Repeat after me: Hardware is NOT software

The Lean Startup / Customer Development, as most startup people know, was started by Eric Ries (author of “The Lean Startup“) and Steve Blank (author of “The Four Steps to the Epiphany” and “The Startup Owner’s Manual“.

The basic tenet of Customer Development is to invest heavily in customer research and underinvest in product development – relying on a low fidelity minimum viable product to test the business model, and iterate until you get it right before really launching into a product development process.

There are four steps to this process, outlined in this illustration from a 2012 post about the Stanford class “The Lean Launchpad on line” by Steve Blank.

 

Customer Development

In brief, the four steps are as follows:

  • Customer discovery – where one works to understand who the customer is and search for problem/solution fit
  • Customer validation – where one starts selling the MVP and iterates until a repeatable sales process is found
  • Customer creation – grow customer base from early adopters into the mainstream
  • Company building – re-look at the organization structure and build out the team to execute the business model

The Customer Development framework further stipulates that these four steps involve the search for a business model using a build-measure-learn philosophy. After a scalable business model is arrived at, and product/market fit is found, execution then follows. Execution is defined as actual product development, involving product management and planned engineering development.

Being a hardware/software product geek, I am nodding my head all the way to Step 2, where I stopped cold.

Let’s look at these four steps and what’s involved to go through these steps with a hardware product. I am going to reuse my connected device consumer electronics product as an example.

  • Step 1: Customer Discovery. This is of course the right thing to do, and any product person worth their salt would start there.  Too many startups become obsessed with the technology or their concept of the product and miss the boat on understanding market and customer needs and wants.  “Stop building the product and get out of the building, NOW!” is what we say to startup teams over and over.
  • Step 2: Customer Validation. This is where we run into trouble for a product with hardware components.  The keyword is “sales” in the “repeatable sales process”. You see, for a software product that, say, has a cloud backend, some big data algorithms and a mobile and web front end, you can start charging from day 1 and begin iterating on the sale process immediately with a very minimal, low fidelity MVP.  For a hardware product, there is no such thing as a saleable low fidelity MVP.  Would you:
    • Buy a brand new eco-friendly dishwasher for your house, if you hear that 30% of the shipped units leak sudsy water all over the kitchen?
    • Buy a new, non-invasive blood glucose meter for your elderly grandparent suffering from diabetes, if it produces a correct reading more than half the time?
    • Buy a new car that gets rid of parallel parking by a novel mechanism that rotates the wheels 90 degrees and scoots directly into narrow street parking spaces, that hasn’t got a complete set of crash and durability test results in their attempt to minimize the time to market?

Oh, by the way, did we mention regulatory approval and product marks? If these words are new to you, I invite you to turn over your computer’s power supply and have a look at the sticker.  You will most likely see a “UL” mark.  That is a fire and safety compliance mark that tells customers your product can be safely plugged into the wall without blowing stuff up.  If you turn over your mobile phone (you may have to open the battery compartment for Android phones), you will see a number of other product marks, like FCC and CE.  The former is an emissions mark necessary for any product with any kind of radio inside (2.4GHz… Bluetooth… Cellular… they aren’t picky) to be legally sold in the US.  The latter is a compliance mark to allow your product to be legally sold in Europe.  There are nationally recognized laboratories that test against safety, emissions and other applicable standards and then issue you these product marks.  Without these marks, you cannot sell the product legally.  And the regulatory approval process for a new product can take 4-12 weeks.

You see, for hardware products, there really is no such thing as a saleable low fidelity MVP. You either make a quality product, or you forfeit your opportunity to test any kind of a real business model.  The best you can do is to collect donation money in return for providing a prototype to customers – you cannot list on Amazon, you cannot get placed in Best Buy or Apple or Target or Walmart without going through all these steps.

The other concept that gives me pause is the idea that you do not need to hone your execution engine to create a hardware product, which again is not terribly applicable to a hardware/software product of any reasonable complexity.  Coming back to the connected device example, the development process is long (measured in months), costly (measured in hundreds of thousands of dollars, frequently exceeding $1M even for simple products) and fraught with risks.  If you do not have well tuned product management, development, and manufacturing organizations that are really excellent at execution, you will find yourself in a quagmire of project delays and budget overruns.  You could run out of money before you can get to first customer ship (the dreaded tail wagging the dog in the Customer Development framework).

Why does hardware not play nice in this framework? Very simple: hardware is not software, and the vast majority of lean/customer development case studies and examples are based on software products.  Software development has a completely different cadence, and with the advent of agile development, it plays excellently well with this whole philosophy.  Hardware – not so much.

So hardware folks are stuck in the bad old days with our stage gate process.  What can we do to embrace key tenets of Lean and outside-in thinking to maximize our chance of success?

  • REALLY invest in Step 1 – Customer Discovery – before embarking on a product development process. Take as long as you need to truly understand the market and the customers.  Look for comparables.  Founders should definitely do this themselves and on a very tight budget.  An idea with no market is a non-viable idea. Learn how to do qualitative research properly.  Conduct detailed / contextural interviews.  Do observation studies. Do immersive studies.  Anything and everything to maximize your understanding of the customer before you proceed.
  • Embrace the “M” in “MVP” when defining the product. Avoid feature bloat – put in the fewest possible hardware features to reduce development time and risk.
  • Architect the product in such a way that the product can become more and more capable with a series of software releases running on the same hardware platform.
  • Incorporate continuous customer research as an ethos while working on Step 2, Customer Validation, which starts with creating a saleable product.  Test anything and everything with customers.  Test with storyboards, foam models, printed models, Duct Tape Prototypes while waiting for the first engineering prototype to come out. Insert 4-6 weeks of customer testing time between the engineering prototype phase and the engineering verification phase. Be prepared to pull the plug on the whole thing and go back to Step 1 if you discover thing about the product that disproves your hypothesis.
  • During the first engineering prototyping phase, employ rapid prototyping techniques to iterate through design concepts as quickly as you can, and test every build as you go.
  • For a connected device with custom electronics, consider making a semi-Frankenstein prototype using external development kits to circumvent the lead times for custom PCBA procurement. For example, you can buy Bluetooth modules that will be too big to fit into your wearable connected device, but you can “fake” the functionality by running a wire from the module to your wearable devices and get going with customer research that way.
  • Conserve runway – drag your feet on building out the rest of the organization outside product, engineering and manufacturing until relatively close to FCS. Consider using contractors instead of full time hires for sales and marketing personnel, in case the sales model changes. For instance, channel sales is different from direct sales and the personnel you would recruit and the organization structure would be vastly different. Hedge if you aren’t sure which way will stick.

Having done all that, you could still end up with a product that doesn’t fly the way you think it should.  Coming up with a second product to test a different business model would be months or years out.  So the proper build-measure-learn loop is long and hard to do for a hardware startup.  But by incorporating lean and customer development principles throughout the development cycle, and making time for customer research, your odds of success will be much better than if you did it the old fashioned way all by yourself inside the building.  Go forth and do some research!

Leave a Reply