leftovers - the medusa and the snail (business bro musings)

This book describes the Delphi method, which produces group opinions by collecting simultaneous rather than sequential responses to a prompt, event, or meeting. These are circulated and revised until the group reaches a consensus; the method restrains the mimicry and groupthink that can undermine open discussions. I realize in hindsight that I partially incorporated this method while collecting feedback as a hiring manager, but oddly I haven't encountered it in other contexts.

Thomas also notes that lowering costs in medicine means attacking the underlying mechanism of disease at the earliest possible stage, which reminded me of something I remember from The Goal - inventory should be kept at the lowest cost stage of the process. The way he says that most major diseases hinge on a single key mechanism also invoked Eliyahu Goldratt's business bro classic for the way it mirrored The Goal's analysis of bottlenecks.

leftovers - the goal (sapporo ramen, part one - bottlenecks)

I mentioned in this post that suggesting seating capacity was a bottleneck for Sapporo Ramen defeated the unstated assumption of The Goal. This is mostly because, in theory, increasing capacity is always a solution to a bottleneck problem - not enough production capacity, just add more machines, right? Let’s challenge that assumption today and consider the most appropriate way to think about capacity in the context of bottleneck resources.

A subtle lesson in The Goal is how primary bottlenecks can remain hidden when constrained by other factors in the system. The seating capacity at any restaurant is one possible example, the inability to serve enough customers possibly obscuring a kitchen's limitations. So, although adding two hundred tables might enable any restaurant to eliminate lost revenue due to waiting customers, at some point the kitchen won't be able to keep up with the new orders. Sapporo Ramen's bottleneck is never really going to be its capacity until seating is decoupled from production capability. For most traditional restaurants, such a feat is all but impossible.

A restaurant could overcome the kitchen limitation by increasing production efficiency. Sapporo Ramen could add seats anytime it increased the rate of ramen production. This is why I identified ramen production as the bottleneck of the organization, not capacity, because one follows the other. Restaurants can raise revenue in these situations by funneling production output into takeout and delivery services, often far cheaper infrastructure investments than adding seats. The restaurant could then examine the performance and decide if adding seats would lead to even better results. My guess is that one vital consideration here would be profitable side items - for example, most restaurants cannot include alcohol in its takeout orders. If the potential of these items was significant, I suppose the restaurant might conclude that bringing customers onto its premises would result in higher profits (1).

Does this thought apply in general? It only applies if physical space is related to production. A restaurant works within these limits by linking kitchen size to seating capacity. When capacity is increased, daily costs such as rent, wages, and depreciation all rise and therefore a restaurant must take in more revenue than it did at a smaller capacity. The first step here is to make it possible to increase revenue, and that can't happen unless a portion of the increased capacity is in the kitchen. Organizations that do not follow this rule might seem to have an advantage, but usually this is true of its competitors. The thought might not apply in general, but if it doesn't then its likely in the cases when capacity isn't a relevant concern.

Footnotes / leftovers...

1. Restaurant geometry

The layout of most seating areas suggests that cost per customer will rise at an ever-growing rate as capacity increases. Let’s say your current restaurant serves four people, all seated around one table. If you wanted to quadruple the capacity and you took into account factors like allowing walking space between the tables or ensuring room for people to sit and stand, you would definitely need to more than quadruple the space. The only counter is to line up four tables to make one long table seating eight per side - does that sound like any restaurant you know? I suspect geometry is why I see restaurants open new locations much more often that I do see them expand the size of their existing premises - multiple small locations simply cost less per customer when compared to one giant venue.

reading review - the goal (sapporo ramen, part one - bottlenecks)

Hi all,

Let's resume our recent examination of The Goal, Eliyahu Goldratt’s 1984 Business Bro classic. I think the best approach for the remaining ideas is to put them in the context of real-life examples. After almost no debate, I chose TOA favorite and Porter Square institution Sapporo Ramen as my example. Of course, I know nothing specific about the restaurant's operations, but I'm comfortable speaking knowledgeably from a foundation of best guesses, wild assumptions, and hunches about ramen lunches.

An organization should always have a simple answer to the question of what it is doing.

Easy – Sapporo Ramen serves bowls of ramen.

One way to start looking for a bottleneck is to learn which parts tend to always be in shortage. Whatever operation is responsible for the production is likely unable to keep up with demand.

There are two possibilities. The obvious bottleneck is seating (well, obvious if you've had to wait an hour for a seat). Restaurant capacity is somewhere between fifteen and twenty people, all in full sight of the five to ten waiting customers. In theory, adding seats would address the shortage. However, expanding capacity must be complex - leases, zoning, buying chairs - all that admin! It might be better to just work with seating as a constant for the purposes of this post. (Plus, there is an unstated assumption in The Goal that ‘bottlenecks’ refer to resource allocation.)

This leaves us with the ramen, a dish with several ingredients – noodles, broth, toppings, etc. There is also equipment (such as the bowl or the chopsticks). I bet only the broth requires any prep work - everything else is likely purchased Sapporo-ready. Therefore, if demand increased, the broth would fall short before any other ramen component.

There should never be idle time for a bottleneck because any lost work through this station is lost forever to the organization.

Sapporo Ramen optimizes revenue if all operational decisions first consider maximal ramen output. This means tools, resources, and staff do other tasks after maximizing ramen production. If a customer waits too long for a bowl, it represents a reduction in the organization’s maximum possible revenue that day.

Quality measurements on bottleneck parts should take place before the bottleneck step because a scrapped part pre-bottleneck is a part lost while anything scrapped post-bottleneck is a cost to the entire system.

There is no time for thorough post-ramen QA so Sapporo Ramen must cull bad ingredients before the chef turns on the stove. The broth presents a challenge because failed broth post-production could dramatically lower the revenue capacity for the next day. (It's possible for inventory to cover this loss but days-old broth is no ideal.) If I were seeking evidence of a robust QA process, I would expect the restaurant's biggest waste item to be discarded broth ingredients prior to preparation. This would mean Sapporo Ramen was as close to 100% certain as possible to producing broth in lockstep with the next day's expected demand.

Bottlenecks waste time if they remain idle, work on defective parts, or produce parts not within the current demand.

Overproduction is a subtle hint of a wasteful process. Ramen sitting on the counter is no good – waterlogged noodles, lukewarm broth, and sinking toppings attract tourists, not customers. Ramen containing 'defective parts' likewise disgusts diners.

Bottlenecks must be protected from supply shocks. Otherwise, there is a significant risk of reduced flow leading to the organization operating below potential.

The ideal inventory level allows the restaurant to meet demand without reducing quality. Storage is a tempting cause for next week's problem. I wonder if Sapporo Ramen has agreements with ingredient vendors, perhaps to pay a premium in exchange for being front of the line whenever sudden demand surges require an emergency infusion of backup seaweed.

The daily supply threat is broth. Sapporo Ramen's main focus should be on ensuring sufficient broth inventory to meet the highest possible demand level. All other ingredients could probably be bought at the local Star Market if needed but there is no equivalent plan for a broth shortage.

If the market is not a constraint, measure productivity and structure decisions based on what maximizes utilization of bottleneck resources. If the market is the constraint, use sales less materials divided by hours consumed.

If Sapporo Ramen sells what it makes, it must maximize ramen production. Otherwise, it must produce within the constraint of its ability to carry inventory costs for whatever goes unsold during business hours.

Someone working a non-bottleneck by definition has excess capacity. The utilization level of a non-bottleneck is determined not by its operating rules but by another constraint in the system.

Restaurant workers always move at top speed, implying a ceaseless contribution to the bottom line. However, not all activity generates value. At Sapporo Ramen, the general job description should read 'ensure order rate matches production rate' and the best employees would be those who contribute the most toward this ideal.

If Sapporo Ramen can produce twenty bowls per hour, the best employees would either generate a new order every three minutes or find ways to increase the production rate. This knowledge should help employees make otherwise arbitrary decisions. When the chef is overwhelmed, a server might top off water glasses to slow the order rate. On the other hand, during a slow period a server might seat a new customer, ignoring empty water glasses along the way, because new customers are more likely to place orders.

Saving time on setup at a non-bottleneck is an illusory way to save costs. By definition, non-bottlenecks have excess capacity.

The result of everyone wanting to optimize his or her own work is chaos. Local optimization must come as a secondary priority to system optimization.

Individual efficiency works unless it slows ramen production. Let's use dish washing as an example. If it takes one minute to wash one bowl by hand and ten minutes to wash one hundred bowls in the dishwasher, then it's obviously more efficient to use the dishwasher - the machine is ten times faster!

Well, not so fast, first this is only true when the dishwasher is full. If the dishwasher is half-full, an employee might wait until it's full before running the dishwasher to ensure maximum efficiency. But what if you need a bowl and no spares are to be found? It's time to wash by hand, inefficient like it's 1985, but if it frees up production capacity then it's the most valuable task.

leftovers - ask the business bro (the goal, riffoffs)

Good morning,

As I promised TOA at the end of the ‘Ask The Business Bro’ series about The Goal, here are a few leftover rules of thumb about monitoring the performance of an organization.

Temporary reduction of released material reveals obvious choke points – those whose inventories remain the longest.

We’ll start with the topic from last time’s discussion – how to identify a flow problem. We discussed the theory behind reducing the release of raw material using the river analogy but this thought gets more directly at how the method will work in practice. If a production process has five distinct steps, broadly speaking the choke point is whichever step takes the longest to complete one unit of work.

Longer leads times automatically increase inventory.

This brings a different perspective to the idea that balanced flow comes from a combination of inventory and excess production capacity. If an organization maintains a long lead time, it implies that it requires more time to complete a certain unit of production. This means anything in the production process remains in progress longer than would be the case if the lead time were shorter. The reliable way an organization can reduce lead time is by increasing its excess production capacity.

Instability results from three basic effects. If the product life is short, overproduction might lead to irrelevant inventories and demand is left unsatisfied for a proportionally longer period of a product’s life. If demand is volatile, inventory stock will move in synch with the threat of shortages. If production load fluctuates, due date performance is the first to suffer.

A few weeks ago, we talked about negative fluctuations and how they tend to accumulate over time rather than even out with positive fluctuations. In general, having an appropriate inventory level is a good solution. However, if the organization operates within the instable conditions defined in this thought, it is advisable to opt for greater production capacity ahead of holding inventory.

Setup, process, queue, and wait are the components of how long a part remains in the system. Bottleneck parts spend most of their time in queue (often waiting for another bottleneck part to process) while non-bottleneck parts spend most of their time in wait (often for the bottleneck to finish processing). Bottlenecks dictate inventory levels through this mechanism.

This thought gets into the basic math involved in calculating the system’s performance and highlights the importance of subordinating the measurements to the bottleneck’s capabilities. If the organization is capable of processing one hundred units a day through the bottleneck, the other measurements must work in tandem to have one hundred units ready for processing by the bottleneck each day.

Saving time on setup at a non-bottleneck is an illusory way to save costs. By definition, non-bottlenecks have excess capacity.

This comment extends the above (and also references the most recent post). Let’s suppose we have a simple operation where there are two teams, preparation and production. The first team, preparation, prepares units for the bottleneck and the second team processes these materials through the bottleneck. We can put one hundred units a day through the bottleneck. The first team considers preparation of one hundred units per day as a normal workload.

One day, the team discovers that through a series of efficiencies they are able to prepare two hundred units. Overall, the total cost rises 50%, but on a per-unit basis they save 25% in production cost.

Good, right?

No.

Imagine walking into a pizza restaurant and seeing piles of uncooked pies, stacked all the way to the ceiling – that’s what will happen here. It doesn't matter that each pie was cheaper to produce because the extra parts will simply create added inventory cost. It’s always important to find ways to work smarter but optimize too much on a local level and the organization is harmed because it cannot turn the increased cost of holding inventory into added revenue.

Parts should only be prioritized if there is a shortage downstream to account for. A buffer system helps preemptively identify shortages.

A buffer system is a complicated way of saying that the next batch of work should always arrive at a workstation before the current batch is complete. This way, there is no lost time in the transition from one unit of work to the next.
 
Using a buffer system means many items will finish prior to the due date. If the release date is tracked, a priority system emerges where the oldest parts are worked on first.

The danger of the buffer setup is prioritization. If a team has multiple options regarding what to work on next, the danger is losing certain options within a repeated cycle of prioritization. A good default system is to work on the oldest units of work unless there is an urgent requirement downstream to prioritize newer units based on other factors.

A good starting point for a flow problem is to use half the current lead time as the buffer. This is not likely the optimal move. However, once the change is enacted, the follow up effort will iron out any details missed in the initial step.

This thought gets at the ethos of The Goal. The ideas from this book are far from perfect and following Goldratt’s words to the letter is hardly the formula for success. However, the framework he creates will work for anyone committed to the process of ongoing improvement. An organization trained to attack the follow up effort and iron out the wrinkles in the initial plan will surely find itself positioned for success in the long term.

Just swerved into a passing truck, big business overtaking, without indicating, he passes on the right, been driving through the night, to bring us the best price…

And this thought from TOA favorite Courtney Barnett’s ‘Dead Fox’ brings home a different point. There is always the temptation to solve problems by putting in the extraordinary effort – with perhaps a broken rule or two along the way – but this always introduces needless risk to the process. Sure, we can pass on the right, but a red light ahead forces everyone to stop again. Was that worth the added risk of a crash?

Even if we take all possible measures in the name of lowering costs, we must ask – since risk is always compensated, why would a process that increases risk be a reliable way to lower price?

Beats me, reader.

Thanks for your time.

Signed,

The Business Bro

ask the business bro (the goal, part 5)

TOA: OK folks, welcome back to ‘Ask The Business Bro’, and maybe for the final time, but we’re here for another round of ‘Q and A’ about The Goal. BB, where did we leave off last time?

BB: Let’s see, well, first, I do agree, this should be the final round, because we really are getting into the weeds here, at least in terms of the detail level. Where did we leave off? I think we were talking generally about how to balance flow through a bottleneck. And really, Goldratt says it best, the rule of thumb to follow is always to use bottleneck capacity to determine the right level of inventory and excess capacity. If you know your bottleneck can handle 100 units of raw material per day, then the organization must design itself to prepare at least 100 units of raw material per day.

TOA: So that all sounds simple enough, well sort of anyway, but I think it leaves out the question of how you might monitor all of this once you setup the organization.

BB: Well, if you go back to our first conversation, I actually answered this.

TOA: You did?

BB: I did. I said it was like lowering the water level of a river until the rocks became visible.

TOA: That was the answer?

BB: Still don’t get it?

TOA: Not exactly…

BB: OK, so think of it like this. What the river analogy really means is that in most organizations, merely watching the water flow downstream hides a lot of the things going on under the surface that dictate how fast the water flows. So you need to identify what parts of the process are hiding the realities of your organization, these realities being the rocks and obstacles so to speak, and make the appropriate adjustments.

TOA: OK, well I’ll try this, let’s see, if the moving water represents flow, then one way to lower the water level, so to speak, is by reducing the release of inventory.

BB: Very good, although maybe more so in theory than in practice. Remember, inventory and excess capacity work together to balance flow, and flow balances only when it fully utilizes the bottleneck. So unless you have the excess capacity required to accelerate the flow of water through your river, you might find that the price you pay for learning where your rocks are is idle time for the bottleneck, and that isn’t the goal, so to speak.

TOA: Ha.

BB: I think part of the issue here is that the analogy brings in an unrealistic assumption.

TOA: Which is what?

BB: The river is made of water.

TOA: That’s unrealistic?

BB: It sure is. Let’s try another version of the analogy. Think about the winter. Suppose the river froze. What happens to flow?

TOA: Well, it would become ice.

BB: So it would stop.

TOA: Right.

BB: Right?

TOA: What?

BB: So it would stop flowing.

TOA: So…

BB: But only on the surface.

TOA: Unless it completely froze.

BB: Well, sure, but let’s say it’s partially frozen.

TOA: Why would it be partially frozen?

BB: Well, because my analogy works better that way.

TOA: This is ridiculous.

BB: Now let’s say the river slowly starts melting. What happens?

TOA: I guess you would have some flow.

BB: What about the ice chunks?

TOA: OK, so first you have some ice chunks, but eventually it would flow.

BB: That’s true, eventually, but the ice chunks are important. Let’s think about those.

TOA: The ice chunks? OK, they would flow at first but not as smoothly as the water, is that it?

BB: Go on.

TOA: And then over time those would melt into water, and eventually there would be perfect flow again.

BB: OK, that’s the right idea. Why would the ice chunks move less smoothly at first?

TOA: Because the ice would bump against obstacles, or each other, or maybe freeze again.

BB: Right, but those problems all go away as soon as the ice melts.

TOA: So what’s the point?

BB: Well, it answers your question about monitoring. One way to think about this analogy is that the batch size is important. If you turn water into ice, that’s like increasing the batch size. That’s like a pizza place saying ‘instead of baking one pizza at a time, we’re gonna do six’. So if you think of it in reverse, melting the ice is like shrinking the batch size.

TOA: I'm lost.

BB: If you think about it, the original analogy’s limitation is that a river of just water is like an organization that works on everything in a batch size of one – there’s just no way to break the pieces down any further. But in reality, most organizations produce in batches. Batches are beneficial in many ways, especially because it creates some visible efficiency, but if it prevents the bottleneck from being fully utilized, it is a big problem. If I order one pizza and the restaurant waits until five other orders come in because their oven is optimized at six pizzas, that’s a problem.

TOA: Are these pizzas frozen, too, or partially frozen even? Does the pepperoni flow?

BB: Oh, quiet. Just understand that one way to monitor flow is to keep an eye on batch size and shrink it as needed until the flow is balanced.

TOA: Even if it is less efficient?

BB: Sure, because the only real thing that matters is keeping the bottleneck utilized. Remember, the bottleneck determines your maximum potential. You could have the most efficient operation in history throughout an organization but if nothing is getting through the bottleneck then you’ll go out of business.

TOA: OK, I understand, so by melting the ice, you see whether the river can handle all the water, and if it can, you need to invest more in melting the ice, efficiency be damned, because without water getting downstream the production of the organization is halted?

BB: Exactly. You could also find ways to simply remove the ice from the river and wait, but then the analogy gets a little too complicated.

TOA: So shrink batch sizes… let me write that down… is that all?

BB: No, certainly not, but at some point you just have to stop asking me and just read the book, right?

TOA: Well…

BB: There are some basic rules of thumb I can put together along the lines of ‘shrink batch sizes to identify flow problems’ and maybe I’ll send those over a little later on. But I think we’ve reached the point where this ‘Q and A’ has done all it can to explain the book.

TOA: OK, well those rules of thumb would certainly be helpful.

BB: No problem.

TOA: Well, thanks for your time. What’s next?

BB: Not sure, actually. I’ve read a couple more ‘BB’ classics recently and maybe I’ll dig into those. I also have some thoughts about email, and I just got a new treadmill desk that I have some comments about. Stay tuned, I suppose.

TOA: Stay tuned, indeed. Thanks for your time, and we’ll see you next time on ‘Ask The Business Bro’.

ask the business bro (the goal, part 4)

TOA: Hi everyone, we’re here again with The Business Bro, he’s got ice on his pulled hamstring and a beer in hand, not sure about that combination, but he looks ready to go again with more Q and A about The Goal. But first, BB, how are you feeling?

BB: I’m good, thanks for asking.

TOA: Before we start today, any thoughts from last time?

BB: Yes, one thing, I think the comparison of my commute to an organization’s production process reveals something Goldratt doesn't really discuss in his book – morale. If you think about my commute, it’s an example of a lot of components working above and beyond to accomplish a goal – the extra spending on the cab ride, the quick transition at the subway, and the infamous sprint here on the last leg.

TOA: Literally.

BB: Quiet. In a real organization, all the workers who made that level of effort would want some kind of recognition or reward. But we see in the example that although everyone truly gave it their best, the organization still fell short of its goal, and therefore probably doesn’t have the resources required to give out extra rewards in recognition of uncommonly good work.

TOA: Because the equivalent of arriving five minutes late for an organization is a loss of revenue or profit?

BB: Right, something like that. Ultimately, the lesson from my commute invokes an idea I read once in The Hard Thing About Hard Things – a good organization is a place where people working hard know that their efforts mean good things for them and for the organization. If an organization is so poorly run that a delay or setback in one part of the process negates all the hard work from the rest of the team, it probably isn’t the sort of place that meets the definition of a good organization. There just would be no way for anyone to know that good work would lead to good results.

TOA: OK, so I guess the logical question is, how does an organization set itself up so that a negative fluctuation in one part of the process can be overcome without extraordinary effort elsewhere in the team?

BB: Well, let’s break the problem down. What is the result of a negative fluctuation on the bottleneck?

TOA: I guess it means the bottleneck works below capacity?

BB: Right. And what would allow the bottleneck to work at capacity after a negative fluctuation?

TOA: Let’s see, if it was somehow still able to access the raw material it needed. So maybe with inventory?

BB: Exactly.

TOA: But I thought you said before that inventory is a negative?

BB: It isn’t really a negative. I just said ignoring the cost of inventory is a dishonest way to calculate the cost of production. The reality is that without those excess reserves, you are almost certainly dooming the bottleneck resource to idle periods, and that means you reduce the overall productive potential of your organization.

TOA: OK, then what’s the way to hold inventory?

BB: The best way is to store it at the lowest cost stage in the process. Of course, to understand this, you’ll need to accept that holding excess reserves of raw material is OK, but if you understand that negative fluctuations are inevitable then this conclusion should follow easily.

TOA: Right.

BB: The problem is a really easy one to solve if the lowest cost stage is directly before the bottleneck resource. However, it does become more challenging the further away you get from the bottleneck. If there are four process steps between your inventory and your bottleneck, for example, then you need to forecast four steps ahead of time in order to anticipate shortages and keep the flow balanced as it moves through the bottleneck.

TOA: And each step involved is subject to its own fluctuations, I assume?

BB: Without question.

TOA: So how do you account for all of that?

BB: Well, in order to balance the unpredictable nature of converting inventory into the material a bottleneck resource requires, you need to have excess capacity in your organization. There’s no other way, really, because if the problem is not having enough of the right material then the solution is finding a way to produce the right material. And if time is a factor, then the solution is finding a way to produce the right material on demand.

TOA: So in other words, you balance between inventory and excess capacity, then use the combination of those to prepare enough material so that you can balance flow through the bottleneck, even during a negative fluctuation?

BB: Hmmm, you know, I’m surprised that you seem to understand this, but yes, that’s a good way to distill the inventory question into a simplified formula of sorts. Really, I suppose it’s also a question of whether it’s more costly to hold inventory or retain excess production capacity, but I don’t think we need to dig too far into that element of the equation.

TOA: I might need to think this through more thoroughly myself.

BB: OK, so break here? There isn’t much left to discuss, I’m afraid, but maybe one more Q and A will be appropriate.

TOA: Yes, if you are up for it, one final chat would be excellent.

BB: OK, same time next week?

TOA: Yup, see you then.

ask the business bro (the goal, part 3)

TOA: Hi folks, welcome back to Ask The Business Bro, and apologies for being a little late this time. Unfortunately, our good man didn’t quite show up on time for this edition and we had to delay the start time. What happened?

BB: Well, unfortunately, I was delayed multiple times on my commute, and the various little issues adding up on my journey proved too much to overcome.

TOA: But you just mentioned before we started that you thought it was good thing you showed up late?

BB: Right, well, I think it’s a good example of what we discussed last time in terms of positive and negative fluctuations not quite evening out.

TOA: OK, explain.

BB: Well, I started the commute by leaving right on time, but then I remembered that I’d forgotten my train pass, so I went back to my building to retrieve it. That meant I walked out the front door ten minutes later than I’d intended. I caught a cab and cut eight minutes off my usual thirty-minute walk to the train station, but I still missed my train by two minutes.

TOA: OK, so you were running two minutes late at that point?

BB: Well, from the point of view of the first portion of my commute, yes, but the real variable here is the actual time I boarded the next train. The trains run every fifteen minutes, so it was fifteen minutes later than I’d intended when I boarded the train.

TOA: OK, so by being two minutes late, you became fifteen minutes late.

BB: Right. The train ran smoothly and I arrived at the connecting subway station without hassle. I usually have to wait five minutes for the subway but as luck would have it the next one was pulling right in as I stepped onto the platform so I was able to board the next subway right away. I got to your station ten minutes later than usual, and then I ran here to cut my ten-minute walk in half.

TOA: And that’s why you were five minutes late.

BB: Exactly.

TOA: So what does this have to do with last week?

BB: The key idea is that my fluctuations didn’t accumulate evenly. I was ten minutes late to leave my apartment but this delay cost me fifteen minutes in the context of the trip because I had to wait for the train. So, even though my initial delay was ten minutes, the way this fluctuation accumulated meant I lost an additional five minutes.

TOA: The cab was obviously a good idea, but didn’t work.

BB: Right. You could compare that to a company rushing production at the end of a tough week. I paid a cab driver to help make up for the initial delay just as you could pay your workers overtime to work over the weekend but it wasn’t enough. I ended up late anyway, and at a higher cost.

TOA: And the same kind of thing on the next leg of the journey?

BB: Yes, the trip from my station to your station was five minutes faster because I cut out the wait for the subway. But even though I’d saved eight minutes on the trip to the first station and an additional five minutes on my trip to the third station, the thirteen minutes of savings still hadn’t made up for the first ten minutes I’d lost.

TOA: And then you sprinted here, which saved five more minutes, but even though you’d saved eighteen minutes overall to make up for an initial ten minute loss you were still five minutes late. So, in a way, you needed eighteen minutes of extra effort to make up for a five minute loss, and yet you still came up five minutes short.

BB (squirming in his seat): Ouch.

TOA: The math makes your head hurt?

BB: No, I think I might have pulled a hamstring while running.

TOA: Wow, are you OK?

BB: I will be, but I need to stretch first.

TOA: Maybe we should pick this up again a little later.

BB: That’s a good idea.

TOA: Let me summarize this, though, just so I get it. Basically, the production process for any organization is like your failed commute, there are so many connections and dependencies that a minor fluctuation in one link of the chain impacts everything downstream in ways disproportionate to the initial fluctuation.

BB: Right. Because of these dependencies, a minute lost in one part of the process doesn’t automatically mean you need to make up a minute elsewhere – it could be two minutes, ten minutes, whatever. If I were lucky, it would have been zero minutes. It’s really impossible to say.

TOA: That’s complicated, but I can see it better due to today’s example.

BB: A far simpler way to put it is that you can be infinitely delayed for anything but you can only be early up to a certain physical limit.

TOA: I’m not sure that’s much simpler.

BB: Well, if you call me at one and invite me over for dinner that night, I could show up an hour early, but not a day early. On the other hand, I could be late by a day, two days, a month – it’s really unlimited.

TOA: I’m not sure anyone is going to be impressed if you show up three weeks late for dinner.

BB: Are you going to let me rest, or not?

TOA: You know what’s funny about you pulling a hamstring?

BB: What?

TOA: In running, a pulled hamstring is usually a good sign that you overdid it somewhere along the way.

BB: What would you know about running?

TOA: A lot.

BB: Oh right, you’ve done those long, boring posts about running, I remember now. Those posts were so boring. I’d rather run myself than read those posts, you know.

TOA: Right. Well, what I’m saying is that an organization that has to increase its effort above a certain threshold risks the equivalent of a pulled hamstring.

BB: Well, I don’t know about that.

TOA: OK fine, I’ll let you rest. We’ll be back soon.

ask the business bro (the goal, part 2)

TOA: OK, welcome back to another edition of Ask The Business Bro, last time we dug a little into the main ideas discussed in The Goal. This week, we’re back to learn a little more about how to apply those ideas. BB, what’s up?

BB: Not too much, I think we left off last time with a question about how to determine excess bottleneck capacity, might as well start there, right?

TOA: Let’s get to it.

BB: As usual, the first answer is deceptively simple. The excess capacity here is just the answer to two questions. First, does the bottleneck resource ever sit idle? Second, could it performer better or faster while it’s working? If the answer is no and no, then you are maxed out, and if the answer to either question is yes, then you can figure out a fairly simple estimate of the excess capacity for the bottleneck.

TOA: OK, so let me see if I get it, let’s say your bottleneck produces five units an hour and it works for seven hours. If you can get up to eight hours, then the spare capacity is five units. If you can get it to produce six units per hour, then the spare capacity is seven units. And if you combine both improvements, then the spare capacity is thirteen units.

BB: Bingo.

TOA: OK, so why is it more complex than that?

BB: It’s not the math that’s tricky, or I should say, the formula is easy, the complex part is getting the right numbers. You might think you produce five units per hour, for example, or that the bottleneck is utilized for seven out of eight hours, but in reality capturing those measurements is the real source of difficulty.

TOA: I imagine Goldratt has some ideas about how to calculate these numbers?

BB: He certainly does. The key to getting the math right is to understand the fluctuations in the process. Each fluctuation creates variation that threatens the reliability of your numbers. If you understand the variation, you can factor those into the calculation and go from there.

TOA: What are some important factors to consider in terms of fluctuations?

BB: Well, before I get into that, I think it’s important to talk about certain assumptions.

TOA: OK, what are the assumptions?

BB: The most important one is that fluctuations average out. This assumption sounds good on paper and anyone who has studied the normal curve in a basic statistics course was trained to think this way. If you crunch your numbers thinking that they could go up or down by 20% with equal probability, you’ll end up with a much different perspective on your operation than someone who thinks differently.

TOA: But why is it automatically the case that those probabilities aren’t equal?

BB: Well, they could be equal, but remember that everything we are talking about ties back to the bottleneck resource. Let’s go back to your numerical example. Suppose the bottleneck resource operates as you defined it, five units per hour for seven hours for a total of thirty-five units. The raw material you need to feed into the bottleneck is therefore equivalent to what produces five units per hour for seven hours.

TOA: Right.

BB: Let’s accept the assumption that positive and negative fluctuations of one unit per hour in the available level of raw material happen with equal probability. On day one, we have a normal day and produce thirty-five units. On day two, we get a negative day and only produce four units per hour for a total of twenty-eight units. We’re down seven units, and we aren’t getting those back.

TOA: Hold on, but what if the next day we have a positive day? Six units per hour means forty-two units, right?

BB: OK, so here we go, six units per hour, we think we’ll get forty-two units because we have the raw material, but remember that we defined the bottleneck as being able to process only five units per hour. In this case, the positive fluctuation on day three doesn’t matter. The unprocessed raw material becomes inventory and at the end of day three we’ve produced thirty-five units for a total of ninety-eight units over three days with seven units of raw material inventory. This means we fell short of the production estimate by those seven units in the inventory.

TOA: But what if we run the bottleneck for an additional hour?

BB: You do the math. We could have ten thousand units of raw material but we can only produce up to the bottlenck capacity.

TOA: OK, five units per hour, so we end up at forty units, with two units of raw material.

BB: The point of The Goal, at least when it digs into the detailed level of how to apply its main principle, hinges on understanding why in most systems fluctuations do not even out. It’s because in most systems, a negative fluctuation accumulates faster than a positive accumulation can make up for it. So even if the probability of the two fluctuations is equal, the reality is that since most bottleneck resources operate at pretty close to capacity most of the time, it’s hard to make up for a bottleneck sitting idle during a negative fluctuation by having it do more work when there is a positive fluctuation.

TOA: I’m seeing the point – you can be stuck in traffic all day but you can’t drive from here to there in an instant.

BB: Right.

TOA: I’m still not sure. Can we go back to the example?

BB: OK.

TOA: Well, what happens if we start with a positive fluctuation? Wouldn’t the accumulated raw material come in handy later on to make up for the shortfall after a negative fluctuation?

BB: It sure would, but let’s do the math. If we have one positive fluctuation, we produce as expected and leave at the end of the day with seven units of raw material inventory. This means we are covered in the event of a negative fluctuation. But keep in mind, inventory is costly, and by holding this inventory we’ve raised the cost of each unit produced. We pay that cost every day until we have a negative fluctuation.

TOA: OK, right, I forgot that we pay for inventory, I guess.

BB: Right. Let’s hammer this point home by looking at a very positive example, let’s say we have a full week of positive fluctuations and we run our bottleneck resource for eight hours instead of seven. At the end of the week, we’ve produced forty units per day instead of thirty-five and we have two extra units of raw material inventory per day. On the surface that looks good, but we're only about two consecutive days of negative fluctuations away from eating up that spare inventory and producing less than the thirty-five units per day.

TOA: But can’t we afford that since we overproduced the week before?

BB: I’m sure it’s true in some cases, but it’s not a given. Remember that we’re working off of an assumed output of thirty-five units per day, so unless the sales team responds quickly and sells those extra units we might not be able to recognize the revenue for those products right away. If the completed product has a short shelf life, we might not recognize the revenue at all. Plus, since we have to buy space to store those extra units, each extra unit above our estimate comes at a slightly higher cost, so we might need to discount the revenue gained from each additional unit to account for this.

TOA: This is getting pretty complicated.

BB: Right. The short version is that if we overproduce, we run the risk of learning the hard way why we don’t overproduce all the time.

TOA: Are there any specific strategies we can use to get around this?

BB: There are, and Goldratt outlines a few, but unfortunately I must say that is where it gets complicated.

TOA: Oh, now it gets complicated?

BB: I think it might be best to leave the tactical discussion for next week.

TOA: OK, that will work for me.

BB: The key thing to remember is that negative fluctuations tend to accumulate faster than positive fluctuations will even them out.

TOA: Yes sir, I will try to do that. Same time next week?

BB: Sounds good to me.

ask the business bro (the goal)

Hi all,

My TOA counterpart has posted a few thoughts recently on The Goal that, I must admit, showed an unexpected level of understanding of Eliyahu Goldratt’s business classic. He asked for my help, however, in breaking down some of the more challenging concepts so I agreed to step in and write a few posts to help him out.

Unfortunately, I realized that the book is so dense with insights and tactics that a traditional series of ‘book club’ posts was not going to work as well for The Goal as it did for some of our past books. I met briefly with the boss and we determined that a loose ‘Q and A’ format might be the best way to both deliver the lessons of the book while also clearly applying its ideas. We’re not sure about this, of course, but nothing beats a try, so please enjoy our first edition of ‘Ask The Business Bro’.

*********

TOA: What does The Goal have to say about running an organization?

BB: The book’s singular focus is on maximizing the constraints that determine whether an organization can meet demand. Everything Goldratt says about running an organization ties back to this idea.

The Goal suggests the best way to serve this idea is to implement an ongoing improvement process. The key question that drives this process is always ‘where is the constraint now?’ and the answer is always a series of steps that will break the constraint. Once the constraint is broken, the organization asks again – where is the constraint now?

TOA: Does Goldratt recommend any specific strategies to find the constraint?

BB: No, unfortunately it isn’t as easy as applying a series of steps, but there are some good principles to keep in mind that always help. Broadly speaking, improving flow will always help an operation. Goldratt compares flow to water moving down a rocky river and the process of improving flow to finding and removing the rocks. A poorly run organization might do the equivalent of wading into the river while hoping to stub a toe. A a world-class organization will do the equivalent of lowering the water level until the rocks become plainly visible.

TOA: But what if no rocks are visible?

BB: Then lower the water level again.

TOA: But doesn’t that hurt flow?

BB: Well, OK, I see, look, it’s not a perfect analogy, but then again, what is? One obvious problem with my analogy is the possibility of confusing flow with total output. True, output is related to flow, but not in the way you are suggesting. The point of the comparison isn’t to maximize the cubic footage of the water moving downstream, it’s to ensure that if you put 100 cubic feet of water into the system, 100 cubic feet of water will flow through the system without unexpected interruption. Flow in this case is how quickly the water moves down the river, not the amount of water that empties out into the ocean.

TOA: OK, I see, so flow is like the delivery time of the pizza, not the number of pizzas sold?

BB: Why are you always thinking about food?

TOA: Hey, I’m just hungry for answers here.

BB: Right, well, if that’s the way you want to think of it, then yes, flow in this case is the time you wait for the pizza after placing the order, though maybe a better analogy is to say it is like the entire process of making and delivering the pizza

TOA: I see. I think I understand some of it, the crust at least. If the whole point is constant improvement and the process for constant improvement is maximizing flow, then sure, you just maximize flow, or whatever. But what if the flow is already maximized? What do you do then?

BB: Well, Goldratt points out that since a bottleneck is a resource whose capacity is less than or equal to the demand placed on it, the flow through a bottleneck should be less than or equal to the demand. Your question suggests that there is some spare capacity, so I would suggest finding a way to increase demand.

TOA: So sales and marketing?

BB: I think so. You just have to make sure those teams understand the excess capacity available on the bottleneck resources so that they do not commit beyond what you can actually produce.

TOA: This sounds too simple.

BB: Yeah, sort of, it’s simple in the same way getting a full night of sleep is, at a basic level all we all know what a full night of sleep is, but in reality a lot more goes into it than just blocking out eight hours to lie still.

TOA: Is that something Goldratt wrote?

BB: No, that’s a BB original.

TOA: Makes sense, you don't need any help to know to lie, still.

BB: What does that mean?

TOA: Nothing.

BB: Do you have anything else?

TOA: Yes, so if the cycle eventually leads to the sales team finding new demand for the product and what they need to know is the excess capacity available on the bottleneck, how do you go about calculating that excess capacity?

BB: Wow, that’s a short question but a big answer. I think we’ll need to cover that next time.

TOA: OK, sounds good. Same time next week?

BB: That will work for me.

reading review - the goal (cost profiles)

The appendix of The Goal notes how the presence of price competition often indicates that customers feel unable to control most aspects of their cost profiles. This conclusion is intuitive – if customers were able to lower other costs, they might choose goods and services based on factors unrelated to price.

An organization with a superior good or service can sometimes be caught on the other side of this problem if they are unable to lower their price point. What is the best approach if customers cannot cover the cost for producing a good or service? To put it another way, if it costs $10 to produce a pizza but customers never pay more than $8, that’s the end of the pizza industry, right?

A clever organization can survive such a challenge if it can lower the cost profiles of its customers. There are simple ways to do this. If customers are unable to afford to travel to a particular store, for example, an organization can lower this aspect of the cost profile by offering a delivery service for a fee that is lower than the transportation cost for a customer. There are a number of more complex examples that grow from this idea – money back guarantees, rollover minutes, rewards programs, and so on.

What these organizations and their methods all have in common is the way they charge a customer for the right to a lower cost profile. The rise of Internet commerce provides many examples for how organizations find ways to lower a customer’s cost profile. A theory-based way to think of Seattle-based Amazon’s shipping service is that it eliminates the need to buy a plane ticket to the Pacific Northwest anytime a customer wants to buy from Amazon. In this approach, a customer who could not afford to buy a book for $20 plus the round-trip airfare would become able to afford the book for $20 plus the cost of shipping.

A more practical understanding suggests that shipping eliminates the cost of driving to any local bookstore that might sell the same book. The majority of Amazon book customers probably consider the cost of shipping to be less than the cost of going to the bookstore. This thought process applies to everything Amazon offers and manifests in its Prime membership program – for a flat annual fee, Amazon will ship anything to you for free. The program’s success hinges on a simple calculation – any customer who feels the membership cost is less than the cost of all the trips made in a year to brick and mortar stores will sign up for the service. This in turn lowers their overall cost profile, which in turn frees up a customer’s spending money for buying products on Amazon that might cost more than the prices offered on local store shelves.

If the program is executed successfully, the customer who is right on the edge of purchasing a single good or service is enticed into making the purchase (and many similar ones in the future) by signing up for the program, lowering their overall cost, and finding that, all things considered, the more expensive good is better for their overall financial health.

reading review - the goal (money, money, money)

In my first post about The Goal, I described the general process of elevating the production capacity of bottleneck resources to help an organization reach its full operating potential. Today, I’ll take a brief look into the back room and go over ways to measure the progress of these techniques.

Goldratt measures an organization’s money using three simple measurements – inventory, throughput, and operating expense. He helps the reader understand these measurements by defining each in the context of money. Inventory is the money invested in what the organization will sell, throughput is the rate of money generated by sales, and operating expense is the money spent turning inventory into throughput. If we return to the pizza restaurant analogy from my prior post, inventory is the cost of every ingredient in a pizza, throughput is the amount of money generated by pizza sales, and operating expense is the cost of cooking and delivering a pizza to a paying customer.

Armed with these three measurements, anyone can understand where an organization’s money is going – the amount of money entering the organization is throughput, the amount of money leaving the organization is operating expense, and the money ‘stuck’ inside the organization is inventory. And since money is usually a reliable measure for the health of the organization, understanding changes in money can make simple yet powerful insights into the health of the organization – productivity is up if inventories go down, operating expense goes down, or throughput goes up.

Understanding an organization isn’t always so simple, of course, because it is rare for one of these measurements to fluctuate while the other two remain fixed in place (if all I know is that one measurement went up and another went down, the best I can say about the effect on the entire organization is that it is ambiguous). The value of these measurements is significant, however, because they provide global context that can illuminate the truth hidden by local metrics. Our pizza restaurant from above might boast of ‘cost efficiencies’ such as cutting down on delivery times or using less energy to run the oven – however, if the amount of pizza ingredients (inventory) measured at the end of each day continues to rise, an astute observer will know that finding a way to turn inventory into throughput is critical so that the reduction in operating expense is not offset by the rising inventory cost.

One up: As Goldratt reminds us readers a number of times in The Goal, a vaguely defined measurement is a step shy of useless. A good number cruncher will always keep this fact in mind. If a colleague comes forward seeking help with understanding a suspicious figure, a good approach is to ignore studying the calculations until all the underlying assumptions baked into the figure are understood.

The thought also links nicely to the old adage about the wizened hammer seeing a world full of nails. Just as the hammer knows it has only one song and tries to sing it whenever possible, a highly trained number cruncher is always tempted to crunch the numbers – this is what a number cruncher knows best! There is always going to be a time, though, when validating the assumptions is going to be more productive than redoing the calculations for a twenty-third time.

One down: I wrote a few posts ages ago after reading Hillbilly Elegy about cash flow so I will not harp too much more on the topic today. I’ll simply mention that no matter how strong a company’s books are from the points of view detailed above, there is usually only one solution for an organization suffering a cash flow crisis – more cash. Whenever an organization with a strong financial profile begins making strange financial moves, it is a good bet that a cash flow concern is lurking somewhere in the background.

Returns on investment are all well and good but if the cash on hand is insufficient then a business learns quickly that bankruptcy is a trump card against even the healthiest calculations of thse measurements discussed above. Cash for an organization is like air for a swimmer - even a gold medalist will lose a long enough race to any amateur if the champion is not allowed a breath.

Just saying: In my first post, I mentioned how an organization should change its decision making process based on whether its operating constraints were inside the organization (production capacity of its key resources) or outside of the organization (demand from the market). If the organization’s production capacity was causing a failure to meet demand, decisions should be made to maximize bottleneck resources. If the market was not generating enough demand, a traditional efficiency measure such as sales less material divided by man-hours was perhaps a better option in order to free up raw resources for marketing or product development.

In terms of measurements, what this idea does is take the three productivity measures outlined above and shift the focus of their metrics from revenue generation to cost reduction whenever the operational bottleneck is maximized. The organization is no longer concerned about whether throughput meets demand because, by definition, throughput is equal to demand whenever the bottleneck resource is maximized. A similar thought process applies to inventory because the amount of inventory needed to maintain throughput should become increasingly stable whenever demand is a consistent leading indicator of upcoming throughput. And if the underlying rates associated with operating expenses do not change, the fall in inventory should naturally lead to a reduction in operating expense.

i read the goal so you don't have to

The Goal by Eliyahu Goldratt (October 2017)

Ah, finally, I get around to posting this one...

I first read The Goal in the winter of 2012 after having a VP recommend the book to me on a business trip (those were the days). He considered the book ‘an MBA staple’ and he thought I would find it very useful in the context of understanding our organization’s ongoing operational challenges.

I suppose I can’t be entirely sure about the MBA bit (though a great friend did read the book himself as part of his MBA coursework) but the VP was one hundred percent right about the latter – The Goal was filled with immediately applicable ideas for our company and I leaned on many of Goldratt’s insights over the next few years as I grew into my role. I will dig into the specifics over several upcoming posts so that you can have a better idea of how to apply Goldratt’s ideas to the real world.

For today, however, I just want to focus on the main idea of The Goal. It’s a tough concept to distill into one sentence but I’ll give it a shot anyway – the slowest rower determines the speed of the boat. If that thought feels a little too casual for a business concept, let's try this one - an organization's singular focus should be on maximizing utilization of all resources that enable it to meet demand.

The Goal uses the concept of bottleneck resources to drive this point home. A bottleneck is any resource whose capacity is less than the demand placed on it. As an example, suppose you are hosting a breakfast at your home. If your guests collectively demand eight coffees and your coffee maker can only produce two cups at a time, then your coffee maker is a bottleneck for your ability to produce coffee in a timely fashion for breakfasts serving more than two people.

Now reader, you may protest – but why can’t I just run the machine four times? And I agree here, in a sense my analogy assume your guests will storm out if they aren't handed the first coffees, but I must note that what works in your kitchen might not be the same thing as what works in the market. An organization’s goal at all times is to ensure the flow of production through a bottleneck resource never falters because each minute lost waiting at a bottleneck is a minute lost to the entire organization. To put it another way, making six customers wait for a coffee risks losing their business.

Let’s think about another example that more closely illustrates the business application – a pizza restaurant. A pizza restaurant with a daily demand for twelve pizzas per day will maximize revenue if it is able to produce twelve pizzas per day. To keep it simple, let's further assume that as long as the pizza is delivered on the same day, it's no problem. If it takes the oven an hour to produce a single pizza, then the pizza oven must work for twelve hours per day to meet demand. If anything happens during the day to cause the oven to work for less than twelve hours, the organization will lose a pizza order at the rate of one pizza per lost hour.

Another way to think about this is that once the oven works for less than twelve hours on a given day, there is absolutely nothing the organization can do to make up the lost time. This is the feature that distinguishes a non-bottleneck resource from a bottleneck. A non-bottleneck has options to make up for lost time because the demand is less than its potential output. A bottleneck, on the other hand, simply must get the job done on time if it wants to meet the demand placed on it. In the above example, the oven is the bottleneck on any day where there are more than twelve orders.

The way bottlenecks are defined in The Goal does not rule out the possibility of there being multiple bottlenecks in any organization. However, there is usually one resource whose capacity shortfall in comparison to demand is the greatest among all the other bottlenecks. This resource becomes a constant target of management attention as it seems to find new ways to prevent the organization from meeting demand. The only way to meet the challenge is to increase the capacity of the bottleneck and to repeat this process for each new bottleneck that emerges (or for each old bottleneck that returns). The nature of how bottlenecks change as an organization elevates the capacity of its resources defines the ongoing improvement process that underlies The Goal. At all times, a capable operations team should ask itself – where is the constraint now and how can we break it?

If the organization is successful in identifying and breaking each new bottleneck, eventually the constraint will move outside the organization when demand for the good or service is less than the organization’s maximum production capacity. At this point, since by definition an organization is able to sell every good or service it provides, it is appropriate for an organization to use traditional efficiency measurements – such as sales less material divided by man-hours – to drive decision-making and increase profits. If demand ever exceeds production capacity again, however, the organization must be ready to subordinate all operational decisions to breaking the next bottleneck.

The simplicity of the main idea covered by The Goal hides the depth of its power. Over the years since I first read this book, I’ve found myself noting its applications in a variety of different environments. In a series of upcoming posts over the next few months, I’ll look into these in more detail. We’ll go to my favorite ramen restaurant, talk about when to ignore a running injury, and consider how the bottleneck concept applies to workplace ergonomics. Before all that, though, I’ll start with a traditional reading review or two and maybe, just maybe, ask The Business Bro for his thoughts on the book.

Until then, thanks for reading.

Tim