Throughput Accounting: Bottleneck Management (#44)
/In this podcast, we cover how to spot a bottleneck, as well as how to manage it to maximize throughput. Key points made are noted below.
Locating the Bottleneck
The first step in managing the bottleneck is to figure out where it is. In a complex environment, it can actually be difficult to find. Here are some indicators.
First, figure out where there’s a work backlog. This should be an area that absolutely never catches up with demand. If it involves a manufacturing process, then there should be a pile of inventory in front of it. But don’t think that this is an absolute way to find the bottleneck, because it could just mean that the preceding work station is super efficient, and is dumping all kinds of work on the next work station in line – which doesn’t necessarily make it the bottleneck.
Another bottleneck indicator is that the maintenance staff is always hovering around a work center. This is because it’s being operated at way beyond its normal run rate, so the equipment tends to break down more often. Or, if the bottleneck is a person, such as a salesperson, look for people who never have any time to take vacation or participate in meetings outside their areas.
And the ultimate indicator of a bottleneck is when it stops working for a while, and company profits sink like a stone.
It’s also possible to deliberately designate a work center as the bottleneck, even if it currently is not the bottleneck. You might want to do this if the work center requires really expensive equipment, or highly paid staff. So, you cut back on the investment or headcount in that area, and now it’s your bottleneck.
Managing the Bottleneck
So, let’s assume that you’ve targeted the bottleneck. So.. what do you do with it? The main point to remember is that the bottleneck should operate at all times.
With this goal in mind, you can make sure that it operates around the clock, and that all employee breaks are covered by replacement personnel. Or, if there has to be downtime for machine maintenance, then schedule it during an employee break period, so that you can overlap two potential downtime periods.
Also, there’s usually some downtime when there’s a shift changeover, so schedule the incoming staff to arrive a little early, so that you have two production crews in the work center at the same time. With the scheduled overlap, there’s no reason for changeover downtime.
This next recommendation is a big one, because it applies everywhere, not just in the production area. If there’re employees involved in the bottleneck, make absolutely certainly that they’re doing nothing but running that operation. They should not be doing maintenance, or cleanup, or filling out time reports, or anything.
This is really useful for bottlenecks located elsewhere in the company, such as salespeople or engineers, because they should have administrative support who handle absolutely everything for them. That means the admin schedules meetings for the bottleneck person, answers voice mails and e-mails, brings in lunch, whatever it takes. The point is to strip away absolutely every activity that keeps that bottleneck person from processing work.
Another way to improve the bottleneck is to review incoming materials for quality issues just in front of the bottleneck. This strips away any materials that would otherwise be processed through the bottleneck and then eventually be thrown away anyways.
Also, be sure to reduce employee turnover at the bottleneck operation. If it’s a boring operation, then raise wages to make sure that people show up for work. Better yet, cross-train a number of people, so that you can easily drop other employees into the bottleneck on short notice if you need to.
And another point – if you have an under-utilized person sitting around in the bottleneck operation, and the cost accountants tell you they’re costing too much money, tell them to go pound sand. Chances are, any part-timer in the bottleneck area is still creating so many incremental throughput dollars that it would be foolish to pull them away from the operation.
And I’m not done yet. You can also off-load excess work from the bottleneck to a backup operation within the company, even if the backup is way less efficient than the main operation. This is a key point, because everyone focuses on the efficiency of an operation, and not its throughput, so they assume an inefficient operation loses money. Not true. In most cases, you can break loose a few employees from other operations to work on a backup work center – and since you were paying them already anyways, there’s no incremental cost to having that extra work center. So, even though it may appear inefficient, it can create a lot of extra throughput dollars.
This also says something about keeping old equipment lying around. If you can put an old clunker to work as a bottleneck backup, then do it – don’t scrap the equipment.
Another option is to outsource the work. As long as the throughput generated from outsourced work exceeds the incremental outsourcing cost, then this is a viable option.
You can also improve the bottleneck just by changing work rules. For example, the upstream work station might have a batch sizing rule which says that it must fill up a crate with parts before sending it on to the bottleneck, where they may already have run out of work. Instead, install a conveyor belt between the two operations, and send parts over one at a time. Or, cut the size of the crate in half. Just do whatever it takes to reduce the artificial buildup of inventory before it moves to the bottleneck.
Another work rule may be to never pay for rush delivery of parts. Tough. Always pay for a rush delivery, because the cost of the delivery is usually way lower than the throughput you’d otherwise lose by shutting down the bottleneck. And the same goes for overtime. If it takes overtime pay to run the bottleneck, then pay the overtime.
You can also focus on the operations in front of the bottleneck. For example, if the upstream workstation sometimes falls behind and doesn’t deliver inventory to the bottleneck in time, then invest in an extra upstream work center, so that you have tons of sprint capacity. And once you have it, run it like crazy for a few days, so you build up a nice inventory buffer just in front of the bottleneck operation.
That way, even if all of the operations feeding the bottleneck crash and burn for a while, the bottleneck operation can keep chugging along, feeding off that buffer.
Obviously, there are a lot of ways to manage a bottleneck. This also means that, as a manager, you need to spend a really large chunk of your time reviewing the bottleneck operation over and over again. If you find that you’re spending most of your time monitoring non-essential operations, then you need to re-evaluate your priorities – because you’re not improving company throughput.