An insider’s handbook for IoT startups

Tips are everywhere. There’s an abundance of generic advice available for how to build teams and culture, how to fundraise, how to be productive, how to stay above the noise… But what about specific advice? Specific tips for what differentiates the leaders from the followers. And even more specifically, a guide to differentiate and succeed as a hardware startup.

We’ve learned a ton while working on Notion over the past couple of years. We’ve had some big wins and some big losses. We read the blogs, read the books, attended conferences, focused on best practices and did our best to be a walking TED talk.

What we really could have used, though, was an insider’s guide; a sherpa for navigating the common challenges all IoT startups will face, a treasure map to the best-kept secrets. Although we weren’t able to track down such a coveted document along our path, we figured it was time to put one together for those who are finding their way now. Without further delay, here are a number of insider tips for building a great IoT company and product.

Be cognizant of the innovator’s feature curve

Ideas! Innovation! Changing the world! There will always be improvements you know you can make to your product. But, as the second tip alludes to, a minimum viable product (MVP) should be the main the focus — what is the least functional, the least capable implementation of your product that is commercially viable? Go build that! Expand on it later.

Take solace in the plot below, which seemingly illustrates most startups (hardware or software). You’ll have an idea, expand on it before you build it, realize you don’t have the time or resources to make it as feature-rich as you’d hoped, make your way through development and, finally, launch — with fewer features than you thought possible at the beginning. But, if you follow the second tip, that will all be okay.

Screen Shot 2016-02-16 at 2.52.49 PM

Clearly define MVP

By definition, if your product is at MVP, it’s viable. Therefore, there are no additional features needed, no additional bells and whistles. Could it be better? Of course! Should it be? No!

Paralysis by analysis can cause many an argument amongst teams.

MVP is key. And, defining what MVP means for your product and team is paramount. To give you an idea, our overview document is 10 pages and outlines more than 150 features in excruciating detail. They range in complexity and depth, but a few examples are: “As a user, I should be able to access support via email” (on the simpler end) and “As a sensor, I should be able to be adhered to any household surface” (on the more challenging end).

From here, we have two additional frameworks: “acceptance criteria” and “definition of done.” For the latter feature above, here’s a reference outline:

Screen Shot 2016-02-16 at 11.25.12 AM

Implement an “80/20 Rule” (or even the lesser-known “70/30 Rule”)

Data makes for informed decisions. That’s easy and well understood. But, how much data is needed to make these so-called “informed decisions”?

Paralysis by analysis can cause many an argument amongst teams. If you’re waiting until you’re 100 percent comfortable with a decision, you’ve waited too long. Startups aren’t afforded the discretionary time to make for such clear-cut decisions. When you’re 80 percent sure (or even 70 percent sure), move ahead. You must accept the risk of being wrong or you’ll never get to the finish line.

Get an early start on compliance

Does your product require compliance? FCC, PTCRB, WEEE, RoHS declaration, UL, California’s Prop 65?

Will you have export restrictions? Do you have more than two batteries in your device (meaning you need an IATA battery handling label on every box)? How will you display your radio compliance IDs on your packaging? On your product?

These are all tough questions that can have somewhat ambiguous answers. Finding the right outfit to guide you through the process can be challenging. Get started early. Be proactive and build compliance research into your feasibility assessments and budgeting (FCC alone can run $10,000+). Here are a few companies that can likely help you: Intertek, Nemko and Shipwire.

Vet costs overseas, but manufacture locally to start

Unicorns won’t just make a single manufacturing run, right? Then give yourself the best chance to succeed and make that initial manufacturing run as simple as possible.

We’ve had to adjust processes for reflow, stenciling, AOI (automatic optical inspection), panelizing, de-panelizing, flashing, testing components, heat staking, assembly, ID assignments, packout, packaging and more. If our contract manufacturer wasn’t 30 minutes away, we’d have had a much harder time working through solutions.

Per-part costs will be higher initially, but all that matters is you know what your upside can be. Sell the dream to investors, “We manufacture for price X now but will be at price Y when we ramp overseas.” Know what price Y is in full detail and be ready to pull the trigger when the POs start rolling in.

Check out PCH Access and Dragon Innovation for more info on how to scale.

Containerize in the cloud

From test environments to server instances, the infrastructure for your product will be complex (to say the least!). As you scale, operating your own data centers can become expensive and unwieldy. The cloud offers the promise of on-demand pricing, but how do you ensure your services run the same on your dev machine as they do in production?

Be thoughtful, be transparent and be forthright.

Enter Docker, a container technology giving you portability, agility and consistency. Containers run independent of the hosts on which they are placed, giving you consistency from test, to dev and all the way to production. With open-source tools like Kubernetes, or AWS’ EC2 Container Service, running containers at scale in the cloud is becoming easier than ever.

Build in over-the-air updates and have a NUT

Hardware is hard. While diligent testing, unit tests and purposeful QA processes help, there are always going to be logic changes and feature adjustments. Get out in front of that need and plan for your over-the-air updating (OTAU) system. OTAUs give you the flexibility and capability to patch mistakes, upgrade services and, overall, provide a much better product as you move past MVP.

Once your OTAU plans are in place, test, test, test! We call our system the “NUT,” which stands for “never-ending update test.” We constantly run sensors through over-the-air changes alongside a series of diagnostics to help isolate any issues that pop up.

Focus on UX

This is your bread and butter. This is how you set yourself apart. If you have a consumer-facing product, make it consumer-proof. Make it intuitive, neighborly, intelligent. Make it purposeful and appealing. This takes work… lots of it. But, you’ll be all the better for it, we promise!

Run user tests in person, ask people not only if they’d buy your product but how much they’d pay for it, be critical and don’t make assumptions. Some great resources: UserTesting.com, Mechanical Turk and InVision.

Get a solid writer on your team

“So avoid using the word ‘very’ because it’s lazy. A man is not very tired, he is exhausted. Don’t use very sad, use morose.” — Dead Poets Society

Messaging is everything. Be thoughtful, be transparent and be forthright. Often times, engineers and creative/expressive writing don’t mix. Figure out how to bring that missing link to your team, whether it be through a professional copywriter or a full-time team member.

If you’re ready to launch, you waited too long.

An articulate, personable website, a useful, intuitive mobile application and a poignant branding campaign can go a long way toward success. Words can get you into accelerators, they can garner investor interest and they can ease the pain of tight circumstances. Write like your business depends on it… because it does!

Launch uncomfortably

Your product will never be ready. It’ll never be bug-free. Your product will never have five-star reviews from all users and perfect 10s for NPS surveys. There will always be improvements to be made, better interfaces to develop, better copy to implement.

But, you know what? That doesn’t matter! Stick with what you defined as your MVP. Get your product out there and let people use it. Get more feedback and iterate from there. A transparent company iterating through launch challenges is much better received than a silent company not acknowledging issues.

If you’re ready to launch, you waited too long.

Use “off-the-shelf” when you can

If you don’t have to build your own Wi-Fi solution, why should you? Just like knowing your overseas COGS, know what your upside is to spin your own component solutions, as well — but don’t do it… yet. You can save time and energy utilizing platforms like Electric Imp to connect your product. It’ll get you to launch more quickly and help you prove product market fit.

Once you’ve hit it big time, cost-engineer down to what you’ve known all along you could get to: Your bottom line will appreciate it, your investors will be happy because you did what you said you could do and you’ll have been on the market that much sooner because you didn’t belabor your team creating something that was already at your fingertips.

Invest in your sleep

This last one sounds a little silly, but stay with me here. Founders and members of early-stage startups tend to eat, sleep and breathe their product. It stands to reason that sleep and energy may suffer after long bouts with launch-the-product-itis. The best way to kick nagging colds, tackle tough travel periods and be at your best is to make sure you sleep. And sleep hard!

A great pillow could be the best investment you ever make. Get a memory foam mattress topper. Get a Hello Sense system to help you track sleep cycles and to wake you up at the best possible time within those sleep cycles. Get an air quality monitor, like Awair, to help you create an environment conducive to best sleep.

That’s it! Twelve insider tips to help you reach the top. We could go on and on (get your packaging figured out before it’s too late, crowdfund like a boss, build a model with recurring revenue, be thinking about unit tests for hardware QA…). It sounds like we just talked ourselves into a revision to the handbook — but hopefully, this gets you started!