Labor Quoting in EMS - Methods and Formulas
In this presentation of "Scratching the Surface Mount" webinar series, Chintan Sutaria reviews 5 methods of Labor Estimation used by the Electronics Contract Manufacturing industry and provides an introductory explanation for utilizing a learning curve for efficiency calculations.
I would like to get started with here with a quick introduction. First of all, thank you very much to everyone who was able to participate today. I appreciate you for taking the time out of your day for talking with us about labor quoting because I know it is a pretty exciting topic for everybody. This webinar is a part of a series of webinars that we call “Scratching the Surface Mount.” What they entail is that we try to share some of the best practices we learn as we talk to EMS companies, contract manufacturers across the industry. Before we get started, I am going to do a quick introduction of myself. My name is Chintan Sutaria, I am the President of Calcuquote. Calcuquote, if you have not heard of it before, it is an EMS focused quoting solution. We help with the end to end RFQ process from requirements when they first come through the door to BOM, materials, labor, overhead and preparing the final quote that you can send out to your customers. My professional background is that I grew up in the EMS industry so, having a family business that was PCB assembly I got to spend time working in the industry and learned a lot about it. I have also spent time working in consulting working with multinational manufacturing companies to help with business process consultation. So, I have gotten a chance through Calcuquote to work with more than 60 different EMS companies across the US and that is how we came to today talking about labor and going over what some of that entails.
Just to give you a quick overview of the agenda, we will start by going over what is the relevant information for labor estimation and then we will get into 5 different labor quoting methods and what formulas are involved in those and we will keep it relatively intro level and then we will get into more in-depth understanding of one of those which is going to be activities based quoting model. Finally we will close by talking about learning curve or efficiency and how you can count for efficiencies while going through the learning curve. We will get a little bit more technical towards the end. At any point, if you feel we are going to fast, feel free to make a comment or chat with us at the end. The other thing I would like to say before we dive into this is when we talk about labor estimation in the EMS industry, from the many companies that we have had a chance to speak to about their labor models, a lot of people are under the belief that they have some kind of secret sauce or some unique way of estimating labor. What I can tell you is that we have yet to encounter somebody's method of doing labor that is absolutely unique from everyone else. The parameters might be a little bit different, the assumptions might be a little bit different, but overall I would say that most things fall into the 5 labor quoting methods that we are going to talk about. There is no secret sauce to it and we are not sharing any kind of company secrets here. It is just the matter of how you apply that to your particular labor model and your company.
- Labor Estimation
So without further ado, we are going to start off by kicking off another poll. The poll is going to be, what information would you need to know to come up with a labor estimate? So you should see a poll on your screen now and you can select one or more of those options and I will just give everyone a couple of seconds to respond.
It looks like a little more than half voted. Before I share the results, working in EMS, contract manufacturing, you are really at the mercy of your customer and what information they provide for you to be able to provide an accurate labor estimation back to them. While it might be nice to have all of these as well as other information, you are not always going to get that, and that is what makes this whole labor estimation process so tricky. You're working based on limited information and a lot of assumptions to come up with that. On the flip side, we do see where information that is not necessarily relevant to the approach you are using is being captured but never having anything done to it. An example would be through a process review, you might find your upfront process requires that you have to have the value of the components before the labor estimation person can look at it. It turns out that the labor estimation person is actually not looking at the value of the components and so it is unnecessarily holding up the process. One of the things that I would encourage everyone to do is really take a look at what information you do need as part of your labor estimation and then make sure that is not being a stopping factor further up the process for progress. I am going to go ahead and end this poll and share the results with you. It looks to me like the number of components is definitely one of the winners here. So people definitely want to know how many components there are. Assembly drawings, single and double sided is also relevant. Then we start to trail off from there. Whether the board is panelized or not, which operation would be involved. Surprisingly, quality and regulatory requirements do not seem to be as popular as some of the other ones where I would think that would be a little bit more significant.
- Labor Quoting Methods
We talked about what information is relevant, and now we are going to take a look at 5 different labor quoting methods. I have put them on a spectrum of what I am calling non holistic to more precise. What I will typically see on the right-hand side is, a little bit more process oriented whereas the left-hand side is more nebulous and relying on more tangibles and more calculations in your head. Either way, this spectrum starts on the left-hand side and ends on the right-hand side along that dimension. What we will do is go through each one of these. We will start with the look and guess and move towards the right.
- Method 1: Look and Guess
Let us start with the look and guess. I have a couple of funny pictures here because look and guess is basically looking at an assembly or gerber files and coming up with a number. Now that can be as sophisticated as looking at it and saying this is what it is, or I have seen some extremely talented, experienced EMS professionals being able to look at something and come up with calculations in their head that somebody who is spending an hour would not even be able to come up with. This is not to say that it is highly inaccurate because it depends on the individual. What it does mean, is that it isn't a documented process, it is not easy to replicate. Some characteristics of this methodology are that using the look and guess approach, you really want to have an expert doing the labor estimating. This would be someone who has been doing this for 10 to 20 years or more. Typically, the profile of the person who uses this approach is someone who owns or started the company so they understand all the nuances. They are well experienced with the customers and they are well in tune with what the production capabilities are. It does require a little bit of executive decision making and a really good memory. For example, if I look at an assembly and price it at $100 dollars today for labor, then I would want to make sure I am consistent tomorrow and the day after considering the same things. What we typically find using this approach, is consistency is rare meaning I can put the same assembly in front of a room full of people using the same approach and they can all have 20 years or more experience and all come up with a significantly different number from each other on how much to charge for that particular assembly. As a result, because the precision goes down, it also makes it hard to replicate how it gets done. So for example, if you have a person in your company that does all the labor quotes, it makes it hard to delegate any of the labor estimation as your company grows or as your quote volume increases. The good side of this approach is that it is much more holistic meaning that somebody looking at an assembly drawing or something with 20 years of experience could look at it and say, "Hey, this is something a little bit weird compared to most of the assemblies we have worked with or there is a capabilities gap here," or just anything like that that would stand out to someone who is well-experienced, well versed and looking at it on a more holistic level.
- Method 2: BOM Cost Plus
Moving on, the next thing that we are going to look at is the BOM Cost Plus method. This approach is that you take the material dollars and you apply some sort of percentage mark up to it. It is a pretty common practice across the entire industry to mark up materials a little bit to count for just the risk and material handling and holding cost. What this approach says is the material markup is going to encompass my labor, my overhead and my material markup, all-in-one. The examples we have here are three different BOMS with different material costs and about a 20% markup on each of these. One thing that I will point out, is that you wouldn’t have a consistent 20% markup on the materials. Most people that you would see in this approach would say that depending on the quantity being quoted and depending on the complexity of the board, and a variety of other things that are more intangible, what they would do is adjust the percentage markup. They might say that it is a 10% markup or they might say it is a 40% markup depending on a variety of things. What that means is that you want to have somebody who has a pretty significant amount of experience so they can determine what that percentage markup is because otherwise you are just picking a number out of thin air in terms of what percentage of markup to use. The other thing that I would say as a result of the percentage markup, some of you have probably come across this situation before, where a customer comes to you and says, hey, I can buy all of these materials for 2 dollars. Why are you charging me $20 for this assembly? Well, the material costs, in my opinion, is not necessarily a relevant consideration in the labor. So for example, I can have an extremely expensive component and it could be very easy to put on the board. While my material cost should be high, maybe my labor cost does not need to be proportionally higher. As a result, because you are arbitrarily picking a number, most of the people that use this approach are doing the look and guess approach and have a number in their head. They are just using a percentage to markup to justify how they got that number in their head. Another thing that I would say is that the benefit of using this approach is it is relatively quick meaning you can take the bill of material cost multiply by a percentage pretty easily. The flip side is that now you have to wait for the BOM to get costed, and you're adding a dependency to the quoting process, making it a little longer.
- Method 3: Price per Part
Moving on to the third approach here. The third labor quoting process is Price per Part. Most of you have probably encountered this at some point or another in your career. The price per part basically says cout the number of placements and for each placement, charge some sort of dollar value to it. I have an assembly here and each component represents a dollar sign here so I could say, 25 cents per component, there are 100 components on the board, so that means I am going to charge $25 for labor. Much like the material markup approach, the dollars per component is something that tends to fluctuate and that fluctuation happens to be based on quantity and complexity of the board. Most of the time what people come up with is some sort of arbitrary measures. So they will say, “Oh for this board I am going to charge 40 cents per component but for this board over here, I am going to charge 15 cents per component.”
Usually, this is because they have a number in their head that they are trying to hit. This is just a way of saying that I have the formula to get there. It's not to say that is the situation in all of the cases, it is just what we typically see. The danger of this approach is that it does not take into account the modernization that has happened in this industry. So for example, if this was the 1990's and things were being done pretty heavily by hand or manually, then this would be a pretty valid approach because of the number of components was directly proportional to how much time it would take you to do the labor. Nowadays, because of equipment modernization, as well as automation in terms of business processes, what we typically see is that there is a high upfront set up cost and the per-component cost is relatively less than what it was 15-20 years ago. As a result, if you are using this approach, more often than not, what we see is that you are overpricing the larger quantities and underpricing the lower quantities because you are not taking into account the high setup cost and lower variable cost. If somebody asks you to quote 5 assemblies, you are going to be looking really good and if they ask you to quote 500 assemblies, you are not going to be looking as good.
- Method 4: Price per Classified Part
Moving forward, we're looking at Price per Classified Component. Very similar to the previous example. In price per part, every component was treated uniformly and Price per Classified Component is more advanced in that you are classifying the components in the BOM so you can break it down to say that some components are harder to place than others. That's what is exemplified by the dollar signs. So let's say the grey ones are BGA's and we are going to charge $3 per BGA whereas the smaller ones in the center are resistors or capacitors that the SMT machine is going to put on and we are going to charge 15 cents or 5 cents or whatever you choose per component there. The good thing about this approach is that you are getting into building a more processed oriented building approach. Rather than saying I am going to pick a number out of thin air, what you are saying is, “I’m going to classify the components, each component class will have some kind of dollar value associated with it,” so that you can be a little bit more consistent. You will still see some kind of variation as you get into higher quantities, so it does not necessarily have to be the same dollar value per component as you get into the higher quantities. Typically what happens in those cases, there will be some kind of price break, so you will say if the assembly quantity is one to 25 I will charge this amount, and if it is 26-100, I will charge a little bit less. So that is one way of taking into account the economies of scale. It still has a similar issue as the previous approach that you are typically going to overcharge for higher quantities and undercharging for the low quantities just because you can't fully capture the burden of setup costs using just this approach.
- Method 5: Activity Costing
Moving forward, we are on the 5th approach which is the Activity Costing approach. This approach says is that we are going to define some activities and each activity will be driven by something and will have a time associated with it and a labor rate and that is how we are going to come up with the cost of the labor. Notice that it does not have an allowance for the value of the component. Some activities are dependant on the number of the components like you see in the pick and place and others are not. The purchasing activity here, for example, is based on line items and not a number of placements. Now you are getting into the more advanced processes. Typically the people who are using this approach have some sort of spreadsheet template that they come up with. Basically, some standardized form that they use to be able to say that we are going to simplify the calculations here. So I wouldn't expect normally somebody to sit there with a piece of paper to come up with these calculations, instead it would be more like, I have an excel spreadsheet and I go in and put in my base assumptions of how many line items there are and how many SMT components there are, how many through hole components there are and how many PCBs there are per panel and double sided boards. Things like that are relatively easy to comprehend and then they will put in assumptions in terms of time. So, they would say, it takes me 2 mins per line item to do purchasing and it takes me 3 seconds per component to do pick and place. What I like about this approach is, one, it is more process oriented. In the previous approaches, we were talking about somebody who has an extreme amount of expertise in the industry, somebody who can look more holistically and deal with tangibles. In this one, it is better suited for a company that has a larger amount of quote volume. You don't have the capacity to only have experts do the labor estimation, instead, you can start to delegate it and leave a final review step. The nice thing about that is, now you are delegating this to somebody so you can process more quotes but you are also making it more comprehensible. What I mean by that is, it is much easier for somebody that is doing the quoting to say it takes 2 minutes per line item to do this operation. By comparison it is harder to say, “this activity is worth 37 cents per part.” The other thing that is nice about this, if you break it out like this, you don't have to strictly stick to shop floor rate concept. We do see a lot of people using a standard shop right rather than an activity rate. This approach does not require that. So I could say that purchasing here is 15 per hour but other places are 35 an hour. It also gives you more flexibility in terms of how you allocate your overhead. For example, pick and place requires me to buy some really expensive machines, I can allocate more of my overhead to that if I wanted to vs purchasing where all I need is a good buyer and maybe a computer.
- Which EMS Labor Quoting Method do you use?
We have talked about five different approaches and now I am going to kick off the next poll to see which approach everyone uses. Feel free to pick one of them. What I would expect is, that given our audience and the customers that we typically engage with, most people would fall into the activity costing range. Those are companies that are on track to scale or trying to build processes that are repeatable and that's where they are looking at attending webinars, learning and building a skill in the organization and implementing software. Based on preliminary results from the survey, it looks like most people do fall into that category. I am going to end the poll. It looks like we have more than half of the results in, but I think that the crowd has spoken. Overwhelmingly it is that everybody uses the activity costing approach. That is definitely my preferred approach, that's what we recommend to Calcuquote customers as well and is more repeatable, consistent process.
- Activity Based Labor Quoting Model
Moving forward, we are going to dive deeper into the activity-based model. One of the benefits of using the activities based approach is that it considers the activities that are going into building the assembly. What that means is, you can get your routing partly done. It is not going to be perfect in terms of work constructions. What is good about it is that in addition to having your labor estimation done, you also have a foundation that you can pass on notes to your doc control or engineering department to where when they come up with the routing, they can look at the quote and say, “Oh, this quote included an allotment for a conformal coating so let's make sure the routing has conformal coating.” It is a good way to make sure that the proper routing is created based on the quote. If the quote did not include conformal coating, and the work instructions do, it would be pretty easy to reconcile between the two and say there is a mistake on how we do the quote here. The second thing, it is more comprehensible to use this approach because you are thinking in terms of time rather than in dollars. If somebody asks me how long does it take to get across town, I can say oh it takes me 30 minutes, or if someone asks me how much dollar value it takes to get across town, now I have to think about the depreciation of the car, the gas, my time. Just too many variables to process in my head. So the activities based approach is much more comprehensible in that way. The other side of that is that it also makes it more accurate reflection of your baseline cost. We talked about a couple of those labor models where you run the risks of not accounting for all of your setup. With this approach, because you are separating out the labor activities in terms of setup vs. variable, what that allows you to do is you can actually amortize the setup across all of your units which makes it more accurate. While you might look a little bit higher at the lower end of the quantity, you are more accurately reflecting what your actual cost will be. The second side of this, which is a really nice side effect, now I can very easily look at this and say I had quoted 30 min for SMT set up, let me go on the production floor to validate my assumption. I can say did it actually take the production floor 30 min to set up the SMT machine or did it take 2 hours. Now you can continuously revise what your base assumptions are, what your labor model assumptions are in a more intuitive fashion.
- Labor Model Creation
The next thing we are going to do is talk about how to create a labor model like this. Since most of you are already using something like an activity-based model, I am going to spend very little time on this. Basically, we like to suggest a 5-step approach here. I have identifying labor activities, categorize and setup variable activities, determine the drive, establish a standard time and calculate a labor rate.
- Activity Costing
So looking at the examples before. Steps one and two are to the left where I am going to list all the activities and come up with whether they are set up activities such as purchasing which you will only do once or variable which is something you would have to do for each board. Then you have your driver, time and labor rate.
- Efficiency Gains for Volume
The final thing I would like to talk about is Efficiency Gains for Volume. Efficiency gains is different from economies of scale. We see people using that interchangeably. The distinction I like to draw is that the economies of scale is how you amortize your setup cost across your assemblies. If I have $500 worth of setup cost up front that I have to deal with, then the economies of scale would be building 500 units vs. 5 units where I can amortize that $500 into those 500 units. So, it can give me a pretty significant benefit to order a larger quantity. The efficiency gain, on the other hand, is the improvements you make in your process through repetition. For example, with inspection, the first board I inspect might take a little longer, but over time, I get better, faster and learn how to do it better. As a result, the time it takes is going to go down as well as the dollars it takes will go down. There are two approaches to this. The first I have exemplified here, the stop approach where from 1-25 I charge a certain amount for inspection and for 25-100, I charge a slightly lesser amount and so on as you get into higher quantities. This approach is probably the most common one we see when we are first talking to an EMS company. What we move people to is a learning curve approach.
- Learning Curve
The learning curve approach starts to gradually reduce that amount of time or dollar value as you get into higher quantities, so you don't see a huge drop off between 25 and 26 pieces because there is no magic behind the 26th piece vs the 25th piece other than that is just an arbitrary number in that step approach. So the learning curve gradually decreases it so even if they are asking for 16 pieces, somewhere in the middle, you properly allocated that you are going to gain efficiencies on that approach. The other thing about the learning curve approach, which is usually true for the step approach as well. You learn pretty quickly at first, so you start to see pretty significant drop-offs in the beginning, and then it levels out over time. The learning curve we would recommend is the U learning curve model. This isn't something that I have invented or Calcuquote has invented. This is something that is an operations management principle. Basically, the U learning model has a simple formula. Hn represents the hours required for the Nth unit of production. It does not have to be hours. It can be minutes, or seconds or whatever. The point is, if I am building 100 units, that n would represent the 100th unit and how many hours it would take to build the 100th unit. The H1 is the first unit of the production, so if I say it takes me 10 minutes to make the 1st unit, then that is what I would put for the H1 values so that would be a known value you would have to populate and then the Nb on the right, the N would be the hundred unit, and b is the natural slope of learning. This is a pretty abstract concept, so what we like to do is think of it in terms of a scale from 1-100 and that is what we call learning curve percentage. Basically what this number does is that it determines the gradient or slope at which it is going to gain efficiencies. A shallow learning curve would gradually drop, whereas a steeper learning curve would drop more rapidly to a bottom out stage faster.
- U Learning Model
The learning curve we would recommend is the U learning curve model. This isn't something that I have invented or Calcuquote has invented, this is something that is an operations management principle. Basically, the U learning model has a fairly simple formula. Hn represents the hours required for the end unit of production. It does not have to be hours, it can be minutes, or seconds or whatever. The point is, if I am building 100 units, that n would represent the 100th unit and so the how many hours it would take to build the 100th unit. The H1 is the first unit of the production, so if I say it takes me 10 minutes to make the 1st unit, then that is what I would put for the H1 values so that would be a known value you would have to populate and then the Nb on the right, the N would be the hundred unit, and b would be the natural slope of learning. This is a pretty abstract concept, so what we like to do is think of it in terms of a scale from 1-100 and that is what we call learning curve percentage. Basically what this number does is that it determines the radiant or slope at which it is going to gain efficiencies. A shallow learning curve would gradually drop, whereas a steeper learning curve would drop more rapidly to a bottom out stage faster.
Overall, we would be happy to talk in more detail if you have any questions. The contact info is there at the bottom so we can definitely get into more details about what this involves. The other thing I would say is if you're looking for more help on this, we would be happy to show you some worksheets that we have built up over time.