A working prototype doesn't mean you're ready to build a hundred of them.
When you make a mistake on a prototype it costs you one board, but in a production run you copy that same mistake a hundred times.
That can mean a hundred boards to rework, or a hundred units sitting in boxes that nobody's waiting to buy.
The good news is that your first 100 units is one of the most exciting steps in this whole process, because it's where your project turns into a product that people can hold.
So here are the 7 things you must do before you build them, including a way to get production-quality enclosures that almost nobody knows about.
The last one on this list sounds like something you do after the units are built, but it has to start before you build a single one.
Read or watch the full version here
#7: Scale up in stages
Your first 100 units should never be built as one big batch.
You want to work up to that number in stages, where each stage is a little bigger than the last one.
For example, you might build one or two prototypes, then four or five units, then ten, and then the rest of the hundred. Those exact numbers aren't a formula, so pick the quantities that make sense for your product and your budget.
The rule that matters is that you only move up to the next quantity once the previous stage is proven to work.
By "work" I mean every unit was put together using the process you plan to use going forward, and every unit passed its tests. It also means none of them needed a special fix, like a jumper wire on one board or some extra filing on one enclosure.
If even one unit needed that kind of hand fixing, you've found a problem that's going to show up again and again at higher quantities.
Design changes will happen along the way, and you want them to happen between stages, never in the middle of a batch. That way every batch is a single known version of your product, and when something goes wrong you know exactly which version you're looking at.
Your parts need to be ready to scale too, so check that every part on your board is in stock at the major distributors and isn't close to being discontinued.
Then price your parts at 100 units, and again at 1,000 and 10,000, against your target retail price. This is the cheapest time to discover your product can't be profitable.
#6: Use the hybrid manufacturing approach
Hybrid manufacturing is the best way to build your first 100 units, and it just means you split the work between professional manufacturers and yourself.
You have your circuit boards professionally made and assembled, since placing tiny parts by hand a hundred times isn't a good use of your time. In most cases you'll have the enclosure manufactured for you too.
Then you bring those pieces together, and you do the final assembly and testing yourself before you box each unit up.
The other option is a turnkey manufacturer, which is a company that handles everything from the bare boards all the way to the boxed product.
That sounds a lot easier, but you give up two valuable things when you go that route this early.
The first is learning, because your first production run is where you find out which assembly steps are awkward and which parts don't fit the same way every time. It's also where you learn what needs to change before you scale up to thousands of units.
When you hand the whole thing off, those problems still come up, but they come up inside a factory where you can't see them.
The second is quality, because when every unit passes through your own hands, you have 100% control over what ships. If there's ever a time you don't want a bad unit to reach a customer, it's on your first 100 units.
One hundred units is a completely reasonable number to handle in-house, even if you're doing it at your kitchen table.
It's going to take some evenings and weekends, but the control and the learning are worth far more than the labor you'd save.
#5: Get a production-quality enclosure without paying for mass production
Enclosures are the hardest part of a 100-unit run, because at first it looks like you only have two choices and both are bad.
A 3D printed case with visible layer lines looks unfinished to a paying customer. Steel injection mold tooling can cost tens of thousands of dollars, which makes no sense for 100 units.
Thankfully there are options in between, and the first one is urethane casting, where liquid plastic gets poured into a soft silicone mold.
Cast parts are molded just like production parts, but they're made from a different plastic with its own look and its own characteristics. That means you need to confirm the cast material meets your requirements, instead of assuming it will.
CNC machining is another option, where your enclosure gets cut out of a solid block, and it has the advantage of using production plastic. It does have limits on the shapes it can make though, and you can sometimes see machining marks on the part.
Then there's the method almost nobody knows about, which is 3D printed mold inserts made from a high-temperature plastic or even ceramic.
An injection mold has a standard metal frame called the mold base, and inside it are the inserts that form the shape of your part. With this method, the inserts are the only pieces that get 3D printed.
You end up molding the same plastic you'd use in production, at a fraction of the cost of steel tooling.
It does have limits, because printed inserts wear out fast and they're less accurate than steel, and there are some shapes they can't handle. When you run the numbers, be sure to include the cost of the mold base and the replacement inserts.
Aluminum molds usually aren't the best option for only 100 units, but they're the next step when you're ready to scale up, before you move to steel molds.
#4: Create a repeatable assembly, programming, and test process
A single working prototype only proves that you can make one unit work.
For a hundred units you need a process that gives you the same result every time, without you troubleshooting each unit.
That process starts with written assembly instructions, even if you're the only person who'll ever read them. Writing each step down forces you to decide on the right way to do it, and then you do it that same way on all 100 units.
You'll also need a set procedure for loading the firmware, plus a calibration step if your product has sensors that need one.
Make sure there's a way to update that firmware after the product ships, whether that's over USB or wirelessly, because your first customers are going to find bugs that you missed. Without that, even a one-line firmware fix means getting every unit shipped back to you.
Then comes a functional test with clear pass or fail criteria, so there's never a judgment call about whether a unit is good enough.
Every unit gets that test, and every result gets logged against that unit's serial number. That log pays off later, because when a customer reports a problem you can look up exactly how their unit tested before it shipped.
Let's say it takes you 20 minutes to program and test each unit by hand. Across 100 units that's over 33 hours of work, and that's before you troubleshoot a single failure.
A simple test fixture, which is just a jig that holds the board and connects to it, can cut that time way down and pays for itself almost right away.
#3: Test it in your customer's environment, not your lab
Your lab is a comfortable place for a product, because it stays at room temperature and nothing ever gets knocked around.
Your customers are going to leave it in a hot car or use it in a steamy bathroom, and some of them will hand it to a destructive toddler.
So before you approve the rest of the batch, take a few units from your early stages and put them through reliability testing.
That means running them across the full temperature range, exposing them to humidity, dropping them, and copying whatever abuse your product will see in normal use.
The functional test you run on every unit only asks whether that one unit was built correctly. Reliability testing asks whether your design can survive the conditions it's going to live in.
I learned this the hard way when a major retailer requested a sample of my own product, and I shipped them a 3D printed prototype during a summer heat wave.
The buyer told me it was awkward to use, and it turned out the prototype had partially melted during shipping. It hadn't melted enough for him to notice, but it was enough to make it awkward to operate.
Now imagine finding that out after 100 units are already in your customers' hands.
Failures like that turn into returns and bad reviews, and those are hard to recover from when you're just getting started.
#2: Run an EMI pre-scan and confirm your compliance path
Certifications like FCC have to be finished before you sell and ship your product, but they don't necessarily have to be finished before you build your first 100 units.
What you do need to know before you build is that your design is able to pass.
A big part of that is EMI, which stands for electromagnetic interference, and it's the radio noise your product gives off. You can check this early by running a pre-compliance scan, or pre-scan, on your prototypes.
A pre-scan is a rough version of the official test, and the main difference is that it doesn't need an anechoic chamber, which is the special shielded room that blocks out all outside radio noise.
That means you can run one in a normal lab with affordable equipment, or pay a local test lab to do it for you.
It won't be as accurate as the official test, but it'll catch the big problems. If the scan finds one, you can fix it while it's still just a layout change, and not after 100 boards already exist.
If you plan to sell these first 100 units, then they need to be certified first.
One way around that is to run customer trials, where you loan the units out for free instead of selling them. The conditions are that each unit is labeled as not yet certified, and that you get the units back when the trial is over.
You also can't advertise those units as being for sale, but you can still share updates and build your waitlist.
You can even take preorders before certification, as long as you tell buyers up front that the product isn't authorized yet and that it only ships after certification.
So have your test lab confirm the right path for what you're planning.
AC line voltage and lithium batteries each come with a requirement that applies even to a small batch.
If your product plugs directly into AC line voltage, then safety testing should be done. It's mostly not required by law in the US, but you won't get liability insurance or retail placement without it.
If your product has a lithium battery, you need UN38.3 test documentation to ship it, which is the United Nations safety test for transporting lithium batteries. The simplest path is to buy a battery pack that already comes with that test report.
#1: Build it with customers, not for customers
Customers need to be part of your first 100 units long before those units exist, and I saved this one for last on purpose.
Everybody's heard the advice to talk to customers, but this goes further, because you build it with customers and not for them.
It starts while you're still developing the product, and not after it's finished.
You share the journey in public, setbacks included, and you run a waitlist so people can raise their hand. By the time your first 100 units are ready, you've got people who feel invested and who are excited to get one.
Those first customers are also how you get answers to the questions that matter most.
You need to know whether they can set it up without your help, what confuses them, what fails, and which features they end up using.
To capture all of that, set up product registration and give failed units a way to come back to you. It also pays to hold a few units back so you can reproduce whatever customers report.
Producing 100 units that nobody is waiting for is the most expensive mistake on this list.
The most valuable things you get from your first 100 units are information and momentum, and they matter more than the revenue.
That momentum also sets you up for the next step after your first 100 units, and for most products that's a crowdfunding campaign. Crowdfunding gets you sales before you build in high volume, so that money can pay for things like your certifications and your inventory.
If you'd like help getting your own first 100 units built, you can get expert help from me and other engineers inside my
Hardware Academy.