Any advice for how to implement real pull across the company?
Dear Gemba Coach,
My boss, the operations VP, has asked me to implement a pull system across the whole company. I’ve practiced lean for many years and am a division lean officer, but not fully comfortable about real pull – would you have any advice?
This is great! A real step forward! Many of the lean tools simply don’t make sense without pull. For instance, why would operators participate in kaizen if whatever improvement they come up with contributes to filling a pallet to push to the next step? Pull brings the customer’s sales pace into the factory, at every cell, so that each team’s work makes sense: rather than filling up pallets as one would a coal heap, an internal customer comes every hour (every 20 minutes if you can) and picks up what is needed to fill the truck, which will deliver it to the final customer.
This constant, steady demand creates a concrete meaning for team members, who now see they must have the right products available every hour, and can start thinking about new ways to make their work better, both in terms of quality (eliminate rework), fluidity (eliminate obstacles), and flexibility (build more products in the same cell).
How to go about it?
Before you even get started, consider two points (that I have learned the hard way):
- You can’t do it yourself. I’m serious – you can’t implement pull. Pull is a completely different way of running the factory, so each plant manager has to learn pull by themselves. You can support them, of course, but it has to be very clear this is a skill they have to learn.
- Pull can’t be implemented. Pull is not a system you can design and implement for the plant managers to run – it’s a way to run operations. Pull is a learnable skill, but it is a know-how. To work, it needs to be understood at:
- Why -- Bring the customer into the plant for work to make sense for operators and reduce the inventory to make interesting VA problems appear.
- How -- Simulate customer consumption on the line for each cell to visualize real demand for operators, but level it to create a “perfect customer” and give them some stability.
- What -- The kanban cards, leveling board, small train, and shop stock paraphernalia levels; plant managers need to learn this. Every attempt I’ve witnessed at defining the system on paper and then driving the implementation has failed. Buyer beware.
5 Must-Have Pull Skills
At the gemba, what are the key skills plant managers should acquire to pull their production? Be warned: acquiring each of these skills is somewhat like learning to ride a bike. At first, they will fall. Your job as a lean officer is not to define the perfect way to ride the bike but to help them up the learning curve, which is mostly about helping them pick themselves up from the floor when they’ve fallen off and getting them up on the bike again – until they ride it.
The first skill is leveling. Pull is very demanding on production cells, because they can no longer interpret the production plan to produce what they can in the time it takes, but need to be ready to fulfill precise demand on an hourly basis. To make it easier for the cell, we have to “level” this demand, to make it stable enough so the team can learn to cope.
In practice this means listing all products produced on one cell, identifying the 10% of SKU references that make 50% of total demand and focus the pull on these (the rest will continue to be treated on demand).
These few references (rarely more than 2 or 3 by line), let’s call them A, B, and C, can be planned for at least a week in a way that the pull is regular every day. This is where we cheat with real customer demand by leveling it (what they don’t take today, they’ll take tomorrow, and vice-versa) by holding inventory in logistics to make it easier on the cell to deliver regularly.
There is no right way to do a leveled production plan. You’ve got to stick to putting one together with sales and logistics once a week and get better at it.
The second skill is to learn to tightly control the truck preparation area by dedicating zones to “false trucks” – areas in logistics, which represent a truckload and will be consolidated hourly from the cells. This involves having a clear manifest for each truck and setting up a “small train” [tugger] that picks up hourly from the cells to consolidate the false truck in a leveled way to reach the manifest at the end of the shift. In the beginning, every aspect of this is problematic since in the early stages the plant manager will need to devote more resources, energy and time on this that he or she usually thinks necessary (like having training wheels to help ride the bike at first).
Again, there is no other way of learning how to do this than doing it. Badly at first, then better and better.
The third skill is to create a “shop stock” for each cell, where teams are responsible for their finished product, like the counter in a bakery. As, Bs, Cs are lined up there for logistics to pick up according to their shopping list (which is, hopefully, leveled). Xs are displayed when they are produced according to plan and picked up by logistics when they are ready, this would be like the special order cake at the bakery.
With this shop stock, the plant manager needs to learn how to teach the teams to run a team-based internal kanban in order to replenish the shop stock to fulfill customer (logistics) demand. This is not easy because, typically, cells produce in batches. Imagine that every hour, according to takt time, logistics picks up 3As, 2Bs, and 1C. But the team only knows how to run batches of 10 As, 10 Bs, and 10Cs – havoc.
This brings us to the fourth skill, surprise, surprise: SMED (single minute exchange of die). Plant managers need to learn to help their teams be more comfortable with changeovers to reduce their own batches in order to get closer and closer to the logistics pull.
This is not as easy as it sounds, first because machines or semi-automatic lines are notoriously not designed for flexibility and secondly because it reveals a lot of interesting quality issues.
Imagine that we’re doing batches of 10 As, and one of these products has something wrong, which needs to be reworked. As long as we’re pushing, the batch of 10 will be ready in the time it takes to make 11 since we reworked one, no big deal.
But imagine we’re now doing batches of 5 As and logistic is picking 2 every hour. One wrong A means missing a customer pull and not being able to deliver – it’s a disaster. Which creates much higher pressure to actually get all the As right, which will reveal deeper quality issues and lead the team (and the supporting engineers) to solve hidden problems with machines, components, training and so on, and ultimately improve product quality for the end user.
It is a virtuous circle but needs a lot of work.
The fifth skill is creating a chain of help to support the team members in dealing with the tension of pull. Essentially this means setting up a whiteboard in each cell for the team leader to track hourly production (and the estimated time for change over) and write down all the obstacles encountered to hitting the planned output. This is NOT about monitoring the cells more closely. This is a tool to help the cell become more autonomous in planning how it intends to work in order to deliver to the hourly pick-up from its internal customer.
Radical Mindset Change
Cells will face immense difficulties, most of them not of their making or out of their control, such as engineering or supply problems. This means the plant manager will learn to create the right support structures for the cell to be able to respond: who is the cell’s manufacturing engineering go-to guy? Who is the cell’s supply chain correspondent? How do we learn to better help the cell respond to pull?
As you can see, the five skills of this skill-set can’t be implemented as a production system. Each plant manager will have to climb his or her own learning curve in their own ways. These foundational skills are the key to creating just-in-time conditions and establishing the roots to real, meaningful kaizen from the team as they change how they work in order to better respond to the leveled pull.
Which brings us to your original question. Your operations VP is probably not expecting the kind of industrial challenges “implementing” pull will appear. Yes, this leads to better customer satisfaction, both in terms of quality and delivery. Yes, it leads to better load for the plant, and lower overall industrial costs. But it also means a radical change of mind about the industrial strategy, from optimizing industrial resources to learning to flex these resources for greater adaptability.
Which brings us to the key skill you need to learn as lean officer – helping the plant managers to see the inflexibilities in their operations’ flows. The tool for this is the material and information flow analysis (MIFA) as opposed to the usual value-stream mapping – but this is another story :^)
To sum up, my suggestion would be to put on hold any plans to “implement” pull, but start working immediately on what would be a development plan for each of your plant managers to learn to pull their own operations – which probably also means to sketch your own self-development plan to help them climb those learning curves, and expect to have to pick them up when they fall and get them to ride that bike again until they get the knack of it. Pull is a skill, not a procedure.
Seeing the Work of a Daily Management System
Daily management systems tap visual elements that expose problems, and also use obeyas as thinking spaces for reflecting on broader challenges, says Michael Balle.
What's wrong with taking a tools approach to lean management?
Dear Gemba Coach,
I keep hearing that a tool approach to lean is wrong, but tools deliver results – how can that be wrong?
How effective is a book club?
How effective is a book club?
It really depends where you want to take your lean journey. What are you trying to achieve? A core part of kaizen is not just looking for the results of the physical improvement, but the opportunity to grow, develop people – or self-develop, so that the learning can be applied to every other work situation.