Blog

Perspectives on entrepreneurship, startups and venture capital from K9 Ventures.

You Can’t Spell Hardware without H-A-R-D

Tweet about this on TwitterShare on FacebookShare on LinkedInShare on Google+Email this to someone

I started my tinkering with hardware first. As a kid I used to trek down to the “electronics bazaar” in New Delhi, known as Lajpat Rai Market, and go store to store to try and find the components I needed for whatever I was building. It was both fun and frustrating at the same time. It was fun as it was a scavenger hunt to find parts, but it was also frustrating as most of the vendors would look at me (not even a teenager yet) and not take me seriously. Computing and software came a few years later for me as access to computers wasn’t as easy or affordable.

But as one of the side benefits of having grown up being comfortable with 555 chips, PNP and NPN transistors, diodes, rectifiers, resistors, capacitors, plastic housings, wire cutters, and most of all soldering, I’m pretty damn comfortable being around hardware. In fact, the smell of melting solder and flux, still smells heavenly to me. (Yes, I’ve since learned that it’s probably toxic and not something I should be taking whiffs off!)

Software is awesome as it gives you instant gratification. You make a change, compile, and voila you can see the effect of what you did. Hardware is different. It requires a higher level of discipline than software does. In fact, in college I used to joke that Electrical and Computer Engineers could write better code that Computer Scientists (not something I would argue about as voraciously now).

But building physical product provides a different level of satisfaction. There’s something cool about holding a physical product in your hand and knowing that you had a part in creating it. I think it’s with that in mind that I’ve been pretty open to investing in hardware companies as well as software companies.

To date K9 Ventures has invested in the following hardware companies:

  • Lytro (f/k/a Refocus Imaging) — light field imaging camera
  • Occipital — makers of the Structure Sensor for 3-D data acquisition
  • Nimble VR — a 3-D sensor for tracking hand gestures in VR (acquired by Facebook/Oculus)
  • Coin — a programmable payment device in a credit card form factor
  • Osmo — a Computer Vision based platform for the iPad at the intersection of Education, Learning, and Entertainment
  • And 2 seed stage companies that haven’t launched yet

That’s 7 out of about 29 investments that K9 has made to date and so that’s almost 1/4 of the K9 portfolio. In addition, I’ve looked at several other hardware companies that K9 didn’t end up investing in for one reason or another. Over the course of working with all of these companies, and having looked at several more, one thing is abundantly clear:

“You can’t spell Hardware without H-A-R-D”

This is not something new and unexpected. I knew this going into it. But having worked with multiple hardware companies gives you a better appreciation for just the breadth of issues that can rise in getting getting a hardware product to market. In fact, I would state that building a hardware company is at least 3x harder than building a software company.

Last year we hosted a K9 Founders’ Hardware Geekout at the The Kennel and talked through a bunch of the issues that arise in hardware companies. Here is the preliminary list from our whiteboard and I’ll provide some color on each below:

IMG_6784

Less Room for Error: In a software company, if you screw something up, it’s generally an easy fix — you can push an update, and within a short time of finding the problem, identifying the fix, you can push an update to fix the problem. With hardware you have less room for error. A mistake along the way can turn out to be incredibly costly and time consuming if it means having to spin a new board, source a new component, or maybe even redo the tooling for the manufacturing. All this implies that you have less room for errors and an abundance of caution and planning is often better than a lot of trial and error.

Expensive iterations: Depending on the product doing a new rev of the hardware may cost thousands of dollars or more and could also take weeks if not at least several days. This is especially true if you’re trying to do something that has never been done and therefore doesn’t rely on off-the-shelf parts. Custom parts can also sometimes lead you down a rabbit-hole of building your own equipment to build those custom parts (and yes, I’ve had multiple companies that have had to do this!). It’s one thing for a company at the scale of Apple to be developing custom machinery for their processes, it’s a whole other ball-game for a tiny startup to take on something like this (I strongly recommend against it unless there is no other viable option for it).

Prototyping: Prototyping has come a long way and is way easier than where it used to be. But it’s still time consuming, and expensive. 3-D printing has revolutionized the ability to do quick revs, but to date I’ve never had a company that’s been able to 3-D print their parts on a commodity 3-D printer. They almost always have to go to Shapeways or some other source to have the 3-D parts printed, which is still a couple of days turnaround time. Sometimes building the prototypes also requires expensive test equipment which cannot be found at your neighborhood TechShop.

Emotional DesignIndustrial Design: Design matters. Especially if you’re building a consumer hardware product; as consumer expectations have been shaped by companies like Apple and that’s the level of design and attention to detail consumers have come to expect. If you’re building an industrial device, you might be able to get away with a box of sheet-metal, but even that is becoming more and more unlikely.

Design is a key component of the product and could well become the reason why people buy your product over a competitor’s product. As Don Norman outlines in his book Emotional Design — humans are emotional beings and the emotions that the design of the product creates can play an important part in how a user perceives the quality and the functionality of the product.

Rarely does a startup produce an awe inspiring product design by doing the design in-house. Almost all the iconic startup hardware products that you see on the market are designed by professional industrial designers. These design houses do a phenomenal job, but it is up to the founders to choose the right firm for their product, and then manage the cost and timeline for the design process. Some of the firms that are well known for their work in the space include: Bould Design, New Deal Design, fuseproject, Moto, and Ammunition.

By User:Wikimol, User:Dschwen - Own work based on images Image:Lorenz system r28 s10 b2-6666.png by User:Wikimol and Image:Lorenz attractor.svg by User:Dschwen, CC BY-SA 3.0, https://commons.wikimedia.org/w/index.php?curid=495592

Sourcing: Ah, sourcing. Step 1 of sourcing is finding the parts you need to begin with. You would think this is easy in the days of the Internet, but you’ll be surprised that just typing a query into Google doesn’t always yield the results you’re looking for. Step 2 of sourcing is finding the parts you need at the right mix of price, quality, and, reliability of the vendor. In at least one case, one of my portfolio companies wasn’t able to source the parts they would have really wanted because Apple had locked up the entire supply chain for that part!

Sourcing may also involve several trips to China, India, Thailand, or other countries and cities that you may not otherwise consider traveling to for vacation. Sourcing also means that you typically need to find someone who can speak the local language, help you get around, order the right food to eat, stay in a less sketchy neighborhood, and most of all, help to translate and negotiate with the vendors you’re considering.

Sourcing also means doing double duty as you almost always want to eventually end up with dual-sourcing — ie. having multiple vendors for the components you need. That’s the only way to ensure that you don’t have a single point of failure in your supply chain and also can provide the necessary leverage to drive down your BoM cost. In at least one case one of my portfolio companies was delayed an additional 4-6 weeks due to a typhoon that resulted in heavy flooding in Thailand, which impacted the factory that made a component of a component! When working with hardware you develop a new appreciation for the Butterfly Effect.

Supply Chain: When building a hardware product you’re at the mercy of so many schedules that you cannot control. You may have everything ready to go, but for some reason the manufacturer of the chip you need may have sold out and now has a 6-8 week lead time on new deliveries! Forecasting production quantities, planning for how many parts you will need and when, accounting for the lead times on those parts is all a complicated endeavour. It’s no wonder that finding a hardware company that managed to deliver on schedule is a rare occurrence.

Manufacturing: Finding the right manufacturer for your product is a process in itself and often requires getting connections to the right people if you want to be able to work with a larger manufacturer. Once you’ve chosen a manufacturer, getting the line up and running with them is another process. And even when you have the line up and running, there can be bizarre problems that pop up. I’ll give three examples (all of which have happened with my portfolio companies):

  • In one case, all the units coming off the manufacturing line were not passing their tests. After much hustling and eventually having someone travel to China to figure out what’s going on, they traced the problem down to a faulty network cable in the manufacturing facility. Yup, a $0.50 RJ-45 cable.
  • In another case, a worker on the line pulled the wrong roll of resistors from the parts room and so 4,000 units got assembled with the wrong part and were consequently defective and required manual work to refurbish them.
  • And in yet another case, the lithium batteries were being stored in a location which was exposed to moisture (You can search Youtube for what happens when lithium comes in contact with water!)
  • As a bonus example, in one case the manufacturer decided to substitute an “equivalent” part without telling the company that they were making this change. Try tracking that failure mode down!

Inventory: When you build physical things, you need to hold them and store them, before you can sell them. This introduces new risks in terms of forecasting, financing, and logistics. How many units should you build? How do you pay for those units? Where should those units be stored so they can be shipped expediently and cost effectively? Although there are many cases of companies ending up with excess inventory, that they then need to liquidate at or below cost in order to recoup the cash that’s tied up in it or due to inevitable obsolescence.

Among the fun inventory stories, twice one of my portfolio companies has ended up forecasting less than what they could have sold. It sucks when it happens as you miss the window for servicing the demand that exists, but it’s a much better problem to have than to end up with a lot of excess inventory which doesn’t move.

There’s another fun story that I can’t get into details about, but it required the founders to make a bold and gutsy move to ensure that they could get enough of of the raw materials for their product for the foreseeable future.

Logistics: Most companies will outsource their logistics to a logistics provider. It’s not worth trying to do this in house. A plug here for Amazon — they’ve taken the logistics to a whole new level and might be one of the best avenues for getting started. But even then making sure that your provider is shipping things out on time, processing returns the right way, etc. is all stuff that could eventually require additional attention.

One of the best logistics stories I was told was where the company opted to use ocean shipping in order to save on the cost of air freight. They packaged their palettes into containers and weeks later when the containers arrived State-side, they found that moisture from the ocean voyage had soiled the exterior of the packaging!

It’s little things like that that require an attention to detail and even then, it’s almost impossible to account for all the things that can go wrong. Sometimes containers just fall off and end up at the bottom of the ocean or get damaged in transit.

Certification: Hardware devices may require different levels of certification. If you use WiFi or Bluetooth, you may need FCC certification. For Europe you need CE certification. Safety certifications like UL certification in the US or CSA certification for Canada. All of this adds cost and time and most of all unpredictability to the schedules for shipping a hardware product.

Software: Oh, and don’t forget that even once you build the hardware device, it’s mostly useless without the accompanying software. As Brad Feld eloquently puts it, most hardware these days is simply software wrapped in plastic. This means that in addition to shipping the hardware product the company also needs to do a kick-ass job at building out its software and services platform. The software becomes just one leg of the stool for hardware, whereas in a software company it is the core product.

Accounting: Having raw material, inventory, credit lines, consignment orders, returns, labor costs, shipping costs, all adds more complexity to even the accounting process. In most of my hardware companies we’ve had to hire a CFO who has experience with a hardware business way earlier than we would normally consider hiring a CFO at a software company.

Packaging: The unboxing experience for your product matters. This goes back to the point that the consumer expectations are set high for design. The same applies for packaging as well. The packaging must not only be attractive, practical, sustainable, and put your product front and center. If you plan to launch through retail channels, then different retailers may have their own packaging requirements to meet their specific store needs. Sometimes even what colors you are allowed on the outside of the box can be dictated by the retailers!

Support: If your product requires installation of setup, then it will most likely require support as well. No matter how much you try to simplify the process, people *will* find a way to trip themselves up and contact you with a myriad of questions. If you don’t provide the support, you will have exceptionally high return rates. So you have to plan on having great user experience, but still provide for support. That support cost needs to be built-in to the cost of the product you’re providing. In my opinion outsourcing support is a terrible idea for a young company.

Launch/Pre-order/Crowdfunding: Do you do a Kickstarter campaign for crowdfunding? Should you do a Self-starter campaign for pre-orders? Should you simply take reservations in order to test demand and prove product-market fit? I’m purposely posing questions here as the answers may wary depending on the company, the product, the market, the funding environment, and more.

If you do a launch of any kind, you will most definitely need a video. The production quality of videos has also gone up to keep pace with user expectations and that also means that the cost of producing such videos has also risen accordingly. (Although I must admit that I’ve become somewhat tired of the formulaic videos that you see on Kickstarter).

The video quality matters a lot. You have to convincingly tell a story within a minute and half. My magic number for the length of a video is 1 minute and 42 seconds. I’m pulling that out of my ass, but so far it’s served well 🙂 But the same video needs to be cut and recut to create a 30 second version that is catchy within the first 2-3 seconds so that it can attract someone’s attention in the social media feed (Hat tip to David Marcus for alerting me to this).

If you do a launch it’s rare for the launch to take of organically, it must also be accompanied by a level of paid distribution (i.e. advertising) on Facebook, Twitter, Google, and be preceded by a press and PR campaign to try and get coverage in the relevant press outlets and blogs.

Marketing: Assuming you make it through the initial launch marketing described above, once your product is shipping, you will then need to figure out growth marketing. What channels work best for you product? Do you continue to rely on online marketing or do you also need to look at broadcast/television advertising (whole other ball of wax!), content marketing, partner marketing,

At a very simple level, how does the copy, the graphics and the color of your packaging impact the sell through in retail stores?

Distribution: Most startups tend to launch on either Kickstarter or on their own site. But very soon they need to expand to figure out other distribution channels. For a physical product, you cannot under-estimate the power of instant gratification, or “Free two day shipping,” or “Ships with Amazon Prime.” If you have hit product-market fit, then you need to reduce the friction for people to be able to buy your product by doing distribution deals. Getting into retail stores like BestBuy, Target, Walmart, Costco, Brookstone, Apple all have their own unique quirks. Some will place an order and buy the product from you outright and then liquidate it on their own if it doesn’t sell. Others will return any unsold product to you.Some will only work through intermediaries/distributors. Some will pay you every week, every month, and others won’t pay you for months.

Returns/Repairs: Stuff breaks. Sometimes due to manufacturing defects, sometimes due to mishandling in shipping, sometimes due to user error,, and sometimes it’s just Murphy’s Law. You have to be prepared to process returns and also maintain inventory on hand to service those returns with replacements or refunds. Nothing is more expensive than the ill-will of your customers and I’m pleased to see how many startups have taken this to heart and done a phenomenal job of their customer service and returns.

Financing/Forecasting: Financing a hardware company has it’s own challenges. Few investors will look at hardware companies for venture financing. Of those, even fewer will take on technology risk and market risk. Most of them tend to look for products that are already launched and are demonstrating some level of product-market fit.

At the same time given all the complexities outlined above, hardware companies require more capital than a software company might require at a similar stage. They also need the capital in a timely manner, as it’s much harder to bootstrap a hardware company when you have to pay for equipment, parts, tooling, manufacturing, certification and more.

There are additional source of capital like venture debt and receivable lines, which lend themselves better to hardware companies, but they all require companies to already be shipping product and be able to reasonably accurately predict what their cashflows are going to be like.

 

So with all that, I think it’s pretty darn clear that hardware should really be spelled HARDware.

That said, I still love HARDware. There is something about building a physical product that gets me excited more than an app or a website (though there are some apps and websites that I get pretty darn excited about too!).

Having funded 7 hardware companies, one does start to see a lot of the pitfalls that the companies could face and the learnings from one company’s experience can help another company to avoid the same pitfalls. That was the entire point of the K9 Founders’ Hardware Geekout — to bring together the founders of hardware companies so they can learn from each other and thereby reduce the chances of making the same mistakes, and increase their chances of being aware of the possible landmines along their path.

K9’s Thesis on Hardware

K9’s thesis on hardware is to focus on hardware that enables software. Selling yet another commoditized widget is uninteresting to me, but opening up a whole new platform and ecosystem by creating the hardware that enables it, is super interesting.

In addition to enabling additional software, sometimes hardware can also enable new revenue models. This is particularly true in cases where the software ecosystem is commoditized/free (think mobile apps), and by introducing a hardware component you can change the customer’s perception of value (Osmo does this very effectively), or in other cases building a recurring/subscription revenue model with a very high attach rate with every hardware device sold (Dropcam did this very effectively and Ring is following suit).

 

In closing I’ll remind everyone that yes, while hardware is HARD, the world’s most valuable technology company, Apple (Nasdaq:AAPL), is a company that makes most of its revenue from the hardware they sell.

It’s not all gloom and doom if you want to do a hardware company, but you must remember that sometimes the greatest opportunities require taking the path less traveled and sometimes even creating your own path (ala Tesla).

 

You can follow me on Twitter at @ManuKumar or @K9Ventures for just the K9 Ventures related tweets. K9 Ventures is also on Facebook and Google+.

  • Mihnea de Vries

    Excellent article, probably the best i’ve read so far about what a hardware startup means. But as a hardware startup founder, instead saying that hardware start-ups are 3 times harder than software startups, i would say that doing software is 3 times easier and hardware 3 times more rewarding.

  • Pingback: Opposing Viewpoints on the Difficulty of Hardware Startups #ManufacturingMonday « Adafruit Industries – Makers, hackers, artists, designers and engineers!()

  • Roi Birger

    Great article!

    Having walked the HARDware path before, I can only state that its a steep uphill affair, yet rewarding.
    If I may add to the challenges mentioned above:

    Properly scaling a hardware business (even if all goes well!) often requires a major change in underlying tech in order to reach a desired bill-of-materials.
    In other words – you can’t really “play” with pricing/margins based on change of strategy alone – improving BOM will cost dearly in NRE.

    And don’t get me started on the dreaded thermal fiasco (“the damn thing is too hot”) – most likely to introduce itself weeks before product release – this might send you weeping back to the drawing board..

    Cheers,
    Roi Birger