The hardware startup: Manufacturing

At some point, all successful hardware startups will face the challenge of manufacturing their products at scale.

By Renee DiResta, Brady Forrest and Ryan Vinyard

May 14, 2015

Man fixing machine at Millville Manufacturing Co. (source: Wikimedia Commons)

Manufacturing

At some point, all successful hardware startups will have to face the challenge of manufacturing their products at scale. Manufacturing involves a bit of a different mindset than prototyping. Prototyping is full of experimentation and invention, while often burning money. Manufacturing is about process and, hopefully, eventually making money. Prototyping is making one of something; manufacturing is making something that can be made over and over again by people you don’t know.

This can be a hard and expensive transition for a young company with limited funds. Often, schedules slip, which means that it takes longer to gain revenue from selling the product. Also, working capital, the money used to fund inventory, is necessary to get the product manufactured.

Learn faster. Dig deeper. See farther.

Join the O'Reilly online learning platform. Get a free trial today and find answers on the fly, or master something new and useful.

Learn more

Here are a few terms you will find used to describe different kinds of manufacturers:

Original equipment manufacturer (OEM): A company that makes the original parts, components, or assemblies that go into a product.
Original design manufacturer (ODM): A company that specializes in manufacturing products for other brands. This approach is often known as a white label, since any company could approach the same ODM and sell the product under its own brand.
Contract manufacturer (CM): A company that handles the manufacturing and potentially the supply chain for another company. A form of outsourcing.
Electronic manufacturing service (EMS): A company that designs, tests, and manufactures electronic components and assemblies.

It’s important to note that these terms have the potential to be used somewhat interchangeably, depending on geography and industry.

Preparing to Manufacture

In your process, you’ll need to determine the best time to engage factories and CMs. If you approach factories too early with concept sketches and vague ideas, you might scare them away from your business, since they might not take you seriously. On the other hand, if you wait too long to engage a factory and already have a fully detailed design, it might not be able to manufacture your product in the way you want, or there might be higher costs associated with features than you thought.

Here are some terms you will encounter as you go through different stages of manufacturing:

Design for X (DFX): The process of making sure your final prototype can be be manufactured efficiently and within budget. This process comprises design for manufacture (DFM), design for assembly (DFA), design for test (DFT), and design for cost (DFC).
Engineering verification test (EVT): The first gate in manufacturing. At this stage you are normally hand-assembling parts from tooling to check engineering function.
Design verification test (DVT): The second major stage in building a pilot. You should have all tooled parts going down a custom assembly line at this point.
Production verification test (PVT): The final stage in production ramp before full-on mass production (MP). At this point, you are mainly focusing on the production line itself, not the product. Most startups will ship product coming off the line, since it is almost perfect.
Standard operating procedure (SOP): Detailed instructions of a manufacturing process spelling out each individual step. Sometimes referred to as MWI (manufacturing work instructions).
Approved vendor list (AVL): A list of vendors to a factory or startup that have gone through the appropriate vetting, audits, and screenings to be a trusted source.
Computer-aided design (CAD): The design files documenting ME and EE parts. ME CAD is normally transferred through STEP or IGES 3D files or traditional 2D drawings for critical parts. EE CAD is transferred through Gerber files and drill files for boards.

Orion Labs, formerly OnBeep, is a great example of a hardware startup that approached manufacturers early enough to get input but was still far enough along to explain to manufacturing partners what it was making. The following sidebar highlights some of their process.

Ramping Up Manufacturing at Orion: A Case Study

Andy Sherman and Star Simpson have both led Production Operations at Orion Labs, formerly OnBeep. Star ran production at Orion as it began ramping up manufacturing of its Onyx product. Andy came on board during this ramp, and Star transitioned to R&D. Onyx is a wearable group communication platform with a button, a microphone, and a Bluetooth connection. Orion engaged PCH Access to help with its first production run in China.

When selecting a partner for its first production run, Star recalls, “We were looking for the ability to scale.” She emphasizes that a startup should “look at what projects manufacturers have worked on and that informed their experience,” since you wouldn’t want someone who has never worked in your industry and would be unfamiliar with your processes. Orion ended up choosing PCH Access to help with its first production run. Having a partner helped alleviate the need for a large team at Orion to get through the first run.

Andy describes the key pivotal role in managing a partner as Orion’s “program manager, someone who works with either the contract manufacturer or the other team, on the ground overseas at your manufacturing site to some capacity.” Orion also needed technical staff in “electrical engineering support on the ground, mechanical engineering support on the ground, quality engineers so this way we can define quality and test plans.” These kinds of technical resources are important to keep in-house so you can control the technical specifications of your product—since only your startup knows what makes your product truly unique.

As far as partner support goes, Andy describes, “On the PCH side, we had a program manager that managed the team over at PCH. We also worked very closely with an electrical engineer on that team, with a sourcing procurement expert on that team, and a couple of quality engineers as well.” Having these resources available through a partner helped Orion keep its technical team lean, while helping production move faster by having local experts on the ground.

Moving from prototype to production also required changes in the design of its product. Star describes the changes necessary in moving from prototyping to manufacturing as “redesign of a board that allows you to test that in an automated way,” in the sense that not only does the board still need to function, but it now also has to be repeatedly testable. They learned specifics related to their part section, such as to “include LED variation within our devices so we had some batches that were interesting.” According to Andy, “We just weren’t very experienced with dealing with variations in bin codes and how we need to approach that from a manufacturing and SMT perspective, and make sure that the same bin code is found across each of the LEDs within any individual device.”

Orion went through two DVT cycles on Onyx to make sure all of these issues were smoothed out. Andy emphasizes the importance of improving a line’s efficiency once you get to the PVT stage. One way to go about this is to “do cycle time calculations on each stage and then see where we have bottlenecks.”

Andy reminds entrepreneurs getting into the hardware field that “in manufacturing, probably more so than many other disciplines, there’s no day off,” as you will have to deal with these kinds of issues that arise day to day on the line. Overall, Star and Andy emphasize the continual progress made throughout manufacturing and the importance of adapting your team and product from prototyping to manufacturing.

Approaching manufacturing, it is important to change your design to accommodate the necessities of mass production. This is lumped under the umbrella term DFX, encompassing design for manufacture (DFM), design for assembly (DFA), etc. This can involve minor design changes such as adding radii or nominal dimension changes for standard tools, or larger design overhauls, such as eliminating unnecessary tooling complexity by moving features to other parts or even eliminating parts altogether if their functionality can be replaced.

Another important consideration for your design is the kind of environments and situations your product will need to withstand. This part of designing is normally called ruggedization. The most common consideration here is the ingress protection (IP) rating of the product, which reflects how much water and dust a product is susceptible to through a two-digit coding system going up into the 60s, with the highest rating technically being IP 6K9K.

Other important ruggedization considerations include the shock and vibrations that the product will need to endure and any specific material or chemicals that the product will be exposed to, especially during normal cleaning.

Once you engage with a CM, it will want you to submit your design files so it can give you a quote. This back-and-forth process is known as a request for quote (RFQ), and it’s important to be prepared with any documentation or information that the vendor will need in order to expedite the process. A typical RFQ will include the technical design files for whatever you are looking to make as well as accompanying documentation setting expectations for process, tolerance, quantity, timeline, and any other requirements. If you are working with a full, turnkey CM, it will want everything, including your ME CAD, EE CAD, product requirements document (PRD), BOM, and possibly even your business plan or financial information if you are starting a larger relationship with that CM.

The bill of materials (BOM) is essentially a list of everything that goes into your product with the goal of capturing the cost of goods sold (COGS), the cost it actually takes to produce your product. The BOM should include the basics of your ME and EE components (PCBs, board components, mechanical enclosures, fasteners, etc.), as well as the packaging and all of the accessories you will be including in the box. The COGS is the bottom-line cost you will be paying per unit, including all assembly, finishing, packout, taxes, tariffs, and markups from any partners. Your BOM and associated COGS are important to start compiling early in your project to get quotes and understand what components are driving costs, but you realistically won’t have a final cost until you are fully in mass production.

Another document that will become important at this point is your product requirements document (PRD). Typically, this captures the specifications and tests that your product will need to pass. Larger companies will start with a higher-level market requirements document (MRD) from the marketing team that will contribute to development of the PRD, but the PRD is often the first formal product documentation a startup puts together. It can be helpful to start putting together your PRD early in your development process, so that your engineering team has product goals to shoot toward as well as a communal document to capture qualitative considerations that can’t be captured in CAD or a BOM. Your PRD can be a good catchall document during development, to capture the intent of your team and make sure everyone is on the same page about what your product is actually supposed to do and how it will perform.

Keep in mind that quotes are free but take time and resources on both sides to produce; make sure to get a few quotes, but don’t waste too much time on this initial process. You will want to save this time to invest in building a relationship with the CMs and factories that you end up choosing.

Once you’ve released your design and issued the initial purchase orders for production, you’ll need to pay for any major deviations that you would like or that the design requires. This is why it’s important at some stage to put a clear freeze on your design and officially release the final design. Once this official release happens, you will need to capture changes formally through numbered change orders (often abbreviated CO, or ECO for engineering change order). Change orders often have a fee associated with them, unless it’s simply a part number or documentation change.

Another important part of building this relationship is to communicate clearly with your factory, especially if any changes arise during production. Andrew “bunnie” Huang (see the sidebar “Boutique Manufacturing Projects in China: A Case Study”) reminds startups that “you are not going to get your design right the first time. You are going to do some modification, so this is where the ECO, the engineering change order, comes in.” bunnie once issued an ECO to a factory right before Chinese New Year, a time of year when factories traditionally shut down for a few weeks and experience much worker turnover. The factory ended up not implementing the change order. But according to bunnie, “They went ahead and actually reworked all 200 boards by hand, free of charge, because I had this fully documented, and they readily acknowledged it was their fault for not implementing the ECO.”

Another part of manufacturing that startups don’t often account for is the testing of jigs and processes for their products. bunnie recommends startups “account for as much time, or more, for the design of the test jig as for the design of the core product itself.” He describes test jigs as “another product,” in the sense that the complexity will rival that of your original product.

When designing a test plan, it’s often too easy to miss the forest for the trees. To prevent that, bunnie suggests that, in addition to working from the engineering specs, you “also visit the marketing bullet points and ask, ‘What are you promising your customers?’ Probably everything that you promised the customer you should explicitly test.” It’s also important to consider this early in the process, so that you include test points in your design and don’t market claims you might not be able to guarantee through a production test.

For complex products, it’s important to break down your test plan into several stages. Test the riskiest parts of your assembly before putting them into higher-value subassemblies. It’s much cheaper to throw away a small part earlier in the process than wait until the full end-of-line test on the product and throw away or rework a whole assembly. Finally, tests take time, which costs money, so it’s important to test just enough to be confident in your product without wasting money on redundant tests. This testing of individual units on the production line is separate from design validation and the external laboratory certifications discussed in Certification.

Once you start your tooling and pilot production, you will start to hit many terms and milestones. The stages of EVT (engineering verification [sometimes seen as validation] test), DVT (design verification test), and PVT (production verification test) differ slightly depending on your factory or CM, but generally, they represent increasing progress through pilot production into full-scale mass production. EVT and DVT are when you will cut tooling and set up the initial pilot production lines. By the time most projects get to the PVT phase, products are good enough to be shipped to your first customers, especially for early startups eager to get something out to crowdfunding backers.

It’s important to protect your intellectual property throughout the manufacturing process. If you are concerned about sensitive IP, you should get nondisclosure agreements (NDAs) signed during the RFQ process. One strategy for this problem, especially overseas, is to purposefully split up parts of your supply chain among more factories than necessary, making sure that no one vendor has access to your entire design intent. Unfortunately, this strategy is difficult for a startup, because you will likely need to trust one or two main factories for your most important components, and especially for final assembly.

When you have ramped up the line, the next step is generally to maintain production levels and concentrate on increasing yield rates and quality concerns. This phase is typically referred to as sustaining engineering. It’s important for startups to understand when they’ve hit this step, so that they can dedicate resources to the important task of maintaining production while also starting to concentrate R&D efforts toward the next product or brand line. Large companies have entire separate groups for this type of engineering, but startups often need to split the same founder between these tasks initially.

Where to Manufacture?

The decision of where in the world to set up manufacturing is important to establish early in your development process. While scale is an important indicator here, there is no “magic number” to decide when to manufacture locally, domestically, or internationally. It’s also important to consider the complexity of your product. If your first product involves some sort of new technology, process, sensor, etc., you will likely need a specialized factory or vendor.

Three major manufacturing locations for US-based companies are China, Mexico, and locally in the US. This section explores the trade-offs among all of these locations. Mexico can be a popular manufacturing choice for US-based companies, because it is located closer than China, has an easier language barrier to overcome, and offers cheaper labor than the US. This can come with some drawbacks as well, as Dan Goldwater explores in “Moving Supply Chains: A Case Study”.

Moving Supply Chains: A Case Study

MonkeyLectric produces a bike light system (called the Monkey Light) that makes riders safer during nighttime rides, while simultaneously turning their wheels into digital art. The company launched with a bootstrapped first-generation kit that it made in a garage and sold on Amazon.com. Once it became clear that there was quite a bit of demand, Dan began the process of finding a factory.

MonkeyLectric considered the US, Mexico, and China. As new hardware entrepreneurs, they had a bit of trouble finding a factory in the beginning, relying primarily on Google for leads. They chose Mexico because costs were lower than in the US and it was closer than China.

“When you’re small and you’re making something and there’s any complexity at all, you’ll need to have someone there at the factory,” Dan says. “If you’re a small startup, that’s a pretty big cost. Depending on your staff situation, it may be too difficult to send someone to China for six months. It was for us.”

A typical run size was 2,000 units. The factory in Mexico would bulk-ship 500 units at a time, typically fulfilling the order within weeks. Although the quality was good and the workers were skilled and efficient, Dan felt increasingly uncomfortable with certain aspects of the factory environment and the management in Mexico. He remembers, “There seemed to be a lot of corruption and bad business practices happening in the town, and we didn’t want to be putting our manufacturing dollars toward supporting that.”

With three years of experience under their belt, the MonkeyLectric team decided to rethink their manufacturing process as they launched a second-generation Monkey Light. They held a Kickstarter raise to gauge interest and validate the market prior to committing the funds to the new design. When it was successful, they moved their manufacturing to Berkeley, California, and Shenzhen, China.

The current Monkey Light kit consists of two primary parts: a rubber-encased LED board, which generates the light show, and a battery holder that mounts to the hub. Each piece had its own challenges. The LED board was going to be exposed to the wear and tear of biking and to the elements, so protecting it while preserving the lightweight design was paramount. The battery holder was inspired by a camping headlight and involved a lot of small parts. Producing it turned out to be a logistical challenge. Dan notes:

There are several pieces that go into that component—small metal springs, connectors, a rubber piece—and five or six factories make the various parts. The ecosystem for making something like it in the US is breaking down, both in terms of tracking down factories making the various parts, and finding people with the skills to put it together. It was easier to get it done in China; we only had to deal with the factory that made the final product, since their people knew where to get the subcomponents. It was also less expensive.

MonkeyLectric decided to produce the battery holder in China. Alibaba proved remarkably helpful for finding a factory capable of producing the specialized design. Dan searched the site for camping headlights, reached out to the factories that listed them, and discussed his desired modifications. He found that most producers had a much larger set of products than the ones they’d listed on Alibaba and were willing to work with him on his idea. (Dan wrote a helpful and detailed account of that process in a post on Instructables.

Although they looked into manufacturing the coated circuit board in China as well, the team ultimately couldn’t find a factory producing anything quite like it. So they set about creating their own production line in Berkeley, California. The boards themselves are made by a contract manufacturer in Fremont, CA—no need to reinvent the wheel. Once they arrive, MonkeyLectric’s team applies the rubber coating and handles quality control, testing, and packaging. The small team is capable of producing a daily batch of 400–700 units, which enables them to maintain enough stock on hand to be able to ship same-day:

Manufacturing locally lets us rapidly iterate. We can test things out—adding glitter to the rubberized coating, trying new designs—in a way that isn’t possible when you’re dealing with an overseas team. It isn’t always true that manufacturing overseas is necessary to be able to price a product competitively.

While overseas manufacturing can reduce labor costs, it’s important for hardware entrepreneurs to consider all aspects of production—feasibility, ease of iteration, quality—when choosing the best location for their startup to set up shop.

Tesla Motors had a uniquely challenging supply chain, since it chose to take on the automotive industry, which meant that it likely needed to be near the auto industry’s existing supply chain in the Midwest. Dave Lyons (see not available) was at Tesla when it was approaching the challenge of where to set up its supply chain. Dave says Tesla was originally looking at China because of the cost, as well as the timing of many consumer-electronics brands moving manufacturing to China. What it found was that partners in Asia “hadn’t had the same shared experience with respect to how to get things done that weren’t completely understood yet.” In other words, they were good at scale, but not necessarily good at implementing new technologies.

Tesla next spent its resources looking into the entrenched automotive supply chains of the Midwest, which have traditionally relied on a multitiered supplier system in whic the OEMs (e.g., GM and Ford) handle all of the branding and marketing for the cars but only the highest-level engineering. The detailed engineering of subsystems and components is the responsibility of what are classified as Tier 1 suppliers (e.g., Bosch, Denso, Behr) that supply products directly to either the OEMs or Tier 2 or 3 suppliers more levels down. This system of responsibility has worked for Detroit for decades. According to Dave Lyons, the problem with innovating new technologies in that system was that Tesla “had trouble figuring out in a regular 2008 car who actually did any of the actual engineering.” This made innovation hard when so many hands touched a design and none had full responsibility for it.

After learning lessons from working with China and Detroit, Tesla ended up with a rather complicated supply chain for its first cars, sourcing the bodies from Lotus in the UK and parts from all over, so it was important to it to do the final assembly locally in Menlo Park, California, in order to keep the manufacturing close to the original design engineers and collapse any potential debug cycles.

While there are many tried-and-true rules for mass production in terms of processes and supply chain, there are also many new trade-offs for entrepreneurs to consider, as bunnie Huang addresses in “Boutique Manufacturing Projects in China: A Case Study”.

Boutique Manufacturing Projects in China: A Case Study

Andrew “bunnie” Huang has been hacking on hardware since his family first got an Apple II. He went on to study electrical engineering at MIT through to his PhD. He gained notoriety on the Internet for being the first person to hack the original Xbox. He published his findings in the book Hacking the Xbox, a great introduction to reverse engineering consumer electronics. His first exposure to mass manufacturing was as the hardware lead on the Chumby, an embedded computer that was one of the first connected, open source devices.

bunnie now lives in Singapore and continues to work on projects related to manufacturing, doing most of his manufacturing out of China. He says that he consciously “moved from the United States to Singapore because I wanted to be close to China. I came to the ecosystem because it matters to me that much. I want to access this particular ecosystem. If you were to pick a locale, this is the locale you should pick.”

That doesn’t mean that he thinks all hardware entrepreneurs should immediately move to Asia, but rather:

If you’re located in Boston, find a manufacturer in Boston. If you’re in Texas, find someone in Texas. And not just in Texas, in your city in Texas. There’s a lot of friction in going far away, right? Loss of communication, inefficiency of process implementation, cost of logistics, soft cost in terms of a plane ticket and your time.

bunnie encourages entrepreneurs to work out new processes and product lines locally before scaling remotely.

bunnie says that there are great benefits to approaching the manufacturing landscape from a different perspective, in the sense that most entrepreneurs have “a temptation to say ‘Oh I’ll just compete on price’ since you’re in China, but then your margins are really thin and you’re constantly cash-starved. It’s really easy to fall upside down in this situation, at which point you have to raise venture capital or you end up not meeting promises.”

Instead of competing in a race to the bottom solely based on cost, bunnie likes to approach the Chinese manufacturing ecosystem differently by seeing what value he can add to products through unique approaches to processes and supply chains. Two great examples of this are two of his recent projects: the Novena laptop and Chibitronics circuit stickers.

This approach of embracing cutting-edge processes is exemplified by the Chibitronics project, in which bunnie, along with Jie Qi from the MIT Media Lab, developed flexible electronics stickers that can be adhered anywhere, just like conventional paper stickers. bunnie remembers:

I didn’t want to get in a situation where I’m essentially trying to be the master of sourcing cheap LEDs, so we have to add some value beyond low cost, which is the peel-and-stick nature of the circuits. So we focused a lot on that process. I wanted to make sure we were doing something that actually didn’t exist in a factory anywhere else. We built, from the bottom up, a line to manufacture that particular process. That’s a unique value in itself.

By examining the process at every stage, even going so far as measuring raw material stock in factories to calculate the most efficient 2D tessellation strategy for wastage, bunnie was able to add value to the product that others would not be able quickly replicate, in China or elsewhere. This is important, because it also shows how having a close relationship with factories can facilitate a depth of information that can help you build value around your partner’s processes.

bunnie says that this is all part of the changing landscape of manufacturing processes, which is indeed shifting, but not as rapidly as public perception might indicate. He posits:

The most important thing to be aware of is that the equation is still basically correct, but the constants are shifting. The upfront cost of injection molding is going down; the per-unit cost of CNC is going down. 3D printing’s quality is getting there eventually, but just one big problem of many that 3D printing has as a production process is the part quality doesn’t have the finish or the reliability that you expect out of a comparable injection-molded part.

That said, bunnie says there are other reasons to choose processes during this convergence, even if they’re not necessarily at the same level as traditional production processes:

One of the ways you look at this is: prototype is production and production is prototype. This whole thing about injection molding versus CNC and 3D printing—they’re guidelines. They’re just top-level, high-level guidelines at the end of the day. Anything’s possible. There are no rules. Just figure it out and figure out what process makes the most sense for your product. Sometimes, you can go with total 3D printing as part of the story, and that can sell more units for you because there’s press-worthiness in 3D printing these days. That’s a completely valid reason to go with a production process.

bunnie has become a keen observer of China’s unique approach to collaboration, something he discusses in “From Gongkai to Open Source” on his blog. bunnie coined the term “Gongkai” to refer to a distinctly Asian approach to sharing that involves information flow among many parties in order to get things made, rather than the Western approach of ideas flowing in one direction and compensation in the other. He contrasts Gongkai to the Western approach to intellectual property, where you are worried that you’ll “end up spending more money on lawyers than tooling.” The open approach allows smaller teams of engineers to develop products more rapidly. bunnie uses cell phones as an example, because China’s open framework is able to produce phones cheaper, faster, and with more variety than seen in Western markets. Overall, bunnie sees the Gongkai approach to IP “as evidence that a permissive IP environment spurs innovation, especially at a grass-roots level.”

Another recent project bunnie has taken on with collaborator Sean “xobs” Cross, Novena is an open hardware computing platform that can be used as a desktop, laptop, or standalone board. As bunnie progressed through his Novena project, it was important to limit manufacturing processes to the project scope. Since he knew from the start it was a project aimed at hobbyists, he avoided processes that are geared toward high volumes. For example, while most laptops have four major injection-molded plastic shell parts, the Novena gets away with only one.

Other unconventional aspects of the project include “the bezel that covers the connectors, we call it the Port Farm. We wanted that to be replaceable, so that as you upgrade the motherboard you can reuse the case.” Beyond traditional future-proofing, bunnie says,

The design itself is driven also largely by cost constraints. Injection molding is really expensive, and you want to minimize the number of big tools that you have. If you look at our design, there’s one major tub that holds all the components; that’s a single injection-mold piece. If you look at a typical laptop, there’s effectively four tubs, so they have four times the core tooling costs that we have.

Making sure that your manufacturing process tracks to your scale is crucial, especially when starting out with smaller production runs for fulfilling crowdfunding campaigns.

bunnie builds on that conclusion by saying that the design language of the product should match the technological philosophy:

I chose these socket cap screws that stick out. They’re really very distinctive and out there. They’re very aggressive in terms of the techy look. That’s very intentional. I want you to see the screws, because you’re going to deal with those screws. That is going to be your experience in this laptop. That’s part of the language of the design.

It’s important to be honest to the aesthetic of your user group, whether that design language ends up being techy, rugged, elegant, or something else that connects with your primary users.

bunnie also stresses the importance of choosing your factory partners and maintaining clear communication and a good relationship with them:

I consider them almost at the same level as VCs in the hierarchy of partnerships. If you can negotiate with the factory and they can treat you like a real partner, they will effectively use their balance book and their buying power to go ahead and reduce the amount of capital you will have to raise to go ahead and get your product out there.

These kinds of benefits are often overlooked by people coming from software or other nonproduct backgrounds. The working capital required to get parts into a factory and products out can be huge enough to sink a startup if it doesn’t have a fast channel to customers and is on bad financial terms with factories, distributors, or retailers.

This is often best established through a good relationship with the factory owner. bunnie says, “If you go to a factory and the boss does not come out and at least poke his head in or say hi to you, it’s a bad sign, because you are dealing with a mid-tier sales rep.” Once a factory boss is convinced your project is a winner, you will find it gets priority handling and you’re working with an A-team staff, something no amount of money can buy.

Overall, bunnie stresses the importance of the relationship you build with your partner factories. This can be facilitated through clear communication and respect, and it will benefit your startup in the long term through access to new capabilities, clearer communication from their side, and even potentially better payment terms and other more direct benefits.

Supply Chain Management

You will need to set up a supply chain of multiple factories. For example, the plastics for an enclosure are likely injection-molded at a factory dedicated to plastics. The electronics PCBAs would be produced at a completely different factory. Assembly usually happens at an electronics factory, since there are cleaner environments and more testing equipment available. Packaging, batteries, specific sensors, and specialized mechanical components would come from even more factories. The product is finally put into the packaging, along with any paperwork and accessories such as charging cords, at a packout or fulfillment center. From here, it is either shipped directly to the consumer, or put in larger master cartons with other boxes if it is being shipped to distributors for retail. Even this is a simplified version of typically multitiered and complex supply chains. A supply chain can go as far back as the materials, or even raw ores, that are mined to start producing parts.

Another important topic to consider when examining your supply chain is the entire life cycle of your product. How will consumers recycle, repair, or dispose of your product when they’re done using it? This is especially important to consider for objects like lithium ion batteries, which can’t be thrown away in typical garbage streams. You can help this issue by providing return and disposal instructions in your manual and online. You will definitely need to include provisions for this if your product involves disposables in your business model that you need to recapture.

You will also want to think through how a user will repair your product, especially for simple operations such as battery replacement and replacing parts that might be prone to breaking. This can allow you to design these repair and replacement experiences to be less painful for users, and also get ahead of anything the public is bound to figure out about your product when iFixit does a teardown (see not available).

Complicated products such as cell phones can have dozens of factories involved in production through many tiers. These kinds of supply chains can get so complicated that buyers in a large company will typically source components from multiple vendors or factories just to ensure that if a vendor can’t meet commitments, it doesn’t impact the supply chain and production can continue. This approach is a common way to mitigate supply chain risk and give buyers bargaining power on price, since both sides know they could get the same part elsewhere.

You can also rely on a supply chain management (SCM) company to set up and/or run this process of finding factories for you, but you will have to pay it a percentage of your BOM and/or an upfront fee. Alibaba has become a popular website for finding factories and suppliers in China, but it can be a mix of interactions; some listings are actual factories, but many are distributors, holding companies, or other middlemen with little knowledge of the actual product.

Importing from Foreign Manufacturers

If you manufacture in a foreign country and then bring the goods into your home country, you are an importer and subject to a series of reporting requirements and tariffs. For the purposes of this section, we’re assuming that you’re importing to the US.

Importation can be a challenging process. The process of bringing goods from an overseas factory into the US generally happens in one of two ways: air freight or ocean freight. (If you’re in the US and manufacturing in Mexico, trucking is an option.) The means by which you import will dramatically affect your costs. In general, air freight from Asia costs approximately four to five times the cost of ocean freight. However, ocean freight shipments can take a full month to arrive, and your shipment might be bumped (shifted to another ship) and delayed several times. The weight and bulk (volume) of your item is an important consideration in deciding which method to use. So is the urgency of customer demand; sophisticated companies often determine a percentage of goods that they’ll import by air to meet immediate demand, and then ship the rest by ocean.

Once your container has arrived at the port, it’s loaded onto a truck or rail to reach the distribution center or warehouse. This movement of the container from one type of transportation to another is called intermodal freight transport. Pallets arriving by air are also loaded onto trucks. To ensure a smooth transition from one mode to the next, you’ll likely be employing the services of a freight forwarder. This is particularly true while you’re a small company; only large shippers contract directly with major ocean carriers.

Freight forwarders will book passage for your goods. An ocean container may be tough to fill up when you’re first shipping; the smallest option is a 20′ box. Consolidators let you ship LCL (“less than container load”). They aggregate your shipment with the shipments of others to secure a more favorable rate. Freight forwarders can also help navigate intermodal transitions (e.g., ship-to-truck or ship-to-rail), including getting the goods from the factory to the port of departure, or from the port of arrival to the distribution center. Their fees may include insurance for your shipment while it moves through transit (if not, purchase cargo insurance), warehousing costs for temporary storage between stops, arrival agent fees, and more. They are middlemen, and there is little in the way of price transparency in the industry; you’ll want to call several companies to identify the best partner. You can also ask your factory if it will help you book transport and compare the rates you receive with its. Some factories will incorporate shipping fees as part of a per-unit cost.

One of the challenges of working with a forwarder is that it’s often difficult to understand exactly what you’re being charged. What part of the rate is the truck or the ocean passage, and what part is going to the freight broker? The process of comparing quotes can be difficult and time-consuming. This is further compounded by the fact that many forwarders communicate only by phone and fax. A few startups are attempting to bring price transparency in the freight space. One is Haven, which enables customers to use its web platform to obtain quotes directly from carriers and bid for capacity based on trending prices.

One common term you will come upon when looking at importing and shipping is free on board (FOB), which designates which port your product will actually be shipped to. There is a big difference in price if your logistics partner is quoting FOB Hong Kong versus FOB Long Beach. This is defined even further by the latest standard Incoterms established in 2010.

What to Look for During Manufacturing

This section covers the most important areas for each technical team to address during manufacturing. The roles on your team will be shifting as you move beyond creatively making one prototype into a more rigorous mindset of reproducing the final design over and over again.

By this point in your development, the project should have been handed off clearly from your design team or partner to the engineering group. If you have industrial designers still involved this late in the process, they will likely be technically unequipped to converse with manufacturing engineers, and might even be suggesting design-level changes that would have an expensive ripple effect to implement. It’s better to let the engineering or operations team handle things from here on out.

Mechanical engineers are usually most concerned with the tooling that is developed for your custom mechanical parts. Tooling is a broad term that refers to any of the molds, jigs, and fixtures that will be fabricated purely to produce your product but not make it into your final product.

The most common (and expensive) type of tooling for hardware startups is often steel molds for injection-molding plastic parts. Injection-molding tooling is important to address early in your release cycle, because lead times can be 10+ weeks for traditional steel tooling capable of making hundreds of thousands of parts. This tooling can cost $10,000–$100,000, depending on the size, complexity, and material of your parts.

For a startup making fewer parts in a first run, you might want to consider soft tooling in aluminum to move faster. 3D printing of molds themselves (typically through a direct metal laser sintering, or DMLS, process) is also currently an experimental process that has been evolving rapidly. Jigs and fixtures also refer to dedicated tooling, but a jig or fixture is usually more common for alignment, assembly, or letting parts cure in a stable position than molding an entirely custom part.

Mechanical engineers will commonly encounter iteration of the tooling to get closer and closer to the design intent. Commonly, the first parts off an injection-molded tool are referred to as “T0” parts. They will not have the texture of the final part and are expected to not be fully correct, but they provide a good gut check for how much iteration will be necessary.

T1 parts follow, after an iteration has been made on the tool, and this is often the stage in which you will add texture to the tool. More iterations would be subsequently named T2, T3, and so on, but if you go through too many iterations here, you might end up being better off starting the tooling process all over again with a fresh piece of steel. This can be costly, which is why it’s important not to wait too long to engage manufacturing experts to make sure you are designing parts that can be tooled.

At some point, when you are satisfied with the parts coming off the tool or any process, you will want to go through a formal first article inspection (FAI), during which you measure all of the critical-to-function (CTF) dimensions called out by your design engineer and produce a first article inspection report (FAIR) documenting which dimensions are in spec and which are not. This is important for your quality team to look back at later and understand how the first parts came off so that they can assess issues later to see if the tool has “drifted” out of spec through repeated use.

This can also be a good time to pull out one golden sample for later reference. A golden sample is a part or assembly that is completely within spec and deemed essentially “perfect.” Golden samples are especially helpful for areas such as color matching and other qualitative properties that need to be evaluated.

During your development process, it is helpful to consider features of your design that you can make tool neutral. Because steel tooling is a negative form of the part you are trying to create, tool neutral refers to the fact that it is much simpler to make a feature larger by machining the tool. Making a feature smaller requires welding additional material back into the tool. Buttons and other areas requiring a good-feeling mechanical fit are prime areas you will want to purposefully design too small at first, knowing you can subtract tool steel later to add plastic to each part. This process of “dialing in” the feel is common and is part of why no specification can replace the value of having one of your engineers on the line when parts are first coming off tooling.

Electrical engineers typically manage the build of the PCBAs during manufacturing, to make sure everything is functioning as desired before putting it into an enclosure. This involves overseeing fabrication of the bare PCB boards to specification; correctly assembling components onto that PCB through a pick-and-place machine or manual insertion; soldering these together through a reflow oven and/or a wave solder process; and finally, inspecting the board before further assembly.

Firmware engineers will work closely with the electrical engineers to ensure that the line is being designed to test the limits of the board. Often, they will need to create custom test firmware for the device as it goes down the line to push the limits of a wireless antenna or run self-tests of specific components, such as accelerometers—all without burdening the final firmware with the overhead of this code. The device will then be flashed with its final firmware before going through the final, integrated end-of-line test.

Something that will become increasingly important when you enter manufacturing is quality, which refers to the fitness of a part or product for its purpose. This is usually broken into two main categories: quality assurance (QA) and quality control (QC).

QA refers to the overall technique of improving your processes and production line. QC is the process of actually measuring parameters to see if parts are in spec. This often happens for parts coming into a factory, to ensure that you are starting with known good parts and not wasting time assemblying broken chips into an otherwise functioning product. This is referred to as incoming quality control, and you will have other QC touchpoints along the production line as well.

Quality processes will be crucial for making sure you have a high yield rate, which refers to the number of parts or products you manufacture that are within spec and can be used. Yield rates vary, depending on the process and new complexities you are adding. You hope to eventually get them into the desired 95–99 percent range, but you should realize that it will be much lower than this during your first run. 70–90 percent yield can be common for first runs, and some more complicated processes can create yield rates under 50 percent.

It’s important to consider that this is also a function of your spec. That is, if you loosen your standards for what is in spec, a 90 percent yield rate could become 100 percent. This is tempting to do in the late stages of manufacturing, when you feel pressure to get the product out to customers, but you should always look back to your original spec and hold your product to a standard you are proud of.

It can also be interesting to consider alternative manufacturing processes and materials during manufacturing, because they can potentially save you time or money, and might also provide a marketing angle if you want to brand yourself as a sustainable or environmentally friendly company. It’s important for new processes to always be innovating, but to avoid risk, you should also look toward processes that have been shipped in a product by someone else. You also want to be careful not to include sustainable practices purely for bragging, because this can backfire if your company is suspected of as greenwashing.

Realistically, resource-strapped startups might have only one or two people wearing all of these different hats during manufacturing, especially if manufacturing is happening overseas or far away from your office. It can be helpful when manufacturing abroad to have a constant rotation of employees going over to check on production, especially at key times such as initial line bring-up and process changeovers.

Certification

Before you begin mass production, you will need to go through the process of certifying your product with different regulatory agencies, depending on the geographical areas you are launching in and your product type.

Two very common certification groups you will need to deal with if launching in the United States are the Federal Communications Commission (FCC) and Underwriters Laboratories (UL). The FCC regulates anything that is transmitting information wirelessly, which many hardware products will do through Bluetooth, WiFi, or GSM. UL certification covers many different product types and industries, including information technology, medical, power and control, appliances, life safety, and security, to name a few. They also do FCC testing and CE for European countries, among many other certifications.

UL has a few pointers for hardware entrepreneurs going through the certification process. Although some products might not be mandatory to certify, startups will also voluntarily certify other products in order to protect their own liability from safety issues. UL recommends that, when working with a startup on a new or existing product, knowing their marketing plan is important, because it goes hand in hand with your compliance plan.

This is important, because your compliance is based both on what claims you are marketing for your product, as well as what geographical areas you are launching in and shipping your product to. Knowing all of this up front helps ensure startups are taking the correct approach to regulatory compliance, as it can balloon into an expensive process. It could go anywhere from $1,000 to $100,000 or even more, depending on the product, types of certifications, and countries in which you intend to sell. That’s why it’s important for startups to consider certifications during their design and prototyping phase, in order to reduce cost and rework, while increasing speed to market by avoiding surprises through the product life cycle.

It can be helpful to work with a local test lab early in your process. They can help you identify which tests you will need to pass by walking through your intended use case with you. This can affect your design, so it can be helpful to get feedback from a test lab earlier rather than later. They can help you as well by doing prescans on final prototypes that can help identify noncompliant areas or anything you can tweak before the final certification test. A friendly test lab can save you lots of time in researching specs, as well as proper design guidelines, and make your certification go smoother by helping with preliminary investigations and prescans.

Ben Corrado, cofounder of engineering consultancy Rigado, has some advice for startups going through regulatory certification. If you are experiencing electromagnetic interference (EMI) issues, Ben says that some products will end up “spraying the inside of your whole plastic enclosure so that it’s metalized,” but, he says, you can often blunt this by “adding additional layers to the board and sandwiching it all to copper.”

Ben adds that “it is a very iterative process. We 3D printed a device and we wrapped the whole thing in foil tape and we slowly peeled back the foil tape.” Iterating different enclosures like this can often be a much more time- and cost-effective approach to EMI shielding than long cycles with expensive analysis software. Ben also points out an important difference when going for CE marks necessary in Europe. During the FCC’s certification process, “all you’re doing is radiated and conducted emissions. When you go to CE, you’re doing immunity as well, so you’re going to have things blasted at you.”

There are also many industry-specific agencies and trade groups with standards that you should consider for your product; examples include the Society for Automotive Engineers (SAE) for automotive products, National Institute of Standards and Technology (NIST) for measurement products, or the National Sanitation Foundation (NSF) for food-related products. Apple’s Made for iPhone/iPod/iPad program (MFI) is an example of a certification of compatibility with a specific company’s equipment. These kinds of trade groups are more industry-specific than the previous general consumer-product certifications, but they will often have standards you will need to meet before anyone in your industry will stock your product.

Another category of certification so complicated that it could be the subject of its own book (and there are plenty of books dedicated to it) includes those provided by the Food and Drug Administration (FDA). This certification is required for any medical products and some food products, and includes specific categories and regulations, depending on how invasive a product is and what level of recommendations is being made. Most consumer products avoid many of these regulations in their first product by branding and marketing as a measurement device, not a diagnostic or medical device.

The most important part of certification is knowing early in your process what certifications you need so that you can make the correct design considerations during prototyping and work with an engaged test house during manufacturing to make sure this doesn’t become a time setback in the ramp-up to production.

Packaging

Another important consideration in manufacturing is your packaging. This is important to think about before you enter production, but it shouldn’t really be on your mind until after you figure out what product you’re making. A package’s primary function is to protect the product during shipment, distribution, and retail, but it’s also important to consider the branding potential available with your packaging.

Once your product is successful enough to sell at retail outlets, your package itself becomes a great way to sell your product. Obviously, there is lots of marketing and branding to make sure people are aware of your product before they go to a store, but once they are on the aisle comparing your product to all of your competition, your package has the responsibility to inform and differentiate your product.

Before designing your retail package, you should visit the retail stores you envision stocking your product and find the areas in which you expect your product to be sold. You can then consider how your product’s packaging might fit in there. For example, is everything sitting on shelves or hanging from tags? If all the boxes in the area are blue, should yours be blue to fit in or red to differentiate? It’s also important to make sure your packaging is large enough to incorporate all of the branding, logos, and messaging you need to differentiate without being unnecessarily large, so you don’t waste material, shipping costs, and retail space on the package.

Beyond retail shelves, it’s also important to consider how your products get packed during shipment. Usually, packages are placed in a larger box called a master carton, and then attached to a pallet. It’s important to consider dimensions, so that you can maximize the number of boxes in a master carton and the number of master cartons per pallet. This can often be the fundamental trade-off to decide while designing your package: do you increase size and maximize branding space on the package, or do you minimize the package and optimize for cost? The right answer for your product will almost always be in the middle of these extremes.

oDuring packaging design, it’s important to note that a packaging designer who deals with the structural form of the box will likely not be the graphic designer who chose the colors and designed the graphic assets of the company (logos, fonts, etc.), which were probably made for your website and launched long before you began considering packaging for the product. Separating these two disciplines can help you not only go faster, but also focus on the importance each discipline brings to the package.

Out-of-box-experience (OOBE) is something that has become more and more important for consumer products. Apple has set a high standard here with the iPhone and iPad; that has driven up customer awareness of the package and increased the cost companies are willing to spend on packaging. Customers should have a pleasant experience opening your box, while also learning more about your product and how to use it while they unpack. There should be a logical flow in how the box is unpacked. You almost always want to start with the featured product for emphasis and bury cords, manuals, and accessories another layer down in the packaging. The package can also sometimes be repurposed for the product so that it isn’t wasted. For example, it could become a charging station, mounting stand, or other natural accessory for the product.

Before you jump to a full retail package with 10 colors, foldout panels, and a viewing window, you might consider a white box approach for your first product. This means that the box has only the base color (likely white or natural cardboard) and printing in one color (normally black) on one or two sides. This minimizes the cost of your package, because this version will likely never be on a retail shelf where the box might need to drive sales. This is also effective once your product has crossed the threshold to online retail sales at Amazon, as fancy graphics and multiple colors are wasted on a customer who is making a decision based more on the online reviews and other added information, not a package in her hand.

Sustaining Manufacturing

Sustaining manufacturing is also an important subject to consider. While your first challenge is getting a pilot production line up and running, the burden it takes to keep that line running smoothly (especially when you scale up quantities) shouldn’t be underestimated. This is typically the point in production where the ownership transitions fully from the engineering team to an operations team.

While it’s important to have operations staff, either from your manufacturing partner or from your own team, involved in the process well before this, it’s also important to take a leap at some point and fully transition ownership to the operations team. They will be better equipped to deal with the types of process-specific issues that arise, as those issues will likely be caused more by tool wear and process drift than issues with the original design. This transition also allows the design engineers to go back to the drawing board and start working on the next product or new brand line for the company.

Now that you’ve gotten through prototyping, successfully manufactured your product at scale, and have your production line humming, it’s time to look to your future products and start developing your next product.

Post topics: Business