How was it made, consistently and well?

How to manufacture if you’ve never manufactured before.

An intro to product

8 min readMay 22, 2018


If you identify with this title, you are probably working at a startup.

In the startup world, there are bits and bytes. Code and physical things.

The goal in a startup is to rapidly alter your product until your potential customers love it so much that they can’t stop telling all their friends about it. I can’t speak for most companies, but most experts say companies are very sloppy in this phase.

“Can’t stop talking about it,” not “Like it reasonably well.”

Once you’ve hit upon a valuable product you rapidly scale.

In the startup mentality, as you scale, you are still expected to perform tests and updates on your product.

This idea is drastically at odds with the old manufacturing style of trying to make a single thing in the most efficient way in large quantities.


With code these updates are easy. You have developers which can deploy code in sandbox environments, eventually moving on to live tests. This means you can test and break code rapidly on a reasonable scale. Then, once you deploy it, you can track, live, how your users interact. The speed of feedback and speed at which you can change a user’s experience is fascinatingly effective with code.

There is still a lot to be learned here, a lot of complexity, and a lot of expertise needed, but with code, you could probably deploy a new product iteration every day, or every week, at your early phases, to every single customer you have.


Physical products trail behind code in this type of agility. To copy something, it must be physically made and then delivered to the customer. If it takes 10 hours to build a box, and you ship out 10 boxes to potential customers, you either have to spend 100 hours building and shipping new boxes, or you have to spend time figuring out how to make the boxes in 7 hours.

This would be easy math, except every time you update the box to adjust for user feedback, you may end up adding a new part that you haven’t figured out how to make yet.

My serious advice here is, if you’re building a functional consumer product, to hack together a version of it that accurately and reliably performs the function needed. You’re looking for Aha! not Mhm. Once you get that reaction, you scale.

Launching into a serious manufacturing run before you a) have a product you know the consumer really enjoys and b) have a price the consumer has already indicated they’d pay (either through competitor products on the market or by actual consumer purchases of your prototypes/small run products) you may run into some serious issues.

These issues include:

Retrofitting lots of old inventory to have a new update

Attempting to sell a product that no one actually wants

Selling a product that people like, but don’t actually want to pay full price for


Most startups deal with these issues by getting funding. That funding allows them to burn through capital while they figure out the answers to a lot of these questions. However, you can avoid burning through at least some capital if you learn a bit about manufacturing first.

Real companies, who have been manufacturing for a long time, do lean. By that, I mean they engage in Lean Manufacturing practices.

There is an idea of one-piece-flow. This idea will be critical for you when making a product that is still in the update phase. The basic idea is that if your product has 50 processes associated with it, you set up your facility to run each of those processes in order. You document each step. You print and post those steps in front of each station, and you develop quality checks to make sure each step has been performed properly. Once one step is finished, it should be of high enough quality that the next step will not run into issues because the previous step was performed incorrectly.

To do one piece flow you will need to know:

What parts your product is made of and multiple vendors from which you can source those parts

What tools you need to assemble or create your products

What vendors you use to order custom parts

How many of each part you need to create a product

Alignment from the executive team

A facility that is built to hold inventory, contain each workspace, and contain work in progress (you do not want to be setting-up/breaking-down stations, or looking for parts and tools)

A facility that is up to OSHA safety code (any employee can whistleblow against you for violating OSHA standards and you legally cannot fire them for doing so)

The downside to one piece flow is that it requires organization and a specific alignment to that process. If you are working in a disorganized environment, or an environment where your team is not aligned with your manufacturing goals, expect to be managing chaos instead.

The upsides to one piece flow is that product essentially goes through the chain. You don’t have to worry about Gantt charting your workflow once you have all of your inventory in house. You reorder based on low inventory, rather than trying to predict demand. Work-in-progress can be tracked by keeping it in a traveler (basically a bin with a piece of paper saying what parts/qc processes have been added/performed.) Once you have the system set up, the management overhead is pretty low.

The other important upside to one piece flow is that instead of batching 20–100 parts and then having to go back and make updates, you should hopefully be producing close enough to the rate of demand that as you make updates, you hopefully haven’t overproduced and thus most of the outdated product is already sold. As your goal is to build a product worth buying, already sold inventory is not really worth worrying about.


Any computer automated process is your best friend, if you are able to get time on machines. Why? It is much closer to the agility of software development. Change the design. Put it in the machine. Wait. Instead of: change the design, physically make the part yourself.

Not all processes fit this model. But you will save yourself tons of time if you start manufacturing as many prototypes with computer assisted processes first, and move on to handmade or external processes later.

Make as much with off the shelf parts. Short of that, modify off the shelf parts, or laser cut and 3D print as much as you can. You can also 3D print molds, but I have much less experience with this process and cannot speak for it.

In the small to mid batch phases, outsource

There will be a point where you think you can handle the amount of work coming in, but it will get increasingly complex.

Most manufacturers I have seen, who are incredibly successful, master one process or one machine, then add more. People love bagging on the Juicero. I really can’t tell you much about it, except that it was a complex product. I would assume that at a certain scale, they outsourced all or almost all of the parts, and then assembled in house. I can’t tell you for sure. As a first time manufacturer, it would have been hell for me.

A business card company is a great example of a simple manufacturing company. To cut business cards on a very simple level, you need a printer and a guillotine. All you need is a consistent source of cardstock and ink. There’s a significantly larger amount of complexity to that process, but if you can master those two machines, you have the beginnings of a company.

The actual price structure for purchasing an industrial scale machine is that you can often:

Purchase a cheaper machine that doesn’t have high output but meets your current needs

Raise funding to purchase high-output high-quality machines

Use current profits to lease-to-own said machines

Take a business loan to purchase outright said machines

Completely outsource work for a specific process or an entire product

With the Juicero, you have to master vendor communication, supply chain for multiple parts, a significant amount of design and engineering skill to understand how all the parts should fit together within the range of tolerances that each vendor provides, and understand the limitations of each process in case you receive product errors that you have to fix.

If you have part wastage (lost shipments, broken parts, parts that do not fit and had to be redesigned and reordered, standard wastage, etc,) you usually make up for that by charging more for the product (yes it costs $399 for $50 in parts, but we spent another $25 in wastage the consumer will never see, and the rest is to cover operating costs in the business,) or you eat the cost, raise more money, and try to improve manufacturing processes.

It’s significantly easier to either:

Buy and assemble your product from 80%+ off the shelf parts, or CAD parts that you understand well.

Make sure your product is a simple product that can reliably be made from a handful of affordable and accurate machines.

If you are not doing one of those two things, do not kid yourself, you are choosing to build a significantly difficult product, which will only spin out into more complexity as you scale. It will be very difficult to “wing it” and you should definitely seek the help of one or multiple professionals that you trust to spend a significant amount of time guiding you through this process.

What people often assume with manufacturing is that once you master something, you will be able to scale and build cheaply, however that is completely untrue. The reason why companies vertically integrate is because they have a need for enough output, that they’d rather handle all of these manufacturing complexities themselves.

At every turn, there is some manufacturing homunculus lying in wait. I have been told that many successful project or product based companies have 20–25% management to worker ratio when building things, depending on the complexity. Many neophytes, myself included, often assume that you need 5–10%.

The issue isn’t the things that you know, the issue is the things that you don’t know. When you source a new manufacturer, they may finish your product differently because you didn’t specify, which then affects the process that your next supplier uses. There may be a tropical storm that wipes out one of your small parts suppliers, and if they are the only maker of that part, you must now remake your product with a completely different part, or wait for them to get back on their feet (months to years.) One of your workers with highly specialized and undocumented knowledge may quit, leaving you to reverse engineer everything they did. You may learn you rapidly need to add a new feature, resulting in a tight R&D schedule that slows down normal production. If you’re using a specialized material, you may only have access to a small handful of suppliers, none of which produce a level of quality that is satisfactory. The list goes on, and is practically never ending.

Your goal in manufacturing is to manage all of this complexity and get things out the door at a low cost.

Your goal in a startup is to rapidly change and adapt to measured customer desire until your product demand skyrockets.

These are completely oppositional goals. One removes complexity. One adds it. The only paradigms I’ve found so far that fits are:

One piece flow coupled with off the shelf parts and CAD parts

Low product/material complexity if at all possible

Short of those two things, millions or hundreds of millions of dollars to burn through and a ton of serious expertise within your team.

Be prepared. Seek out expert advice and assistance whenever possible. Lower your complexity whenever possible.

If you’re working out of a non-standard manufacturing environment, treat it like a manufacturing environment and organize it like one.

Take lean manufacturing training on a regular basis.