order had been an open book finance company for a couple of years already. Our financial statements are available for any employee to look at. But we go a step forward, and make employees responsible for the content. We take key operational and financial metrics and we assign ownership to those key lines. It can be anyone in the organisation: just because it says revenue, it doesn’t mean it will be a number owned by an accountant. We might assign a revenue number to somebody who is a front line employee. Being open book put us already over the bump, and led us to the understanding of one of lean’s key concepts: that there is intelligence in the workforce. The front line had shown they were able to understand financial metrics – why could they not be owners of the processes as well?
The final reason for adopting lean was, well, chance. Back then one of our managing partners was taking an MBA and read The Toyota Way. He suggested I read the book as well. I thought what it said was natural for us. I often say lean is half philosophy, half mechanics: while most organisations struggle with the philosophical implications, I figured we had that side covered. I thought that all we had to do was applying the tools – it was an oversimplification, of course.
RP: How did you get started?
TR: We contacted Professor Liker at the University of Michigan and asked him if he had anyone interested in us as a project. He had a student who was working on his dissertation on lean implementation in variable demand businesses. We were perfect for that. Can lean be adapted into a system with extreme variability?
He came in and ended up spending three years with us, revamping our production line and taking a lot of metrics along the way. We were a chapter in his dissertation.
RP: How has lean helped you?
TR: It would be easier to say how lean has not helped us, as the methodology is everything in terms of how we do things now.
The first project we took on was making the kits just in time. We set up a marketplace and routes to replenish it. We picked the kit items, sent them down the line and then we had a work cell that did the assembly just in time.
That immediately showed us we didn’t need a lot of space to build and store the kits, which immediately took off the pressure of space constraint. However, lean has to be consumed all or nothing: after those first steps, the next bottleneck showed up and we tried to address it, then the next one, then the next one, and so on. It has been 10 years of addressing the next problem.
RP: Specifically, how did lean impact on your ability to respond to the demand variability you experience every year around Christmas, when the number of employees grows tenfold?
TR: Fifty per cent of the business happens in December at ZMO. We have about 45 to 50 employees who work with us year-round, but during the holidays the number goes to 450.
Lean gave us an absolutely obsessive focus on standard work: it is the only way for us to onboard 450 people, get them trained up enough that they are effective while they are working with us.
My challenge back in the day was getting someone off the street and making them 80% effective in 30 minutes or less. The way we did it was through a careful division of labour and through a strong focus on defining the standard work.
Because part of my background is in IT, I came to Zingerman’s as chief information officer. We decided to build our own order management system. We brought its development in house, which allowed us to control how it behaved but also to integrate it with the production floor.
At that time we used to print all the work orders for the day out in one batch and we would bring a big pile of paper out to the floor. Then we went to printing them just in time – that’s how we established takt for the line. We used to pre-print labels for everything, and we changed that into printing those just in time as well.
We had the advantage of building and maintaining our own software. But you have to be able to build the system manually before you make a software out of it.
TR: Yes, but unfortunately that is not an option for many companies, particularly companies our size. We are a $11 million business. We happen to have a staff of developers because we realised this would be core to our success. Had we not had someone like me in house, who had an understanding of IT and software, we wouldn’t have gone down that route.
RP: Can you give us another example of how lean has helped you?
TR: Trying to get as many things close to when they are needed as possible has been great for us. We might have 200 of those assemblies as active products at any given time – they all have names and descriptions.
We used to have 200 pre-printed labels on rolls and there was a station that did nothing but picking the label for the kit we were making. It’s all the classic problems with inventory: the moment we wanted to change the product on the line, production told us, “We have 2,000 of those labels – are you really going to change them now? Can’t we use them for a while?”
When we moved to printing each label where it got assembled, many issues (for example picking the wrong label) went away. The driving force was putting a computer at the kitting station to print the label.
Then we found a ton of other things to do with the computer – for example, the worker can scan a barcode on the work order and on screen we display the illustrations on how to assemble that kit. This comes back to the idea of developing someone in 30 minutes or less: I cannot rely on them memorising the assembly recipes for 150-200 kits.
RP: How do you schedule what you produce and when?
TR: Takt is our “magic number”. We have a box count for the day and a set of standard takts that imply staffing. If we are going to do, say, 500 boxes today that means we are going to run on a 24-second takt; that means I need this many people here, and this many there, etc. That used to come from institutional knowledge – you would have had to work here for years to know how many people you need. Takt allows us to calculate this pretty accurately.
We also make some products that need to be handled in some way before they are shipped. Take bread, for example: it comes from the bake house unpackaged. You need to put the loaf in a bag, put a label on the bag to identify what product is in it before it is shipped. Takt allows us to do “time slices”: the bread bagging area starts one hour ahead of the line and produces in its first hour what the line will consume in its first hour.
We also have the ability to switch into a mode where we are using the order management system while responding to kanbans, a hybrid scenario: the kanban comes back asking for more products; we present the kanban to the order management system, which knows how much of that specific product we need on the day and prints out the appropriate number of labels. If the kanban calls for 10 loaves but I only need two, I get two labels. If it says 10 loaves but I need 100 more loaves it gives me ten. If the system says we don’t need any, the kanban is filed until the next cycle.
RP: What about leveling production?
TR: It’s the first thing we do, starting at the beginning of the process – picking items. Items are stored in racks near the line, which are assigned to different pick zones. Our IT system will level orders across the pick zones to make sure that each picker pick a similar amount of orders. By using this model we ensure on-time delivery from our warehouse to our customers.
RP: Is it difficult to apply lean in a small company?
TR: I would argue that our size was actually an asset. I didn’t have to get buy-in from a lot of people. That’s not to say we didn’t have our struggles in terms of adoption, but it was relatively straightforward for us to implement a lot of this because the group of people we needed to engage was small.
If you take the approach that lean is about buying software or hiring consultants, however, being small is a liability. But that’s not what lean is.
On the other side, Zingerman’s Mail Order is one of seven businesses. It was fairly easy to apply lean at ZMO, but the others are not practicing. And it’s not like we haven’t spent the last 10 years talking about how great lean is! I teach classes, we do lean simulations using Lego bricks, but it hasn’t caught on elsewhere. ZMO has also generated the largest revenue within the business for the past six years.
I would say that people like the problems they know, as they are less scary. In the classes I teach the first thing I tell people is that, if they to apply lean, they need to be okay with change at a personal level or it won’t work.
Recently, a consultant has started to advise us on our environmental performance (which is very important to Zingerman’s as a whole). He pointed out that it would an advantage to be able to produce a value stream map of the entire organisation. I am working on it nonw and I think it could represent the first step towards the adoption of lean outside of Zingerman’s Mail Order as well, even though the interest didn’t come from the desire to improve production but to become more environmentally friendly.
RP: How do you keep your relationship with suppliers working?
TR: That’s where being a small business becomes a liability with lean: we don’t have any leverage with our vendors. The way we maintain the relationship is very much in person, through a lot of contact. We try to build relationship with vendors, and that includes bringing guests over on tours of their operations around the world. We can control things once they arrive on our dock, but it’s still up to us to figure out how to coordinate the rest.
RP: How has the small size of Zingerman’s Mail Order supported the development of a culture of continuous improvement?
TR: As an organisation, even before we learned about lean, we knew about the importance of creating a culture. That knowledge was used to adopt the open finance book approach, for instance.
With lean it was about using the tools we had to create a new culture, but it’s very difficult. Even in a company our size, with 30-40 people, there a lot of individuals who do not approach their lives from a CI standpoint. Building the argument that there is a difference between knowledge work and labour is not totally straightforward, and it takes time.
RP: What makes ZMO’s lean journey unique?
TR: The high variability we face; our corporate culture, which provides for the democratisation of the running of the business; and our proximity to the University of Michigan. I have students all the time visiting us. This creates a different environment, one that encourages learning in a number of ways.
Zingerman’s is big on learning, but also on teaching. There is an expectation that every employee will find something they can teach. This is a huge component of self-development here.
There is no aspect of the business that one can’t become an expert in. You take classes on olive oil, finance, governance… and then become a teacher.
RP: I assume this contributes to the creation of a multi-skilled workforce as well.
TR: Absolutely. Cross-training at Mail Order is a necessity. One of our ways to deal with variability, for example, is “Help your neighbour”. We have a signal that highlights when a cell that is behind, and the cell upstream or downstream will send people there for a cycle to support. Training people in a number of different operations and making them as knowledgeable as possible is essential to how we manage variability.