Wow, this is now way more orders than we can immediately deal with (the experts would call this “reaching capacity limits”). So what do we do? We can mix pretty much all the dough at once, and we have access to additional delivery vans – so do we just go for it? Make lots of money? Buy the shiny, or get tickets to that thing we like?
From a management perspective, even though we know the oven is limited, we should aim to minimize downtime between bakes… right? We were often told at business school that switching costs can be reduced by larger batch sizes…and then of course there are those economies of scale and economies of scope. But what does that mean? If we know the solution, why don’t we just do that?
Well because we’re using a bakery to explain this, you already know that the oven limits what is possible. But for arguments sake, even if the dough wouldn’t turn bad just sitting around waiting to be used aaaaand even iff we had a quantum oven with infinite space, our service quality would deteriorate because we still only have the same number of bakers.
On the flip side, we don’t actually want to be processing one order at a time either, we want to be making the most of our available capacity, because if we processed each order in series, then that leads (ha!) to longer wait times.
In this (theoretical) state where we can do everything at once, our customers would still have to wait a very, very long time for their orders to arrive. Check this illustration (kudos to Henrik Kniberg for the idea), comparing the leadtimes for the preparation of smiley donuts:
| Donut | Start Time | End Time | Leadtime Increase |
|---|---|---|---|
| Red Smiley | 1 | 3 | |
| …with batchsize = 3 | 1 | 7 | +133% |
| Green Smiley | 1 | 6 | |
| …with batchsize = 3 | 1 | 8 | +33% |
| Blue Smiley | 1 | 9 | |
| …with batchsize = 3 | 1 | 9 | – |
So if we went down the suicide large-batch route then the bakery would basically be locked down while we work on the big batches. Imagine, when we’re baking, our mixers sit idle and our delivery van just rusts on the drive. Betty would be trapped in her own special hell of an eternal loop of full speed and idling. There must be a better way…
The KanBun Flow
It’s at this point where people sometimes freak out and start “getting involved” to help solve the issue where we’re facing impossible odds, only by the grace of their experience and charisma can things be survived.
How about we just “make it happen” and juggle the orders? Is multitasking a good thing? Ehhhhh, no. It is definitely not, as I described here (it also massively increases lead times!). Don’t do it.
Customer orders need to flow through the work process, which means flowing across the stages of the Kanban board. In fact there’s even a concept called the “one piece flow”, and it’s something like the holy grail in manufacturing – although that might be a bit much for Betty’s needs. So rather than starting to work on all orders at once or multitask, maybe it’s better to work only on the high priority orders?
Much better, but that means that we need to figure out how to identify the priority of orders so that we can keep both productivity and customer satisfaction high. So there are still capacity limitations which we can write next to the column name. Since we only have only one oven, which constrains overall output – that’s the bottom line that KanBun has to work with (remember that in our simplified example, all orders are of the same size and complexity!)
Kanban is all about minimizing the Work In Progress (WIP), so plan your capacity and commitment around what is possible to deliver.
The more work in progress (WIP) we have, the longer our clients have to wait, and the lower my ability to respond to change. Just imagine we just prepared a lot of dough “just in case”, and then the customers change their mind. Or taste. So we need to reduce our work in progress as best as we can (for those interested, you might look up the Goldratt’s Theory of Constraints, queueing theory and especially Little’s Law).
