This is an entrepreneur's guide for hardware investors. The goal of this Guide is to show the differences between hardware and software startups and why the standard investment tests you can use for software often don’t work with hardware, or at least they are not enough. I couldn't find any meaningful resources for those who want to invest in hardware or are generally interested in hardware startup mechanics. So I decided to write one.

The Guide doesn’t touch standard themes like product fit, team, lean and other important issues. If you’re interested in those, you can find more information about them elsewhere. They still apply and they need to be considered when analyzing a hardware business, but it’s beyond the scope of this Guide. It covers the following topics:

• The 5 phases of hardware startup's life
• Resources that separate hardware startup from the rest
• Target markets
• Evaluating traction and sales
• Margins and BOMs
• Intellectual property and patents
• Manufacturing and risks
• Capital requirements
• Pricing models 
• Competitive positioning

So if you’re interested in learning more about the differences between an early stage software vs. hardware startup, the hardware specific risks and ways to mitigate them, do read on.

The 5 phases in a hardware startup’s life

In my experience, there are five well-defined phases in the process of making and launching a hardware product. That’s important because it significantly differs from a software startup.

Although you’re able to create a viral game or service in a few weeks’ time, there are natural limitations when comparing a hardware product development cycle with hacking software solutions. The reason why I’m outlining this is because it’s very inviting to compare companies and, for instance, think that a company is not as efficient if it doesn’t ship in three months, or if it delays a launch when producing hardware.

The sketching phase

Sketching an idea is not being a startup yet. The ideas for a product are often driven from two sources. The first is tech driven. Founders have some cool tech that is looking for a problem. The second possibility is that there is a problem and the founders are looking for a possible solution. Both are valid sources for ideas.

At this stage, the company usually has sketches and ideas. That's as early stage as it can get. Interestingly enough, some of the successful crowdsourcing campaigns have been in such a stage, working only with renders on Kickstarter or Indiegogo, but are still used as a reference as being a successful hardware startup. Usually they don't ship at the end, or at least they have really long delays. They face a long list risks and only the really lucky ones will make it to become a reasonable businesses.

The prototyping phase

In the prototyping phase, most hardware startups go public with a PR push, because they start feeling the need for funding or at least exposure. Quite often the prototypes are made with development platforms like Arduinos or Raspberry PIs. Almost all of the crowdfunding companies are in this stage. They have a working prototype, usually 3d printed casings, some proof of concepts. The software side is lacking, but the hardware somehow works. The majority of preorders happen in this phase.

The company is usually 12 months away from shipping the product, if they're lucky.

The production phase

In the production phase, the company is setting up manufacturing processes, designing and tooling casings, packaging, writing manuals, manufacturing, working through certification and so forth. Risk-wise there is not a lot of difference to the second phase, however the company's ability to execute comes to a real test: if they are not really resourceful, they will fail.

If an investor believes in the company and the product, and it’s obvious that the company won’t make it on their own, now is the time to pour the money in and help find the person or providers who’ll carry the company over this phase.

The company is 6 months away from shipping the product.

The manufacturing phase

The manufacturing phase comes with a whole bunch of challenges. Particularly in the first batch there are issues with quality and delays. User support really comes to the test at this stage. A lot of problems are caused by external factors which need to be managed.

Production cycles take time, usually 8-12 weeks. And if a company runs out of stock in between two batches, it can’t ship, so it can’t sell. The only way around it is to increase batch size. So instead of 3,000 units, the company needs to produce 15,000 of them at one time. And that takes working capital, which you, as an investor, need to provide in the initial phases.

There is one additional risk involved. As long as the company produces small batches, it’s somewhere below ground on the totem pole of importance with suppliers. Orders will get cancelled, changed and misplaced. Even worse, I’ve seen poor PCB manufacturing, fake chips on the circuits, forged power supplies, electromagnetic influences on stability of the circuits and a whole bunch of other, unforeseen issues.

And all these will affect the ability of the startup to push forward. The trick is to analyse the situation and correctly identify the issues that would make or break the product. So ask a question: Are the drawbacks that will happen part of the natural process of a hardware startup development or are there real issues with the product, company or something else?

Quite often investors are less forgiving to hardware startups than to their software counterparts. The reason is that with software, you just publish an update; in hardware, you’re often dependant on others, but so are your competitors.

The company is shipping, but often with delays.

The scaling phase

SScaling the company when it’s successful enough is the last phase. When it can start working on larger batches, distribution partnerships, has enough working capital to finance production, it stops being a hardware startup, but rather becomes a proper hardware company. And it should be viewed as such, from a risk assessment perspective to its ability to perform.

Hardware vs. Software Startups

Hardware is hard. For a reason. But that's also an advantage.

In broad strokes, there are a few areas that are important in early stage hardware business and are not part of a software startup resources kit. It is not a complete list, but rather a list of additional resources that a hardware startup needs, on top of everything else, to be successful. All other resources like product management, software engineering, usability design, user support, sales, marketing, are still required as they are in any other business.

All this means that hardware takes longer, is more expensive, but also somehow more defendable. Anyone who wants to compete on hardware needs to go through the same process. The point I’m trying to make is that it cannot be modeled by software startup models, at least not in the early stages.

Hardware engineering

For a hardware product you obviously need to design hardware. This means circuits, electronics and everything else that comes with it. Designing hardware is a long process and has one significant disadvantage: limited iteration capabilities. If a software startup can write and deploy code, Facebook style, on a daily basis, that’s not the case in hardware. Sure, there are prototyping and simulation tools for hardware, but nothing compares to a real working circuit. And one needs several versions of those to learn about the product and iterate. That takes time.

In software, the startup can iterate at least once a day. In hardware, at best, you can do one iteration a week, often not even that, more likely once every two weeks. That reinforces a need for mature engineers who are diligent in what they deploy and how often. Some resent me for saying this, but I do believe that old school engineering practices apply here. A bit more planning and a little less improvisation is an asset here.

Design engineering

Depending on the target market, industrial design is important. For consumer products, it is one of the key areas. And industrial design is demanding. Some startups opt for external designers, some for in-house ones. Whichever model a startup chooses, iteration is key. Finding a way to integrate industrial design with user experience, UI and everything else demands a lot of iteration. Simply ordering a design from an external firm without an option to iterate and develop further is bad practice.

If a software firm needs to figure out UI and user experience, a hardware startup needs to add the actual product into the mix. And it’s not just the looks, it’s also the feel, that's important, so selection of the materials, their characteristics and manufacturing processes.

And since the company is designing something new on limited resources, it’s likely they have to be very resourceful. Often suppliers are far less innovative as one would hope. So all responsibility for dealing with materials actually falls on the startup, which can be a job on its own.

I call this design engineering, and it’s often neglected. It needs to be synced with industrial design. One has to think about usability, feel, durability and of course safety. Choosing a plastic that melts at lower temperatures than overheating safety limits is not only bad practice, it can burn someone’s house down!

Firmware engineering

Firmware is the hidden virus in any hardware firm that does electronics. Firmware is the layer that makes hardware alive and appealing. A company without firmware resources is dead before is even born. Sure, it's easy to make firmware that makes hardware work, however writing firmware that's efficient, allows for remote update, safe to use… that's hard and the firmware person/team needs to work closely with the hardware and software engineering teams.

Do not underestimate this! It requires deep engineering with C, sometimes assembler, understanding processors, interrupts, memory allocation and so forth. On the firmware level, an engineer does not have the luxury to initiate another AWS instance in the cloud, but instead they work with 64 or 128KB of memory! If done properly, the firmware rarely uses high level libraries, or standard open source resources.

The reason is very simple. A hardware startup usually builds its own computer with non-standard components; in software terms, they need to build their own operating system.  There are a bunch of drivers, real-time operating systems (RTOS) and components that can be reused, but it's still fairly low level, at least compared to web software development. Imagine that every software startup would need to build its own infrastructure and an operating system (or at least parts of it) first, and you’ll get a sense of the effort a hardware startup undertakes.

Logistics

Hardware products cannot be wired overseas like software. It takes resources to take care of logistics. Packaging, shipping, customs, declarations, warranties... That's actually one of the few things that you can safely outsource, but at a price. However, it’s still a good practice to do the initial steps in-house to get the feel of the process. It’s much easier to outsource when you know what you’re outsourcing.

Logistics is important because not only it deals with getting the product to the customer, but also with getting the product from the customers. With software, in case of an unresolved issue or a warranty case, you only need to return the money to the customer - in the worst case. In hardware, all this applies, and you have to deal with the logistics of getting the faulty product back from the customer. And it costs extra money on top of the refund.

Certifications

When putting hardware on the market, certification for different markets is necessary. A product for consumers cannot be placed on the market without appropriate certification. To make things worse, there are different certifications for different markets like EU, US, Australia... If the company is active in a specialized industry with additional requirements, multiply the effort by 10!

The company needs to go through it, there is no way around it. It will take up to 2 months after the initial product is ready and will cost time. And you can’t do it with a prototype, you need to have the end product machined and produced. There are some risks involved. For example, in my startup, the certification process exposed that a standard charger that was sourced in EU (not China) and supposedly had all necessary certificates, was actually fake. The cost: manically finding an alternative and an additional month of the certification process.

Most importantly, it will significantly limit update possibilities that are possible on the product. If the product is changed, even if you just switch an equivalent component or part of the device, it needs to go through the whole new process of recertification, which translates to another 2 months and more costs. You never have this kind of upgrade cost with software unless you work with defense, aviation or medical systems that need thorough verification.

Production

Creating a new product requires organizing manufacturing as well. It takes time and experience. When a company is scaling, outsourcing manufacturing to an overseas manufacturer makes a lot of sense. However, outsourcing overseas is not optimal when the company is still designing a new product and has to figure out how they are going to manufacture it. If the company is well funded, it can afford to hire experts to do it for them. But if it's well funded, it's out of scope of this guide. Then the company doesn't need your help as an early stage investor anymore.

So, the company needs a person who sources the components, organizes the tooling, negotiates prices, lead times, payment terms, and coordinates manufacturers so that everything works. Sourcing is incredibly hard when you deal with smaller batches, which happens in the first few production runs. With building up a single product, I've seen synchronization issues of sourcing from around 25 different suppliers, delivery delays, lead times of up to 19 weeks, cancellation of products, falsified electronic chips, fake certification documentation, mismatched deliveries, and quality issues. Since there are a lot of things that will go wrong,  production responsibly is likely a full person job! So that’s another dimension to the hardware startup framework.

Target markets

The key to product positioning is to understand what startup's target market is. Talking broadly, there are three target markets and designing for them greatly influences recruiting, design and production processes.

Target markets clearly define what should be the core advantages of the startup. In consumer products, high tech comes secondary to usability and design, while in business products, utility and manageability is key. These things define what kind of talent, plans and timeframes the startup needs to put in place, and it’s a great tool to analyse the startup’s strategic ability to plan and sync their resources with their market.

Consumer oriented  products need to be well-designed, have a simplified UI and as seamless a user experience as possible. This also means that the startup has the longest path to production, since there are a ton of details that need to be thought through and defined. The startup has to nail the product, software, packaging, manuals, web site and support. Its defining corner stone is usability. On average, consumers don't have the patience to read through manuals.

When looking at consumer hardware, an investor needs to look at the product advantages in terms of usability and design.

Business oriented products need emphasis on robustness and manageability. Business buyers are often not the users of the product; they are particularly interested in utility, ROI and longevity; coolness and high tech comes second. Therefore, an investor needs to assess the startup's ability to sell to businesses, and emphasize maintaining quality and utility. One particular aspect is mass deployment which is often neglected. Are there processes in place that would allow deployment to business, how long will the sales process be and how is production synced with them?

This all will affect time to validation, traction analysis and will need to be financed.

Tech oriented products target engineers. Their core driver is high tech, not design or handling robustness. They thrive on tech, specs, and close developer relations. They can be an excellent source of partners, however markets are often limited. Tech-oriented products need to demonstrate clear tech leadership and a clear roadmap for future innovations. An investor should look closely at tech talent, among other things. The business case is that they provide a product that their clients will use to access other markets. The Raspberry Pi is a great example of that. An API or a software library is the closest software business analogy.

Traction and sales

Traction is obviously a proof of growth. In the hardware world that translate to sales, measured in number of products sold/shipped. Period. The reason for this is that if you want to compare it to anything, you need to have a comparable unit. Money is not one because you have no clue about the margin, so you’re comparing sales of different products.

Many investors try to compare traction with free software services, or other famous startups. I don’t need to explain how wrong it is to compare free distribution traction to a paid one.

Early adopters

When checking the number of supporters on Kickstarter, you see that the volume of supporters for successful Kickstarter projects is between 2000-8000 in majority of cases, maybe even less. That's it. That number, almost by definition, are early adopters.

When looking at traction (sales in volume, or orders) one has to look at it from a healthy business perspective and try to filter out all the cool/PR/hype noise. So, if a startup has several thousand users/orders without using a crowdsourcing platform, it's as successful as if they were successful on Kickstarter or another crowdfunding platform. Even more, if the numbers are equal, it's likely it's even more appealing since it hasn't benefited from media exposure yet.

I often hear this comment: "We're looking for a meaningful traction!" I haven't heard an explanation yet of what meaningful is, but rather something like: "Well, company XXX raised 1M USD on Kickstarter!" Without knowing how many supporters it has, the number does not help you in comparing it to other companies.

Obviously, if a company does not use PR, distributors or buying ads, and still generates traction, that's way more valuable than if it invests a lot of money just to generate initial traction. A typical trick of Kickstarter campaigns is to use Google remarketing to increase conversion. Which is great. It just needs to be noted when comparing traction and true product appeal.

Evaluating traction

Comparing traction to other hardware startups makes more sense, but there are fairly few precedents available.

What do individual numbers actually mean and how to evaluate them? If we categorize traction into categories, there are signups, reservations, orders or done-sales. Startups usually track one or more of the following categories and it’s important to understand them to evaluate traction:

• Signups: no money involvement by a user, the transaction costs is one email signup.
• Reservations: many startups opt for reservations, particularly when they don't know when they will ship. With reservations, a user pays a certain reservation fee (eg. 10 USD) and pays for the rest of the price when the product ships.
• Preorders: taking orders is valuable, however if it's far from the shipping date (like several weeks), it still allows users to cancel the orders.
• Done-sales: after the product was shipped, the sale has been made. There is a small chance that the product will be returned and a refund requested, however that chance is irrelevant in assessing traction.

As you can see, the traction pipeline could be organized as a classical sales pipeline/funnel, incorporating the chances of dropoff/conversion and thus assessing the health of traction. If you look at numbers of signups, reservations, orders or shipped products per month, you could use a conversion factor like: signups 5%, reservations 65%, pre-orders 90%, shipped 100%. Multiply the factor by the reported number and you’ll get a comparable unit of shipped/sold products, or number of customers. With that you can easily see and compare the traction of the product and compare it to the potential or addressable market.

Hockey sticks

With physical products one must be aware that traction is highly dependant on distribution/marketing. So a few questions need to be asked: What’s been the PR/exposure effort to date and how has the startup approached media and markets? How about distribution, is the company using distributors or partners? One important issue is also marketing: Google Ads and similar might work, however one must know how much advertising is being used to generate traction?

Due to the lack of viral properties, and with only its second-best equivalent - word of mouth - there is no such thing as hockey sticks in hardware. There are simply too many obstacles like ability to ship, stock, marketing, distribution and of course the need to sell and actually persuade people to swipe their card.

If you look at shipping data from the big electronic companies, there is always a wave pattern, never a hockey stick. A peak in the wave is an introduction of  the new generation/product, and a drop is when the product matures and is waiting for the next generation. Unlike with software, one cannot upgrade a physical product with one keystroke to the next generation and build a new audience on top of it.

A hockey stick implies viral/exponential growth where every user generates X new users by using the product. In hardware terms it would mean that every customer generates X new customers. In physical terms there were only a handful pyramid schemes that succeeded in that and almost all required profit sharing. And they never lasted.

I almost always get a question about traction in my investor pitches, and never a question about context. And I'm still puzzled about why not. Just talking about traction without context makes very little sense to me since it makes the number completely incomparable.

Margins and BOMs

Margins and bill-of-material (BOM) are very important concepts in hardware startups, but often ignored. In vast majority of cases, hardware startups make products that need to be sold to their clients. Margin and BOM are inversely correlated. Even though in theory we have different BOM's, in the hardware startup world the term BOM is most often everything that it costs to build a unit without direct costs of the company. Material, assembly, packaging... Subtract that from unit price and you get the margin.

A margin is obviously a core driver of the company's income. Asking for revenue without understanding margin is pretty meaningless. It's a common mistake that software investors make. In today's early stage software world, the majority of cash outflow goes to paying salaries and maybe some server costs. In other words, BOM in software startups is almost 0 and the margin is almost 100%. In hardware, one first has to pay suppliers and what's left is the income for the company. From THAT one needs to pay salaries, servers and profits.

It is quite dangerous to compare the software and hardware business, because one needs to understand the balance sheets of both first, or face some serious risks of uninformed decisions.

It is useful to put the numbers into perspective. For example, it's not uncommon to have an investor talking to a hardware startup that sells 1,000 units per month, and comparing it to traction of a software startup that has a million users on a freemium model.

Let’s look at the economics of two potential companies: a hardware startup HARD ltd. sells 1,000 units of their product a month for $100 at 30% margin. A software company SOFT ltd. runs a subscription service that runs a freeware system with 1 million users, typical 4% conversion rate and $10 yearly subscription costs. If we ignore costs of salaries and servers (assuming they are comparable), we can calculate HARD's revenue of $1.2MM/year with income of $400k. SOFT's income roughly equals revenue and it's $400k. That’s of course based on the assumption that lifecycles of the HARD’s and SOFT’s offerings are comparable. You can easily adjust factors, but the idea is that sometimes incomparable things can be reduced to the common denominator, which in business is usually economics.

Even though it's clear that financially the businesses are fairly comparable, some still maintain they are different because one has 1M users and the other a mere 12k in the first year. Looking at this example, how do you decide?

Economics of distribution

How the product gets to the customers is often the key to success. In the software/services world people like to talk about virality or similar mechanisms for distribution. That's way harder in the hardware world.

All these things cost money and again it depends on what kind of market the company targets. If consumer, sooner or later, the company will need start building distributors and retailers. If corporate, integration partners will likely be the target. Similar goes for PR and marketing.

If you look at it from a business perspective, the rule of thumb is that every degree of separation between the company and an end user costs roughly 25-30% of the margin.

If a company sells directly to consumers, it needs a margin of 25-30%. If it uses distributors/partners that sell directly to clients, the margin required is 50%-60% to finance those channels. If the company targets retailers that work through distributors, that's another degree of separation. So we're talking about 75-90% required margin in order to show sustained business that can finance its distribution channels.

Another issue is returns. With a physical product, the company will likely face returns. If it’s lucky, we’re talking a few percent. So that needs to be accounted for as well. The difference between returning the money to an unsatisfied software customer and to a hardware customer is huge. In hardware, you need to get the product back, send a new one to the customer, and pay for both shipments. That means that one return eats up profits of 3-5 successful sales. It’s not unmanageable, one just needs understand it.

An investor needs to be aware that different hardware businesses require different approaches and have different financing needs. One simple cannot just define one rule and smear it over all.

Intellectual property / patents

What about IP and patent? That's one of my favorite questions that I like to debate with investors. I've seen offers that conditioned investment by a requirement to patent a part of tech. Here are several layers to think through.

First, if the company has been bootstrapping, it's highly likely that they cannot afford a patent process. I've talked with an investor who was trying to convince me that patenting was cheap. Clearly he never set foot out of his glass tower and tried bootstrapping or filing for a patent.

Second, a startup’s resources are extremely stretched, so patentability is something a startup can start thinking about when they have money for. So the right question to a startup is not: "Do you have patents?", but rather: "Is there anything patentable and how much will it cost?" and then add this number to the investment offer.

Third, until a startup is successful and has significant funds/revenues/traction, it is highly sensitive to any kind of litigation. Here there is a big difference between continental Europe and US. In US, basically everybody carries its own costs and there is almost no first-use protection for IP. In continental Europe, there is first-use precedent, lawyer fees are reasonable and often with a ceiling, and whoever loses carries all costs of the winning party. So at least in Europe patent litigation is slightly more just, but that's not the point.

The point is that, on average, starting a patent litigation in US costs at least $1MM. That’s the sort of money a startup does not have. So if an investor is worried about that, they should put up that money on the table in case of the patent lawsuit. At least in the very early stage. Putting conditions on patents during investments, and not providing contingency cash, is counterproductive.

Manufacturing

Quite often we hear that we should manufacture in China. The rationale behind manufacturing overseas is of course the price.

When you invest in a hardware startup, you should understand that optimizing the manufacturing process requires consideration of many variables: the most important ones are volume, quality control, and time. They are all functions of the phase the startup is in.

I do not understand how anyone can justify going overseas for the first 4 phases of the hardware startup process. The volatility of sketching, prototyping, testing and first batch manufacturing phases is so high, that you simply need to be very close to the manufacturing line. Of course, if you plan to be based in Shenzen, that's perfectly fine, but the majority of startups I know are based in Europe or US. They are at least 6 time zones away from China or Taiwan. How anyone considers iterating on a such long distance lean, is beyond me.

Preparing and working on hardware manufacturing requires a lot of iterations, which consume time and that the startup usually doesn't have. A startup usually doesn't have a polished production plan, so there are many details that need to be ironed out in the manufacturing process. So engineers simply need to fly over there several times a month, or live there. And that's a waste of time. You cannot be rapid and lean if your manufacturing is half a world away, in other words, it’s way easier to bootstrap on gas than on airplane tickets.

The second variable is quality control, which anyone, who has ever ordered a part from cheap suppliers (and even some expensive ones), understands. Mistakes will happen, in particular when you work with 30+ suppliers, which is often the case for an average product. The trick is to have control over what's going on. It's much easier if you have your suppliers close by, particularly the ones who are not just reselling stuff, but the ones who are manufacturing things like packaging, plastics, mechanics and so forth. You need to be there.There is no other way than to get a product in your hand. If that happens overseas, it takes several days of sending it back and forth, or a plane ticket and flight overseas. And if you have to do it several times in a month, you might just as well move there.

The third variable in deciding about going overseas with production is volume. A startup working on small production batches is likely to produce their product in thousands. Which in hardware terms is small. Some of the larger suppliers actually have minimal order volumes of 10,000. So you have to rely on smaller ones that cannot utilize economies of scale, the price difference per unit is not geographically conditioned and rather small in small batches. On top of that, there are always set up costs for production runs, initial tooling, production templates and other initial costs, which at small volumes significantly influence cost structure. Variable costs difference per unit gets really small at small volumes. So why would you expose yourself to risks related to time delays and quality problems if there are no financial advantages at small units?

A blunt question: "Why aren't you manufacturing in China?" is meaningless, unless you put it in context of the phase the startup is, time constraints, quality control risks and batch volumes. Having looked at all these parameters, you'll have a much better sense of the startup’s ability to perform and where they need to optimise.

Capital requirements

On top of the usual 12-month-runway-to-get-to-the-next-level, at a hardware startup one must account for initial costs and working capital.

Margin is linked to bill-of-materials and ordered volume. In other words, if you order more units, you get them at a lower price per unit. The amount of money needed to finance that is the working capital that is usually required upfront. There are only a few suppliers that are willing to finance early stage startups, so the goods need to be paid several weeks or even months upfront.

And the financing for it needs to come from somewhere. Initially it's likely that investors will need to step in. A new company will have a hard time finding a bank to finance working capital. Crowdsourcing helps, but it's often not enough, because it’s based on an optimistic view of the economics of early stage hardware, ignores initial costs and neglects to account for all the unforeseen issues in first three phases in a startup’s life.

Initial costs

Another interesting resource that is different from software startups are the initial cost requirements. They are significant in comparison to a software company. There are development kits for processors, which often cost several thousands of dollars, a 3d printer doesn't hurt, but some 3d printing services can be outsourced.

An interesting example is plastic tooling. Unless the company buys off the shelf casing, or it has a really simple casing for their device, this is a whole new can of worms. Tools for plastics need to be machined for every plastic case or part separately and it takes between 4-12 weeks to be machined. The tooling costs range in several ten thousands dollars. A typical plastics tooling manufacturing process requires several test run cycles before the tooling is ready for production. One expert told me that in their company an average number of trial runs of a plastic tool is 4, and I know a company that did 17 trial runs before the tools were accepted. The mould/tool for plastics is heavy, so it can require a forklift to take it from the machining to production engine, do a test run, inspect it, return it back for polishing and so forth. That means that after the tool was delivered, it took 17 trials before all the details were polished. Oh, and every trial run takes at least a week!

Tools like oscilloscopes and other electronic equipment are needed as well. All in all, initial setup and tools can easily run several months and cost 100k-300k. Compare that to a software company, where the initial equipment are actually just computers for developers!

Working capital

This is probably the most underestimated resource for a hardware startup. Working capital is needed to order batches of products that the company sells to customers. The company has to pay for goods, that it can sell. If everything else remains equal, working capital is on top of the capital requirements for a comparable software company. At least for the initial batches. Banks will finance the working capital, but only when you have a running business or funding. Usually not for a few months-old startup without any customers.

To put it into perspective: let’s say that a company does an initial run of 3,000 units for a device that costs $100 and they have a 30% margin. That means they have to come up with $210k upfront just to pay the suppliers. And that does not include tooling and initial production costs of $100k and certifications costs. That is why a convertible note of $25k does not cut it!

The best model to take care of working capital is that investors invest into the first batch and immediately help the company get a working capital credit line from a bank. It makes little sense to finance additional batches with equity. But for the first (sometimes second and third) batch, there is often no other way.

Again it’s an advantage as well, because all of their competitors need to go through the same process.

Production batches

Another important issue is the nature of batches. Unless the startup is extremely lucky to have just-in-time production capabilities, they need to order production in batches. Again, each batch requires working capital, but also means that it carries certain lead times to be manufactured. That's a production cycle. It's a major curse for early stage startups. Usually, production cycles are between 8-12 weeks, so if the startup is successful in selling their product fast, and if they don't have enough capital to finance at least 2 production cycles, they will have a hard time keeping their market presence, because there are not a lot of clients that will put up with several months of backlogs.

So, as an investor, you have to realistically assess what are the startup needs and support them with initial batches. If you're not prepared to do that, don't invest in hardware startups.

Fundraising

Fundraising for hardware is different than fundraising for a software startup. If one looks at three (often misused) stages of startup investing it’s obvious that the differences are huge.

In a seed stage, software takes $25-$250k to fund 3-6 month worth of salaries for engineers, office costs and some servers. The aim is to build a working product that can prove some traction. It has to work, ideally it will gain 10k-100k users and prove to investors that it’s the puck part of the hockey stick!

Put this into hardware startup’s perspective. In order to place a product into the client’s hands the startup needs to design everything, produce the initial batch (which is usually expensive), certify it and sell it. And selling it is very different to persuading people to try an app!

The trick with evaluating an investment is that costs of acquisition need to include BOM of unit + all initial costs which need to be done uprofont. That means that a hardware startup needs 6-9 months to create a workable hardware that can be put in users hands, finance working capital for the initial batch, initial cost and sell the units. It can’t just give them away. That means that a hardware seed round is likely to be in range between $500k and $1MM, which puts a lot of pressure on initial capital. But the key here is to know what you want to validate? Usability of the product, team’s ability to execute, sales ability? If it's traction, do put it into perspective!

Series A is the round where in a normal software business, the company starts scaling. In hardware, the company needs to perfect the product and it creates the second generation. More importantly, it seriously needs to start looking into sales/marketing channels and optimize the value chain. It’s becoming a proper business. If in software $3-5MM can cut it, it’s unlikely that in hardware that will be the case. If a software Series A will take the company through 18-24 months, hardware is at most half that period. In that time one needs to decide if how they are going to support it.

However there is even a more important issue on the table. Since the hardware business is capital intensive in the beginning and it needs a large Series B,C,D when successful, smaller early stage investors get hugely diluted and therefore cannot produce meaningful returns, even if the company is really successful. A natural reaction for smaller investors is to push founders into more lean models and press them with unrealistic traction scenarios, which causes a drop in quality and increased risks.

Pricing

The fun discussion with investors is often about pricing. The elasticity of prices for hardware is something that most people think they understand without any proof. Since hardware startups produce physical objects similar to the ones in stores, we like to compare them to others.

The right price

However, often pricing suggestions by "informed" individuals have little to do with the real world. As long as the startup sells to early adopters, they can charge whatever they want, if they can sell it. There is no right price! When they go mass-market, they'll need to adjust pricing to whatever the retailer thinks it makes sense. But going mass-market will totally change economics, and since new product revisions take a year, there will be plenty of time to figure it out.

Of course it makes perfect sense to model things, however putting price pressure on a startup that sells a product in several thousand units and has a backlog makes very little sense to me.

I've got a seed investment offer by a well known angel/VC that conditioned investment by slashing the prices immediately by 2/3, even though the company was backlogged by orders and sold out the first batch in a few weeks. The investor thought that the product was too expensive, because “he felt that way”. Needless to say, that gentleman hasn't invested in a lot of hardware startups yet, even though he’s a very successful software investor.

Pricing is extremely subjective and if a software product has a market comparison, in hardware that's not the case. This has been proven many times, by smaller startups or big ones (think smart thermostat pricing when Nest came out, or iPhone pricing). Price elasticity is subject to market proof, not feeling or theory. And it should be managed in that way.

It's very simple: hardware startups need to compete on innovation, not price! Whoever pushes a small batch startup to compete on price, has no idea what they are talking about.

Pricing model

Another important consideration is the pricing model. The basic decision is between continuous and one-off charge.  One-off charge is the natural model. You make a device and you charge for it. The challenge happens when the device is connected online and requires some server side storage or processing which causes additional costs.

Often, influenced by software products, people consider subscription models to gain a recurring revenue stream. Since financing hardware and leasing it to customers is not a viable option for a startup, there is an option of charging a subscription for a services linked to the hardware.

I've heard several times that this is the only way to keep sustainability in the model and heard a few investors declaring that the only way hardware is interesting to them is by having a subscription model link to hardware. I disagree.

It is a valid option, though not always. The important variable to consider is also the product’s lifespan. If I can increase the margin to incorporate future cash flows of subscription into the model, I’d rather charge one time than force users to pay monthly as well, especially if it's likely a small amount. The case is particularly strong if it's reasonable to expect that a happy customer will buy a newer model after a few years. That's where you can achieve continuity. Smartphone producers are masters of that model.

I saw a hardware startup that wanted to charge $3,000 for a solution and then $25 per year of cloud subscription on top. I think that's a distraction. It's not always such a clear cut case, but my point is that subscription is not always the solution. Plain economics often is. Calculate a product’s lifespan and see what's the best model to maximise revenue from a client. Do not forget that invoicing costs.

Competitive positioning of hardware startups

There are a lot of difficulties and challenges when dealing with a hardware business; however, I find a some quite important and different from software startups.

A hardware startup's competitive position is usually better than its software counterpart, in particular because of issues mentioned above regarding capital and timing requirements. If a startup needs to go through a pile of issues before it can launch a product, you can be sure than when launched, any competitor will  face a significant uphill battle. There is of course a downside to this, if a large incumbent decides to go after the same market. However, big companies are usually slow and limited by a lot of internal committees and limitations, so this can be an advantage. Speeding up time-to-market is the key, and here is where an investor can make a difference by supporting the team.

Exits seem to be more natural in the hardware business, because big electronics companies are used to buying technologies and market share. These are mature markets and they know what they want. That also means that exits are less opportunistic (smaller) but better justified (cash flow). Which again for a VC can cause issues, but can be much more reliable. There are always exceptions in the unicorn club like Nest or Oculus Rift, which in proportion to the number of funded hardware startups actually makes hardware startups one of the most lucrative investments. However, the numbers are still too small to draw a conclusion. In any case, my personal take is that hardware is a real business that can be measured, evaluated and therefore pitched to a wider investment/exit audience.

Switching costs in hardware are higher, unless the startup is producing a $5 gadget. People seem to develop a closer connection to a device at hand than to a service online, especially when it's not an integral part of their daily life or business. We seem to have a reflex against throwing away a perfectly good device, while looking another way is not a problem when a switching an online service takes only signing up on another web page. So locking in a customer is actually cheaper after you overcome the sales process costs. That's quite the opposite to software services. Signing up is actually cheaper, but keeping a customer can be expensive. So again, it's not better or worse, it's just different.

Why I wrote this Guide? 

It's up to investors to figure out what makes sense for the business they invest in.  However, directly copying software models when analyzing hardware is dangerous and often leads to bad decisions.

I have always been wondering: Where do hardware investors get their training from?

There is plenty of resources available for entrepreneurs. Blogs, books, accelerators seem to be popping up like mushrooms ranging from very useless advices like: “Hire a world class team!” to actionable frameworks. A training or a book does not a successful founder make, however it does straighten up the vocabulary and at least gives a small taste of what they’re getting into if deciding to go the startup route.

But I couldn’t find anything similar for investors or those who are interested in working closely with hardware startups. And yet most of the people I talk to seem to be very confident in their experience when discussing the business of hardware. There are some exceptions, however in general they couldn't have gotten it from past deals, because there have simply not been enough hardware startups or exits around to gain meaningful domain experience. What usually happens is that investors (at best) try to apply the wisdom from software investing into hardware, by adding atoms to bits. Unfortunately it’s never that simple.

Am I qualified to write it? Probably not. In past, I ran several software startups and fundraised in Europe and the US, lead a successful accelerator, and now I run a hardware startup. I’m in contact with investors on a daily basis, but have never been a VC.

The issues outlined here will give you the tools to better understand your investment and evaluate it. In my view, a hardware startup business is an order of magnitude more complex than a software business with similar resources. However, I also believe that every issue that makes a hardware startup more complex is actually a potential competitive advantage. Because a competitor will have to deal with the same issues. And this is exactly where investors can make a difference!

The Guide is published on this site and will be occasionally updated. So please do comment on it. I’ll do my best to keep it alive and current. The Guide is published under the GNU public licence. So, basically, do whatever you want with it, just reference it wherever you use it.