Everything I Learned Bootstrapping a Consumer Electronics Product
The world of consumer electronics is not a friendly place for bootstrapping. Consumers are used to paying low prices thanks to the economies of scale of mass-manufacturing. But bootstrapping typically involves small-scale manufacturing in which large fixed costs, such as $10,000 plastic injection molds or FCC testing fees, are no longer divided by large numbers.
Nevertheless, I waded into these waters with a small music controller called the VMeter. So far, I’ve sold about 1500 units, the first 500 of which were made by hand in the US, followed by two runs of 500 in China.
Although I wouldn’t call this product a success (just broke even so far), I learned a lot about business that went counter to the engineering instincts I had starting out. Here’s everything I learned about pricing, distributor discounts, risk, agile development in hardware, advertising, FCC testing (which I failed and then passed), packaging, sourcing parts, micro manufacturing in China, and why I’ll never give anyone a free sample again. I hope it helps other engineers striking out on the entrepreneur trail, and that more experienced travelers will pass advice back my way. (see also: technical design tour video)
Margins and Pricing: the 4x Rule
Your retail price should be 4 times the cost of making the product. This is something many others recommend, but it seemed extraordinarily high to me before working with this product. As an engineer, I assumed that if you made a better mousetrap and spammed reddit, the world would beat a path to your door. But the much more likely scenario is that you will have to pay 50% of your retail price to advertise. In the case of the VMeter, I give 40% discounts to distributors that buy 100 units or more, and then I purchase advertising on top of that to drive customers to those distributors.
If this seems like a steep price to pay for distributors, understand that they offer numerous benefits. Distributors hold inventory that enables you to do larger manufacturing runs. They give you an air of credibility. They can ship to customers in other countries without your customers having to wait weeks and pay import taxes. And if 40% still seems high, I was routinely contacted by wholesalers—people who buy and resell to distributors—expecting upwards of 75% off retail. The truth is, when you buy something at BestBuy, there’s a good chance that the majority of what you’re buying is some form of advertising.
The other reason to build in large margins is to cover risk. What if an entire manufacturing run has a problem that was your fault? What if a distributor returns a huge order? It’s not just about how much profit you can make, but thinking through a worst case scenario. If you lose 100 units, how many will you have to sell to cover that loss?
On my last run, all the top plastic pieces were yellowish and had paint that chipped off. I had to pay for all new acrylic and to have each package opened, fixed and re-sealed. Every run had some sort of multi-thousand dollar surprise like this despite my best efforts to inspect samples. Until you make the exact same order multiple times in a row, it’s probably prudent to plan on these types of surprises by building in higher margins.
Your First Product Will be the Wrong Product
Now that I’ve preached about this 4x rule, I should admit that I haven’t followed it (yet). My last run of 500 units cost upwards of $25 per unit (40% of the $60 list price), and the initial prototype runs were even more expensive. I’m close, though—a 1,000 unit run will get me down to $15.
And this brings me to one of the toughest questions to answer when bootstrapping a product: “How many should I make?” By making higher quantities, you reduce unit cost and increase profits, but you also risk being saddled with inventory you’ll never sell. This also feeds into the initial margin you’re targeting. If it will take a run of 100,000 units to reach 4x margin, maybe that product isn’t the best bootstrapping product.
While I don’t think there’s any right answer to this question, I want to make an argument for smaller numbers, even if that means temporarily breaking the 4x rule (but not for too long!). The primary reason for this is that you’ll likely need to make 4 or 5 versions, starting with an initial version that’s completely wrong.
I initially marketed the VMeter as a PC volume control. I sold exactly zero of these, and had many friends tell me they wouldn’t pay $15 for such a thing. I later rewrote the software to make it a music controller, (going back to its inspiration, the stribe touch controller), and had significantly more luck.
There’s a development framework from the software world called “Agile,” in which you basically accept that your initial product idea will be wrong, and that it will require many rounds of feedback and iteration to hone it into something useful. While new versions of software can be deployed at the push of a button, iterating is much more difficult in hardware, because long lead times and tooling charges make each change costly. Despite this, I still think you should try for smaller runs and faster iterations, validating features and use cases before ramping up.
Taken to the extreme, you can try to sell ideas even before making them, an idea promoted by the Lean Startup book. Had I done that, I could have learned that no one wanted an expensive volume knob for their computer by faking the software functionality in a video.
However, even if you can validate interest in an idea, there’s no substitute for feedback from real paying customers. As I sold 10, then 100, then 500, I continued to receive feedback about bugs and use cases I never ran into or thought of, all requiring both software and hardware changes. The VMeter was detecting touches from its backside because people were holding it in their hands. Every time a Windows 7 PC restarted the sensor went wacky. The pressure sensor didn’t always return back to 0. I only learned these things by having real customers—and lots of them—pound away at the device.
Given the high potential for change in any new product, I recommend being cautious about any investment, whether it’s the size of a manufacturing run, development time on a particular feature, or one of the many presumably necessary expenses: FCC or other compliance testing, manufacturing optimization (injection molds or other tooling), fancy packaging (unless it’s retail), and even circuitry to prevent rare problems, such as ESD (static shock) protection. The biggest risk is not that the FCC will fine you, but that no one will buy your widget.
Advertising
While there are plenty of products whose developers simply emailed engadget or posted to hackernews and received of tens of thousands (or more) of free views, I don’t think this is the norm (like not having distributors). Even with a lucky boost, you still need a long-term strategy to bring in new customers.
VMeter’s marketing strategy consists of two parts: banner ads on Google and SEO through tutorials. I wrote tutorials with excruciating detail on every configuration step for every major music program. Even though the tutorials were written for the VMeter, they can also be used for many other MIDI devices, and often show up as first page results for “MIDI setup xyz program.” These tutorials, however, took well over a year to generate much traffic.
On the Google Adwords side, I mainly relied on banner image ads and some video ads. The text ads were upwards of $1 per click, which only would have worked if the VMeter was $600, not $60. It took about 100 clicks to get a sale, so I tried to keep all clicks down to 10 cents (I suspect that text ads are more expensive because they’re easier to create). Text ads are also tricky for new products. Google ranks your ads not only by bid amount, but also by relevance. It will give preference to products that match exactly what people are searching for. But if you have a new product, no one is searching for it. Your best bet is to list competitor products or general keywords, and have these same words on your landing page.
More than half of my advertising was focused on driving customers to my overseas distributors rather than sales on my own webstore. I don’t count the distributors as part of a marketing effort, as it’s rare to get in their newsletters or homepages. Their main benefit to me was having them buy large quantities of inventory (which they paid for upfront), and making it faster for customers to receive product. On the topic of distributors, the majority of mine reached out to me rather than me sending an email to them. In some cases, I had customers in other countries that made the request, but I suspect many of those distributors just saw my ads.
Along with distributors, I also wouldn’t include Kickstarter as advertising. I had the idea that if you simply posted a project, it would automatically get noticed by the world, but this isn’t how Kickstarter works. Very few projects get included in their email blasts or highlighted on their homepage. To carry out a successful kickstarter, I suspect it takes a substantial advertising effort ($10,000 plus?) or an incredibly unique and impactful product. But if you have those, do you really need Kickstarter?
I never got to test a Kickstarter campaign because my project was rejected by staff before it ever saw the light of day (“not a good fit”). While I could have gone with another crowd-funding platform, I’m glad I didn’t. As mentioned before, I learned so much about my product, its use cases, and manufacturing during the first 1,000 sales. If I had taken a short cut to mass production, I would have mass-produced a lemon.
Looking back I wish I had spent more time on marketing rather than on engineering. As an engineer, the answer to the problem of “why aren’t more selling?” is almost always “add another feature.” When all your have is a hammer, everything is a nail. Another way to increase sales is to determine which feature your existing customer base already loves, and figure out how to find more people with that problem. Conduct surveys to see how people use it, write emails, and record famous people using it that same way. Getting better acquainted with customers is always a good idea.
The last piece of marketing advice I have is the one I feel most strongly about: never give out a free sample.
I gave away countless free samples to people promising me great publicity on their super popular blog, youtube channel, etc. Zero of those ever resulted in any publicity. I think this is because giving something free to someone tells them that it’s worth nothing. Instead of free samples, I started charging full price or giving a small discount at best. Almost 100% of people who paid for a VMeter wrote about it or gave me feedback. The irony is that I would have paid people to talk about the VMeter, but charging them was the only way to get that to happen. (Wired might get a free sample).
FCC Testing, UPC Barcodes & USB Vendor and Product IDs
Before I did my first run of 500 in China, I decided to get my device FCC tested. Based on the touch sensor design, I was worried I had potentially created an antenna that was broadcasting enough noise to interfere with other equipment (and get the FCC’s attention). I had also just read about a similar product’s company getting a $1 million fine for skipping the testing. Looking back, I’m not sure the cost was justified compared to the risk that I would need to change the product and get it retested, but I wanted to know if I was in the ballpark.
The VMeter failed its initial test, at a cost of $1200. I immediately panicked because I had no idea how to go about fixing it, and thought I would be stuck in a terribly expensive and long loop of redesigning and re-failing. But the testing company recommended that I have one of their engineers try to fix it at $200 an hour. While this also seemed scary, since I would be writing a blank check, and because they didn’t know anything about my design, the engineer fixed it in less than two hours. I had another version made with the extra parts he added and it passed. My total cost was $1800. I have a lot more details about this whole process in the accompanying video.
It’s amazing how many $1,000 to $10,000 certification fees you can pay when launching a new product. I had just gotten a large order from ThinkGeek, and they mandated that all products have a barcode. If you buy from the GS1 organization, this costs a minimum of $750. This gets you thousands of barcodes, which is ridiculous when you just need one. There are some resellers of barcode blocks, though. I bought one from qualityupc for $30. Some retailers, such as Walmart, won’t accept these resold codes, but if you’re selling to Walmart, you’re probably not in bootstrap mode.
Another similar fee for USB devices is for vendor and product identifiers. The minimum order is $2,000 for 65,000 product IDs. Instead, I bought a single one from mcselec.com for $10. While this may be something the USB organization looks down on, for a bootstrapped product, it doesn’t make sense to spend $2,000 on something that doesn’t directly help sales. I believe some chip manufacturers will let you use their USB IDs, which might be an even cheaper route. If you want to put the USB logo on your product, you may also need to get USB compliance testing done on your device. Not only is this another $5,000 or more, but the circuitry needed to obtain compliance might only be necessary to pass compliance, not to actually work (for example, the 47uF capacitors on the popular arduinos are over the max size allowed by the spec.)
Micro-scale Manufacturing in China
I’m not sure it makes sense to manufacture 500 units in China, or put differently, to place a $5,000 manufacturing order (my total cost was more since I was sending parts). It’s so very little money for a manufacturer to care about learning your product, and definitely not enough money to justify flying over there. But on the other hand, making that many in the US for me would have meant a profit margin too small to cover anything unexpected, or to afford advertising. I was willing to make 100 at a loss or by hand in my home, but 500 is a lot to lose money on and too many to make yourself if you have a day job.
On finding a manufacturer, I got very lucky. I had a friend who had made a similar product, and he connected me to his manufacturer (recommending us both to each other). I believe the more typical path forward is to find a broker.
On my first run, I shipped all the parts, including fairly heavy laser cut acrylic pieces. The manufacturer made the circuit boards, assembled and tested the units, and packaged them.
For packaging, all I had to do was send them a picture of a cardboard insert, and they prepared a vacuum molded case. It wasn’t Apple-pretty, but it looked like something I didn’t make in my garage, so I was very happy. All the package designers I contacted in the US all wanted thousands of dollars just for a design, but my total cost for 500 units of packaging in China was just $360. No tooling.
Since I shipped all the parts, I only paid the manufacturer about $3,000 for the PCBs and labor. I suspect I was extraordinarily lucky to find a manufacturer willing to deal with someone for so little money. I paid more to FedEx to have samples and final parts shipped back and forth on that first run.
Sourcing
On my next run, I tried to have the manufacturer source as many parts as possible. This was mostly a failure.
I assumed that he could source the electronics much cheaper than I could, since the parts wouldn’t be making a round trip from China to Digikey and back. I was hoping to avoid both Digikey’s profit margin as well as the 30% import tax on all of my shipments since those parts would be shipping right back out in three weeks. (To reduce that import tax, I marked down all my shipments to about 30% of their true value; this was risky as I was also marking down the insured value.)
The manufacturer’s price for most of the parts wasn’t that much cheaper, and they seemed unable to buy less than 2,000 processors when I only needed 500. The most expensive parts were the 38 blue LEDs. At 500 units, I was emptying Digikey’s and Mouser’s entire inventory (19,000). Kingbright, the manufacturer, was willing to sell them directly to me, but said they had to ship them to Hong Kong and back again to comply with import rules. In other words, so I could pay that 30% tax I was trying to avoid. I ended up buying most of the parts myself from US distributors again.
There were other sourcing issues, too, such as never quite getting the right shade of acrylic, or finding an epoxy silk screen that wouldn’t chip off, or getting the right kind of stickiness on the rubber feet. I ended up replacing all these items after the fact, and at a substantial cost.
I assume that if I was ordering 2,000 units instead of 500, many of these sourcing issues would magically work themselves out. There’d be enough money on the line for everyone involved to have more incentive in the hunt for parts, and maybe even enough to fill out whatever paperwork is necessary to avoid that import tax. The manufacturer was fantastically responsive to every email, and sent me many samples, but I think on the micro scale, I was better off just shipping exactly what I wanted.
I think the larger lesson with regard to bootstrapping is that the parts you build prototypes from in the US aren’t necessarily going to be the ones you can use in China. So plan on a major sourcing effort at some point in the ramp up, and ideally do this in partnership with your manufacturer.
With regard to timing, even shipping everything via FedEx, it took four months to get my first run done and two and a half months for the second. A lot of delays were due to things like testing samples, troubleshooting the chip programmer from screenshots in Chinese, and other first-time troubles. But even without those delays, it’s not uncommon for components to have eight to twelve week lead times. Planning on a lead time of four months or more is probably reasonable.
Conclusion
Bootstrapping a consumer electronics product is certainly not easy. You’re competing against products backed by huge corporations in a manufacturing world designed for minimum orders of thousands. But there’s potentially hope that being small and not having corporate red tape will allow you to iterate faster. The winning product is probably not the first one you release, but an idea triggered by that release.
I hope this helps some other entrepreneurs and/or engineers. I’m very much still finding my way, and would love any feedback from more experienced people.
If you’re interested in more details about the VMeter, it’s entirely open source, and there’s an accompanying video that details its technical design and development.
