Dear Gemba Coach,
We've done a lot of kaizen work in production and have something of a pull system running. We've started measuring on time delivery at every step of the process, and it appears that suppliers account for a large part of our instability. Should we extend the kanban to the suppliers? Is there a lean way to rank suppliers?
Suppliers typically account for 40 to 60 percent of costs in bought-out-parts and materials, so it isn't too surprising that they account for a good proportion of problems as well. While ranking suppliers may help, before looking into tools such as supplier ranking or supplier kanban, let’s try to understand the problem a bit better. For that let's go to the Gemba.
If you've got a pull system running, I’m assuming that you've installed a leveling box, shop stocks at the cells, and since you mention kanban, you're probably using withdrawal kanbans to pick up parts from the shop stock, according to the leveling board. In my experience, not a day goes by without some delivery issue, without the leveling boards filling up with red cards signifying that the desired box is missing or late. And so an analysis of the cause of lateness in delivering internally is likely to help us figure out what is going on at the supplier’s.
So what is the number one cause of late internal deliveries? Operators who have experience with the pull system will invariably say leveling. Customer demand varies daily, and no matter how hard we try, having a perfectly leveled production plan and delivering parts on time is often more of an art form than production (which is why the MRP can be such a drag if it's not used properly). Whenever a sudden order comes in, or if the customer suddenly stops picking up parts, or if they change their mind about an option from one day to the next, it’s battle station scramble to be able to get the parts out, which throws off the leveling. The upshot is that other parts then get affected (production capacity being finite), which triggers further instability unless one knows how to bring the production plan back to the right conditions.
I was talking recently to a logistics manager in Toyota who concurred that departure from the leveled plan is the number one cause of internal and external mis-shipments. Particularly if you’re using kanban to pull, the moment you depart from the strict stock replenishment sequence, the whole system will be affected, leading to low tide on some part at some point.
Before looking at supplier kanban or starting to blame suppliers, let’s go and see what kind of information we send them. For a few references, we can plot the following data for the past two or three weeks:
- Our own customers daily demand
- Our daily production plan
- What we really produced
- What we ordered to the supplier
- What they delivered
More often than not, we'll find that although we have worked hard to level our own production plan, procurement still sends supply orders on the basis of "resupply when you've hit the safety stock level" - essentially an unleveled message to the supplier. From the supplier's perspective, this amounts to receiving a series of "scoops" - you're not exactly sure when the order is going to fall, nor for how many parts.
Suppliers haven't got any reason to be particularly lean in any case, so the chances are they have not calculated an average demand, know their economic order quantities, and run the batches to fill their finished product stock (even worse if you’ve got consignment stock - no one owns it, physically, so who should look?). The upshot is that when they get a larger than usual order, they ship what they've got in stock and reprogram the rest in their normal batch and queue system - you'll get it when you get it. If you complain about this, they'll typically pull out the file of all the instances where you've sent a plan for so many pieces, but only ordered half on the day, for one reason or other.
The only way to install supplier kanban in such circumstances is to have the political clout to strong-arm your supplier into holding enough stock to cover whatever unleveled demand your production needs. This is far from the lean ideal on two counts: first, it creates unnecessary stock in the system (someone will have to pay for it), and second, such kanbans reinforce adversarial relationships rather than encouraging teamwork with the suppliers.
How shall we go about tackling this problem? He first step is getting the system under visual control at three key steps:
- The leveling of orders to suppliers
- The components stock in the warehouse
- What's in the trucks
Checking how well we level orders to suppliers is a matter of 1) sending them the leveled production plan where it concerns their parts (which involve running the Bill Of Material through the plan and sending the info part by part) and 2) tracking some measure of leveling.
In the warehouse, the visual management step involves getting all purchased components into flat storage, with a standard number of boxes corresponding to the supply frequency. With the flat storage you usually need a controlled overspill area to visualize over-delivery. More importantly, warehouse staff must be trained to "jidoka" in warehousing: they have to spot problems (typically, the dedicated stock placement is empty or overflowing), have a system to react immediately, and then look for root causes (which are more likely than not leveling issues). So this means improving the daily teamwork between procurement and the warehouse.
The third point that can be visually controlled is the pick-up truck itself. Logistics can produce a clear manifest per truck: a list of all the containers that should be in the truck. The truck driver checks this list at the supplier and if there's something wrong, he's got to call you before leaving the supplier premises - which gives an opportunity to call the supplier and discuss issues, in the spirit of teamwork.
Still, you'll argue, if suppliers are bad none of this is likely to help. Well, yes and no. First, before blaming suppliers, we can start by cleaning the window in our own operations, to have a better idea of what we do to suppliers. Second, many suppliers tend to serve best the most demanding customers (it's all a matter of waiting queue, in the end, except that no customers knows where in the file he really is), so just talking frequently to suppliers on very specific and detailed topics tends to improve service. Finally, when it comes to ranking suppliers, we're pretty sure it's about them - and not us.
The usual way to rank suppliers is to count mpms on one hand (missed deliveries per million) and ppms on the other (bad parts per million). With these two numbers you can plot suppliers on a chart and start having different types of activities to try to get them to improve. However, I'd strongly suggest doing the previous cleanup work before going to suppliers to tell them to clean up their act. I still bear the (embarrassing) scars of going to suppliers and complaining about their delivery service, and watching them pull out the thick folder and start listing all the disastrous things we’ve done to them, from late payment of invoices, over-ordering and not picking up, or not telling them of product changes early enough for them to do something about it and so on. Hey, this is lean: sweep your own porch first!