This is the basis of a daily struggle startups deal with constantly. The product you want to manufacture is cost sensitive so it needs to be manufactured in volume, but you haven’t validated the market to justify the expenditure of manufacturing in volume. You have a classic Catch-22. You could raise money by convincing financiers that the product is that good with an untested market and risk the large production run. Or you could start smaller, with less capital, and sell fewer at a higher cost and lower margin. We will focus on the latter option.
First Sell 1
Many entrepreneurs and startups are so focused on the big picture. It’s great to have vision, but can you sell 1? Go out and try to sell 1. You’ll learn a lot. In short this is called validation. Once you’ve sold 1, sell 10. See how that goes. Lose money on them if you must, but sell something. It’s important at some stage you get money to change hands. Many ‘customers’ will be happy to use your product for free. If your product is providing a value to your customer, they should be willing to pay for it. Paying customers are more likely to tell you the truth about your product.
Now you have the problem that you need something tangible to sell them. If you are going to sell 1, you now need to make 1, and keep in mind that this order isn’t necessarily important.
Go out and try to sell 1.
Design for 1
Many would just consider this a prototype, but it should be a little more than this. It doesn’t need to be at the level of a MVP, but it should be close.
For electronics you are buying mostly off the shelf products, assembling, and making it work. Program it up, add your special sauce and go.
Mechanically, you’ll be looking at 3D prints and traditional machining for plastic and metal components. It doesn’t matter what this version costs to make, the point is that you made it, and you can sell it.
This version is usually built by the engineer who designed it.
Design for 100
Now that you’ve sold a few and have received invaluable feedback from your early customers, iterate. Design at this level will likely involve a combination of custom and off the shelf components. There may be some design compromises, but it’s starting to get to where it should be. It’s still not perfect, but you should be able to start to produce your product and sell it with a little margin.
Electronics are usually a combination of modules and custom circuits. Assembly is still cumbersome and only partially automated. Engineers or skilled techs are still intimately involved at this stage.
Mechanics have not changed much from the first stage. However for plastics and some metals, casting becomes viable. This process can start from a 3D print and it is used to create mold, which then liquid plastics or metals can be poured into the mold to create your parts.
Overall assembly can start to be handled by technicians, but the engineer will still be close by.
Design for 1000 to 100,000
Now we are cooking. You have feedback from your customers, you’ve tweaked and refined the design, you know what is going to sell, and now you want to make a lot of them.
Electronics are full custom to the chip level, no more modules. There may be some custom components used like ASICs or LCDs, but still electronics are mostly off the shelf or commodity.
Mechanical aspects have moved into high volume production. Usually plastics are injection molded at this point, metal designs are casted (investment or die-cast), optimized for machining, or possibly re-engineered to plastic to take advantage of injection molding.
The assembly line utilizes a lot of automation and as much unskilled labor as possible. The process is very important as well as quality control. A fall out of 5% may be acceptable at low volume, but at higher volumes, you’ll be spending a lot of money manufacturing trash.
Design for 100,000 and Beyond
When your design is manufactured in this volume, you have reached a HNL (whole ‘nother level). Think of Apple. They don’t spend much time looking around at what they buy to integrate into their devices, they just design, build and manufacture whatever they need. If they want a button, they make it. Custom silicon or screens, they specify it and work with vendors to get it made. There are still commodity electronics components used, but only when they are a perfect fit for what they need.
The Math
How you design a product is directly related to the quantity the product is manufactured.
Here is the key, the setup and customization charges amortize out across the production run. We call this NRE, non-recurring engineering (more below). In every level of design and manufacturing, there is NRE, this leads to the mentality of the more you make, the less it will cost, down to a minimal level. Here is how the math works.
NRE + cost per unit * volume = total expense / volume
or
(NRE + (cost per unit * volume)) / volume = actual cost per unit
As volume increases, the NRE term becomes insignificant. Don’t believe me yet? Here is an example.
A client has spent $100,000 on engineering (NRE) and wishes to produce 2000 units. The units cost $100 each at that volume. Plugging in the numbers, the total expense is $200,000. Dividing by the volume, the real cost per unit is $100. They all sell and the product is a success and the client determines they should build 100,000 units next time. Without more NRE, and assuming a 20% reduction in cost due to volume, the total expense is $4M or $40 each (remember we don’t need to amortize the NRE again). That sounds great! Look how the cost dropped. However if you spend another $100,000 in NRE and the cost per unit drops to $25, the total expense is $2.6M and the cost per unit is $26. Now that is savings. Spending money to save money.
Spend money to save money.
NRE
NRE can take many forms. It may be a simple design change from a module to a discrete solution. Modules cost more because they have added value and the hard work has been done. If you put in the hard work (NRE), you can reduce your BOM cost by purchasing the less expensive discrete components rather than a module. Another example on the mechanical side is when moving from 3D printing or castings to injection molding. ‘Tooling’ is the general term for creating a tool to make a part. Injection mold tools can run from $2000 to over $100,000. The parts they product may only cost pennies to a few dollars. You are basically paying for the raw material of the plastic and the time, which is much less than the labor and time intensive other methods.
A big place where NRE can pay off big is with DFM, or design for manufacturing. Making strategic design changes to reduce assembly time and to introduce automation whether with software or machines, can reduce your manufacturing costs, especially with labor intensive product.
A note on volumes
Every product is different. We’ve had clients where 100 units a year was a massive business and others that weren’t really doing until they hit 10,000. The numbers here are relative. The important thing here is to validate and change your design as your volumes increase. In some cases you may have more custom design at low volumes or source items off the shelf at high volumes.
The takeaway here is that it takes money to save money. Figure out where you are in this process and design and manufacture accordingly.