S2 Ep. 5: Decoding Manufacturing Modes: A Comprehensive Guide

The Four Manufacturing Modes and How Integration Fits Each One (S2 E5)
The Integrate Intelligently Podcast, Season 2, Episode 5 — with Jeff Brickler and Scott Brickler (CADTALK)

Jeff: Welcome back to The Integrate Intelligently Podcast. I'm your host, Jeff Brickler, accompanied as usual by my brother, founder and CEO of CADTALK Software, Scott Brickler.

I'm excited about today because we don't have a guest — this is an educational episode, and coming from a background in education, I love the chance to share what we know. We're going deep on manufacturing strategies, maybe even out of my depth, and how different types of manufacturers actually work.

Before I came to CADTALK, before I knew anything about manufacturing, I assumed all manufacturers were the same. You manufacture stuff — what's the difference? The reality is they're quite diverse, and I don't just mean different products. Different methods. There are four main methods we're going to cover today: the foundations of each, how they work, and what we've seen with our own customers, including how CADTALK evolved from serving one type toward others. Since I've been here we've seen all of them, and often mixed.

Scott, you're the subject matter expert. Everything I've learned about manufacturing came from talking to customers and to people like you, not from school or from working in it. So before we get into the four, what is a manufacturing mode?

What Is a Manufacturing Mode?
Scott: Think of it as a strategy — how you go about making things, which usually has a lot to do with the kind of products you make.

Picture a spectrum. On one side you have mass production. On the other, one-off, highly customized products. These modes exist somewhere along that spectrum. And almost no company is purely one mode. They usually have a mix.

But the way you structure your business changes depending on which mode you mostly operate in, because the problems change — especially on the ERP side. There are different strategies for cost accounting, for how you measure things, for how you estimate. If I had to boil it down, most of it comes back to how much predictability there is in your process, and how much variability the products you make introduce.

Jeff: So ERPs would need to support all these modes. Otherwise they'd only work for one type of manufacturer.

Scott: Most ERPs are based on APICS standards — I'm blanking on what the acronym stands for, but it's the organization that developed the standards for how accounting works around ERP and MRP. Most ERP systems are compliant with that, so they can generally handle most types of manufacturing.

That said, some ERPs are geared toward certain problems and modes more than others. If an ERP is structured around make-to-order or make-to-stock, it tends to be less friendly to engineer-to-order thinking, since those sit on opposite ends of the spectrum. And there are even philosophies within each mode.

Make to Stock
Jeff: Let's define each one. First on my list is make to stock. If I weren't working in manufacturing, I'd assume it means exactly what it sounds like — you make something and put it on a shelf. Is that right, or did I make that up?

Scott: That's basically it. You have a standard product, maybe in a couple of different options, and you make them because you know there's a certain amount of demand in the market. It's mass production. You make to stock, and the stock gets depleted through the supply chain as people buy it.

The distinction from make to order is that you're making it with the intention of putting it on a shelf for retail purchase. Examples would be clothing lines or office supplies — you know a certain number of pencils get consumed a month, so you make that many pencils for stock and replenish the shelf at the store. You're producing to a plan based on what normal demand looks like.

Jeff: So my Diet Sunkist here — everyone knows I drink these — they're not making that just for me. They're making it for a lot of people.

Scott: Exactly. Although you may be their biggest consumer.

Jeff: So food products generally. Kroger, or Albertsons, or Publix buys from a manufacturer who makes the cans, packages them, puts them in a warehouse, and distributes them to us. Same with anything else on a shelf — electronics at Best Buy, speakers, clothing.

How Better Data Is Turning Make to Stock Into Make to Order
Scott: One thing I find interesting about make to stock, and this is my opinion, is that you see less and less of it, because computer systems have gotten so much more sophisticated.

Walmart is the example. With their tracking systems, they turned a lot of make to stock into make to order, because they could track down to the level of the transaction at the checkout and trigger ordering from that. Before, you were guessing at a stocking level, and that guess trickled down through the supply chain to manufacturers who made things in bulk.

I'm not a supply chain expert, but my experience is that you're seeing more make to stock converted into make to order, just on a mass scale. Make to order usually means I buy this thing and they make it for me. But those large suppliers ordering different batch sizes based on dynamic demand is closer to a make-to-order scenario than a make-to-stock one. People would argue different opinions on that.

Jeff: But it's a spectrum. As technology improves, make to stock can move toward make to order — maybe not fully, but it trends that way. Before the data was good, manufacturers made a bunch of stuff, put it on the shelf, and waited for someone to buy it. Without that data you're always trying to find the balance, and with lead times on spinning production up or down, you end up with too much or too little. Overstock sales on one side, shortages on the other. I assume that still happens, just less.

Scott: It still happens. Think about the Christmas hot toy — there's no way to predict what it'll be. When we were kids, showing our age, it was the Atari, and they couldn't make enough. Or the Nintendo. And if they guess wrong on production, they don't have enough to sell.

Jeff: Or too many.

Scott: Which could be worse, depending on your cost ratios. You can make a lot more, but if your cost is low, the extra hurts less — it depends on your gross margins on materials.

Back then they weren't getting pre-orders for Ataris. They were just guessing. Now, as the feedback loop gets tighter, they can measure consumer demand — there's a lot of buzz around this on social media, and we can correlate that buzz with a certain level of sales. It's still technically make to stock, but the estimates are so much more accurate because the feedback and the technology are better.

Make to Order
Jeff: So make to order — a customer calls and says they want to buy something, and the manufacturer makes it.

Scott: Typically. Examples would be furniture or machinery. It's still a fairly standard product that they make. The difference is they're not making it ahead of time or predicting demand. They wait for actual demand to show up, the order kicks out, and they make it for that order. They'd rarely make it just to put on a shelf.

Make to order also tends to mean they're actually making the components, not just assembling. Everything is driven by an order of some sort, whether from the end customer or from up the supply chain. As I was alluding to with make to stock, it depends where you sit in the chain — you might be making it against an order from a grocery store that puts it into their stock.

Generally it's a standard or configurable product, something with a few features and options to choose from, that you're building for that customer.

Jeff: Let's touch on configuring. When I hear that, I think of a configurator. We talked a couple of weeks ago with a customer whose goal was to let customers configure anything right on their website, and minutes later it's on the shop floor being made. So a configurator would be applicable to a make-to-order company.

Scott: Absolutely. A configurator typically shows up at a make-to-order or assemble-to-order company, because there are features and options and they don't want to make to stock — the number of variations would be prohibitive.

In the case of PBC Linear from last time, there are hundreds of thousands of combinations. You can't guess your way to stocking that. You wait for the orders.

Jeff: So if I order a specialty color car — same car, but a one-off color they don't normally do — that's make to order.

Scott: Right. With Toyota production, people think of cars as made to stock, and some are, because there's predictable demand for certain models. But a lot of the time, especially for Toyota, they're made to the order, because somebody ordered it through the dealership and then it gets built on the line for that order. A Ferrari would generally be made to order rather than put on a showroom floor for someone to buy.

Assemble to Order
Jeff: You mentioned assemble to order and engineer to order. Which next?

Scott: Assemble to order, because it's similar to make to order. The difference is how far along the parts already are.

Dell is the example. I like the term configure to order, because it encompasses both make to order and assemble to order — the real question is how far along the raw materials are.

In make to order, I might pull bar stock, make the part, put it together, and sell it to you. PBC Linear has to make the shaft to a certain length, because pre-making shafts would tie up inventory at lengths you're not sure you'll sell.

With a computer manufacturer like Dell, they know they'll sell a certain number of RAM chips and hard drives pretty much all the time. So they buy those components and assemble them for the configured order, and they know which combinations of pre-built parts can be assembled for the orders they get.

The reason to do that is speed. If the manufacturing is already done and you can build a lot of different products from pre-built components, you can still deliver a highly configurable product in far less time, because you're not cutting metal. You pull components from stock and assemble them into the end product.

Jeff: So an assemble-to-order manufacturer, true to the name, orders components manufactured by someone else, keeps them in stock, and puts them together based on the order. Like calling Dell back in the day and saying I want this hard drive and this screen — they have the components, they assemble, box, and ship.

Scott: Exactly. They may manufacture some components — fabricate the case, maybe, though I think they outsource that. Again, it's a spectrum. But the point is the parts are pre-made, so lead times are shorter. If you ordered a Dell and had to wait six months, you wouldn't buy it. You need it in two or three weeks.

Jeff: And the flexibility they can offer is lower. Not low, but not wide open. Lots of options, but limited — the limit might be a hundred, where PBC Linear's could be nearly unlimited.

Scott: That's the trade-off between flexibility and speed. If you have to make everything, it's unlimited, but you can't deliver as fast. It's an optimization problem, like everything in manufacturing and coding.

If I make components ahead of time, I'm adding cost to something I hope I can sell, so I'm taking on cost risk. If I keep it as raw material, that's the lowest-cost form it can be in, which lowers risk but increases time. You're always balancing cost, risk, and time to find the mix that gives you a competitive advantage.

Engineer to Order
Jeff: The last one is the one you have direct experience with, and I think where the idea for CADTALK originally came from. What is engineer to order?

Scott: You're making a more customized product — something engineered for your specific use case.

I came from the packaging industry. Packaging machines are considered engineer to order because every product you're packaging is a little different. We made the machines that put bags of cereal into cartons. You'd think cereal boxes are all about the same, but they're not. They're shaped differently, they have different flaps, different sizes, even different colors and barcode placement. All of that is slightly different.

In engineer to order you're usually not starting from a clean slate. You start from a more standard design — like buying a custom home, where you typically start from a template that's known to work, so you're not reinventing the wheel, but you can change it to what you want.

With the packaging machines, they'd let you pick the motors and the controls. Some customers had deals with Allen-Bradley where they got discounts if the machines used those controls, so even if we standardized on Siemens controls, we'd swap them out for Allen-Bradley. That's what engineer to order lets you do. But generally there's some part you have to engineer specifically for that customer order, and that's what defines it.

That's where CADTALK fit best, because there's so much engineering change happening. When I worked at the packaging company, there were about 500 people and over 200 of them were engineers.

Jeff: Wow.

Scott: More than a third of the company. So a lot of engineering is happening — something like 3,000 new part numbers and 3,000 engineering changes a month. You're cranking out designs, because you're making them per order. Even with baseline products, you're generating variations for each customer's specifications.

Jeff: Customers used to tell me: we make the same thing, but different. And I heard that packaging equipment designs could change depending on what floor the machine was on, or where in the country it was, because of humidity or temperature. Maybe that's less true now with better climate control. But the type of box, the ink used, where it sat in the shop, where it sat in the country — all of it mattered, because small factors could make the machine run inefficiently or not work at all.

Scott: There's a funny story about that. One customer I can't name had two machines on the same floor packaging exactly the same thing, and the program that ran one machine would not run the other.

The reason is they weren't made at exactly the same time, and the guides and other components differed slightly. They got the same end result, but they didn't get there the same way. So they were customized even at the software level, because there were different parts somewhere. There are constant little tweaks, and when you're running hundreds of boxes a minute, very small variations make a big difference.

Where Integration Fits in Each Mode
Jeff: So CADTALK came out of your time at that packaging company. But something that comes up on sales calls and at trade shows is a customer saying, we don't make the same thing every time, so we wouldn't need CADTALK. And I'd think, wait — you'd need it less if you made the same thing over and over. People sometimes have it backwards.

So what's different about these modes in terms of CAD requirements? We have customers across the whole spectrum.

Scott: Start with the intuitive fit, then there's the less intuitive one, which has to do with cost savings.

An integration typically provides several benefits. The most common and easiest to understand is time savings — how much time am I saving by bringing this data over? That's what everybody picks up on immediately. In an engineer-to-order scenario there's a lot of engineering happening, so transferring that data is enormously time consuming and costly.

Over time we started seeing it with customers doing make to order, or configure to order. Configure to order looks like engineer to order; it's just that software generates the configurations instead of an engineer. Same problems, lots of variations, so speed helps there too.

But with make to stock, what's driving it? You typically see more PLM in those customers, where they're working toward a model year or running a long engineering cycle to create a product variation. That's where the benefit of consistency really shines. Software is simply more consistent than people doing it by hand, and it's faster as they iterate. People like that the software does it consistently and accurately.

So it depends which benefit you're after. Integration has multiple benefits, and you value them differently depending on what kind of business you are.

Jeff: So the use cases differ by mode. Engineer to order: I'm generating lots of new parts, I need new SKUs in my ERP as fast as possible to get to the shop floor and deliver. Make to stock: the customer has a PLM, and they still have to design every model year — a long development cycle — but the quality of the data going into the PLM is critical. Getting it out of CAD and into the PLM isn't so much a time issue as a data quality issue.

So engineer to order is really about speed, because of long lead times. Make to stock is about quality — the new Nintendo system or the new iPhone, developed over time. And make to order, like the customer we discussed, waits for an order and has software engineering it, but still has to get it into the ERP. They wanted it to go from customer order to ERP in minutes, so speed mattered — though speed and quality both mattered there.

Assemble to order we don't see as often, but I'd think they'd still use a PLM, at least to manage components.

Scott: They could. It depends. You do see PLM with assemble-to-order products, and you typically see configurators too. Not always, but you can.

And with a configurator, you generally have a bill of materials being generated on the fly. Any time there's a configurator, there's the opportunity for a completely unique bill of materials to be generated — and that unique BOM has to get into an ERP somewhere. That's typically where the integration point is for us.

So even when the parts aren't being made — even when we're not getting the advantage of figuring out manufacturing steps, because those are already done — there are plenty of customers who don't make any parts at all. They buy everything and assemble it. They still have bill of materials variations to deal with.

From Engineer to Order Toward Configurable
Jeff: When I started at CADTALK ten years ago, engineer to order was the main customer in your mind. Over the last ten years I've seen all these other types, with similar needs — better data quality, faster into the system, streamlined processes. And no manufacturer is purely one thing. Was the company you worked for purely engineer to order, or on the spectrum?

Scott: We were always trying to get to more of a configured product. One of the initiatives I was on right out of college was getting to a configure-to-order version of the machines. The problem with engineer to order is it tends to be expensive, because you're paying for engineering. So how do I get to a more configurable machine? You may not eliminate all engineering, but you reduce it, and then you can compete on price — which everybody has to do somewhere, even if we hate it.

It was still a spectrum. The question was how configurable versus how engineered. My projects were about driving less engineering and more configurability into it, which led to an interesting constraint.

Here's a side story. At that company, they'd start with a basic design and change it for the customer order. They also did something called conversions, where they'd take an old machine already out at a customer site and retrofit it for a new product.

The profitability on conversions was really good — and my understanding was that the best engineers were on conversions. Which sounds backwards, until you ask why. Because they had more constraints. The machine is already built, and the more you take off and re-engineer, the more expensive it gets. When it's just a design in virtual space with no parts made, there's no cost to making the part different. But when there's a physical thing sitting there, you get more creative about reusing what's already there and making it work.

So one idea I remember having was: could you treat the original design as if it were a conversion? Force the engineers to think the way conversion engineers did, to make the existing machine work, which requires less engineering. I thought that was a cool concept, because it shows that constraints bring out the genius in people.

The premise was driving engineering out of the problem and configurability into it, and the theory was to do it by increasing constraints — you can only change these parts, or only this section of the machine. Which would drive more ingenuity.

The reason it didn't happen is that those problems are really hard to solve, and engineers were under time pressure to deliver designs. It's easier to change the design than to find a genuinely creative solution that fits the constraints. That was the push and pull. But that's typically what you see in engineer to order — a push toward configurability. There's always some engineered portion; you might say 50 or 60 or 80 percent of the machine is re-engineered, and the goal is to drive that percentage down.

One Size Doesn't Fit All
Jeff: I learned a lot. I intuitively knew some of this from talking to so many customers, but I'd never sat down and thought through the different types. Anything else to add?

Scott: One side note. Coming out of an engineer-to-order company and then going into consulting and software, you tend to assume the rest of the world works the way your world worked.

There's nuance in how you apply solutions across these modes. If you come from engineer to order, you'll say we need to do this because it cuts lead time — but those constraints and problems aren't the same at a company doing assemble to order or make to order. Everybody's doing manufacturing operations at the core level, but their constraints differ. You have to understand, based on the mode, what they're optimizing for. If you know those nuances, you know how to apply the right solution at the right time.

That was a really good lesson going from working at a company to consulting and ultimately to software. It fueled the question I always ask now: what is this customer really optimizing for? What constraints are they dealing with? They may be engineer to order, but their real problem might be competing in their market, or lead time. You have to understand that, and the modes give you a framework for thinking about it.

Jeff: That's a good point for anyone in a consultative role, in implementation or in sales. What matters to them changes based on what kind of manufacturing they do. One size doesn't fit all, and we tend to use our own experience to push people toward certain things when it isn't always the best solution.

Thanks everyone for listening. Subscribe to The Integrate Intelligently Podcast, leave comments and feedback, and if you're a manufacturer or a consultant who works with manufacturers and you'd like to be a guest, reach out.