Hacking the consumer electronics hardware start-up
Do not apply if you want an easy life
I recall being shocked at the suggestion by one of my lecturers (on the small business module) who said “if you want to earn good money and have an easy life get a job in a successful large company and hide”.
He went on to say that some people have pathological issues that compel them to work harder than everyone else and to take more risks. He spent most of the term explaining why starting a small company and growing it into a medium and then large business is the hardest thing you can do and that it should be avoided at all costs if you wanted an easy life (which he suggested was exactly the thing you should want).
Little did I know at the time but having an inherent bias for creativity and hard work would later have me leave the confines of a leading global tech company (which btw was definitely not an ‘easy life’) and jump straight into running a consumer hardware start-up focused on creating and launching an innovative new product into a new market.
Discussion of pathology aside it was the best decision I’ve ever made — enough time has now passed that I can objectively share some of the unique insights from that journey which will hopefully be useful to others embarking on a similar path.
Running any business can push your limits but there are some unique additional considerations to factor in when you are a tech start-up focused on entering the consumer domain with an innovative hardware product to create a new market.
Understanding your starting position
All start-ups share a similar range of challenges but for those in the consumer tech hardware space these can be more acute and detrimental to success.
Hardware businesses are significantly costlier than software businesses to get up and running and operate. Success needs access to deep pockets and time to create new products and categories; the routes to funding are more limited than pure software plays and the stakes are higher (on the flip side the opportunity is also high if you get it right).
Start-ups don’t have existing global brands, distribution channels and manufacturing they can leverage. These are very costly to build for product businesses as unlike software it’s hard to provide free trials or freemium pricing models to scale out initial users, ramping up manufacturing takes time and specialist expertise, managing global consumer product brands is very costly (circa £100M/year to run a global brand at scale) and getting ramp-up volumes wrong can kill your business before you even get going.
In the early stages hardware start-ups can be locked-out of the best technology solutions from component manufacturers who provide the best (and lowest cost) parts. As a result, bill of material (BOM) costs for prototypes and initial volumes are much higher than they would be at scale. Having to buy parts through distribution channels adds additional margin and lead-times that have to be absorbed. Business cases are predicated on getting to scale for profitability; all economies of scale rely on hitting volume. It’s hard to pivot a product business if it doesn’t hit your revenue targets.
Hardware product development cycles are much longer than software; product flexibility reduces rapidly over time as commitments are made on electronics, mechanics, manufacturing and launch dates — each major milestone adds significant dependencies making it harder to make changes and adjust to on-going user feedback. Constantly balancing the need for flexibility against the need to lock-down the design means you have to pay critical attention to ‘smoke-testing’ early versions of the proposition before making material commitments.
Running a consumer tech hardware business actually means you are simultaneously running specialist software and hardware teams who have to come together to create the end solution. You need to create a product, it’s software, apps and probably an accompanying online service — it’s like running a few start-up businesses in parallel where they all have hard-coded dependencies and disparate working styles and cultures.
You can start in your bedroom or shed but success needs specialists and a laser focus on your early users. Agile methodologies can benefit software but jar with hardware development cycles. The void between prototype and production gets deeper and wider the more a business looks to innovate on technology and market application — i.e. the further you move away from existing technology solutions and existing consumer behaviors.
At the extreme of these dynamics are start-up businesses that are simultaneously creating new technology AND markets.
Not all consumer hardware start-ups are the same
Start-ups that are faced with high levels of market AND technology risk (the ‘Create’ segment) have significantly more uncertainty to manage — there’s many more unknowns.
‘Create’ businesses are developing new technology solutions to meet market needs that are still evolving (or do not yet exist). There’s simply more unknowns.
Known unknowns — things you know are challenges that you don’t yet have a solution for but they’re at least on the plan.
Unknown unknowns — critical obstacles that you are not yet even aware exist. The more a project is pushing on new technical and market ground the more likely and the more crippling these are likely to be.
‘Pivot’ businesses have the benefit of fitting existing technologies to new markets (most of the uncertainty is on the go-to-market side).
‘Disrupt’ businesses are developing new technologies to meet existing market needs (requirements are known and uncertainty is on the technical side).
‘Compete’ businesses are applying existing tech to existing markets and competing on pricing/features (not usually a start-up focus), low risk in terms of uncertainty but not an appealing sector to enter.
Getting a ‘Create’ product to market takes resilience, adaptability and a constant re-assessment of what’s important — it’s also the most amount of exhilaration and frustration you will ever feel, all at the same time.
The following section digs into my hard-earned tips for hacking the product development process in a ‘create’ business.
Hacking the product development process
ACCESS DENIED
The leading chipset manufacturers are busy looking after the large volume consumer electronics giants (Apple, Samsung, Google, Amazon), you can sometimes buy their chipsets through distribution but you will not receive any direct support to fix bugs or add functionality. In early development stages you are likely to have a constrained choice of technologies.
‘Create’ businesses are usually pushing chipsets to provide new or unproven functionality which is likely to unearth a range of low-level bugs and discrepancies in data sheets that either take time to work around or inevitably lead to changes in the way functionality is implemented. Chasing global technology teams around the world for support on low-level bugs, that no-one else cares about, is not a good place to be (believe me).
Some chipset providers offer support via ODMs but this requires the handing over of technical IP details and comes with significant development fees. In ‘create’ projects this can add higher levels of risk as the ‘unknown unknowns’ will be very costly to work around when you are paying someone else to fix them — it takes the right team, culture and attitude to work through development challenges and maintain the original product vision.
Creating your own bespoke hardware operating system may appear to give you more control and flexibility on new features but this comes at the major expense of time/effort/cost; your team will end up spending their time re-writing drivers for basic functionality rather than focusing all your efforts on developing new product capability (and IP).
WORKAROUNDS
Work out who your strategic technology partners are early on so that you can work hard to sell them the vision and ensure they are on hand to support when needed, you need an aligned strategy with partners.
If you can’t get your first-choice partners interested enough then you’ll need to look wider — having good relationships and trust built over time with leading suppliers will pay dividends as you hit inevitable setbacks on the journey. Once you build dependencies on hardware components and partners they become critical to your success (whether you like it or not). You cannot afford to have them let you down but some will and you’ll need to quickly find new alternatives.
Understanding where you’re pinch points are will help you stay on-top of your strategic supply-chain risks — you will likely not be able to mitigate them all but at least you’ll be aware of the implications for when one or more of them go sour (I had a major manufacturing partner lose their main facility to fire and had a few of our critical components go onto global allocation through critical ramp-up periods).
If you can start with tried and tested technical reference designs even better — the further you deviate away from normal configurations the more likely there are a range of nasty surprises waiting for you. Try to prove out as much functionality as possible using development kits before you look at laying down new designs — this will also allow for early user testing of the proposition before you start to create new dependencies.
Try to find the best match of existing technology to your requirements and deviate as little as possible; to innovate there will be some areas where you have no choice but make those decisions knowingly; spend time and money on the things that will make your product (and IP) stand-out.
Don’t be swayed by new chipsets that are just coming to market — select an architecture that is proven, in full supply with an active codebase. Ideally you want to select a platform that you can build your next 1–3 products on as that’s what it will likely take you to be successful in the ‘Create’ sector. Over time more options will open up to you but by that point starting a whole new hardware design will be a tough decision.
DANCING VS MARCHING
In the early stages of production, you want a manufacturing partner who can fit with your iterative development culture, working with you to help hone the manufacturing process whilst you’re still refining the product and proposition. Most partners say they specialise in prototyping and getting you to production but the reality rarely lives up.
Electronics manufacturing service providers (EMS providers) are not setup to dance, they are built to march, and march they will to whatever tune you give them until you tell them to stop.
In the initial stages you want a partner that can flex with your needs; you want this:
But you ultimately you get this:
False-starts and on-going hiccups on the prototype and production lines can quickly stress any good-will or strategic support that existed within the partner for your project. Any loss in credibility soon filters down to the production team and you’ll start to see your production slots being bumped and less willingness for flexibility.
What started out as a partnership to get you to market can end up being a struggle for production time, costs and support.
Designing products and supporting operations for manufacturing requires specialists — if done properly this can be costly and time-consuming. If you start too early you will likely invest too much time/money into a product design that is very manufacturable that no-one wants to buy (ie. the product needs to change) but if you start too late you may end up with a product that everyone wants that cannot be manufactured cost effectively.
WORKAROUNDS
Don’t fully engage a manufacturing partner too early in the process. Prove out as much of the product proposition, mechanical design (3D models) and electronics as you can first. Use smaller partners to get you by until you’re ready to commit — then fully commit.
Once you’re clear on the product requirements engage expert industrial and mechanical designers and hire a supply-chain expert with experience taking consumer electronics into scale. Engage with EMS providers through the process but don’t officially appoint them until you have a more proven and manufacturable product design which they can then fine tune. Your supply-chain expert should have a good network of providers that can be called on to check your designs as they mature and to get tentative pricing on mechanics alongside electronic components — pay very close attention to design for manufacture/test/assembly (you want a product that can be assembled very quickly and easily). Spending time on good design upfront will save you lots of time and money in the long run.
When you’re ready, try and find a manufacturing partner who has local facilities that can be utilised in the early ramp-up stages along with larger scale/low cost facilities that can be switched on when the time if right. This will speed up problem solving and support when it’s most needed and means you (or your supply-chain head) can get to the factory quickly to work through the inevitable production issues that will occur through the first few months. This also avoids needing to hire local language managers for offsite facilities or having to spend lots of time out of the country at a critical time in developments.
It also provides some additional protection for any early IP worries associated with going offshore prior to you launching the product.
MANAGING LEAD-TIME RISK
The single biggest thing to manage once you prove out the initial design is your working capital and lead-time risk. The more certainty you can put into your foreword production forecast the happier all parties will be.
It’s usually inevitable that your most strategic components are also the most-costly, are on the longest lead-time, have the highest minimum order quantities and in the case of your mechanical parts have the highest upfront costs for tooling. You need to understand the lead-time time and cost dynamics of all your components in order to manage your ramp-up volumes effectively. It’s likely there will be anywhere up to a 6month lag between placing an order for a strategic component and having volume ready in the factory (or office) to start building.
This is why Kickstarter and other crowd-funding platforms have been such a hit for new hardware product ideas — they have created an environment where taking money for pre-orders many months ahead of shipments is culturally accepted. Taking pre-orders allows start-up businesses in this space to more accurately forecast initial volumes and ramp-up accordingly — it is still tricky to predict a shipping date as companies tend to take orders ahead of a product designs being ready (or in some cases even started) but anything that adds certainty into the order book is extremely valuable for working capital management; some crowd-funding customers may be happy waiting for product but components suppliers always want paying on time.
Tooling commitments for major mechanical parts create tangible project milestones as they are associated with relatively high costs and instantly create fixed dependencies for later stages of the design. If you need mechanical parts ready for initial user testing in June, then you likely need to start tooling in Feb/March (usually done in Far East with tools shipped back by boat) which means you need mechanical design done in Jan so the tooling company has time to design the tools. You may need to commit to the final product design 6 months before you start building the first real products with real parts (ie. outside of 3D prints and other prototypes). You will also likely be committing large amounts of cash upfront on molds so you want to make sure all mechanics dependencies are set before you do that — or you risk having to scrap parts, write-off tooling costs and incur major time setbacks (making mistakes or changing design can be very costly later on).
Again, the more bespoke your design the more cost, time and risk involved. If you are creating your own camera system, as we did, it’s closer to 9 months to get a final lens and camera module design and only then can you start tuning your image pipeline (software that runs the camera) and assembling your real product. We had no choice as we needed a small format, high quality, wide angle lens camera module for our application and there were none available to buy — we had to find the right suppliers and work with them to design our own (not a simple task when you want to start with relatively small volumes).
The early rig for focusing our camera modules.
WORKAROUNDS
Build a simple critical path chart with all your major dependencies listed and connected so you can make informed decisions about when to start tooling parts and when to start placing orders on components.
No matter how many times someone tells you it will only take three weeks to spin a new electronics board (PCB) it always takes twice that — there’s just too many process steps involved between the intricate PCB design, the PCB fabricator, creating the stencils, the SMT line, the components themselves (availability, fit, heat profiles) and test processes for it to run smoothly first time. I’ve been hit by all these at various times and sometimes all of them. When you get your first electronics boards back in the office you also need to allow time for the engineers to fire them up — if they don’t power on first time there will be much head-scratching between the hardware and software teams until someone works out why. If you’re lucky it will just need a quick swap over of some resistors with a soldering iron or update of software and if you’re unlucky you may be back to the start of the process again (PCB layout mistakes are very costly in time and money).
If the tooling manufacturer says he will have parts back in the UK for you on the 1st June they likely mean you’ll have their T0 samples which will require modifications; at best removal of flashing and polishing marks from the tools, at worst they may be out of tolerance and need tools to be adjusted or mold-flow may need changing to get a more even surface (optimization of pressure and heat of the materials injected into the molds). From the point you receive initial parts it’s safe to assume you’ll need another 3–4 versions before you are happy (ie. T3’s). Once you’re happy you then need to place an order and wait for the first shipment of volume parts to arrive — the whole process can still add months from the point you receive first samples.
As part of our product design there was a small hidden LED screen underneath the product case which required a semi translucent black material to be used for the parts (a very new material). From the users perspective it gave a seamless product finish with no signs of technology until the screen menu was required. From the mechanics perspective it added two extra months of mold-flow adjustment to remove the strange reflections within the material we were seeing under office lighting (and the ordering of a large batch of a very specialist polymer material) and many sleepless nights looking at the implications of changing the design.
Just make sure you allow enough contingency time for the creation and tuning of parts — when your supplier says the part will be ready on date x make sure you check what they mean by ‘ready’. If you’re doing something for the first time then it’s practical to assume it will need some rework of some sort to get into spec and likely 2–3 more updates after that for you to be happy.
As much as possible stick with standard parts and use your common sense when it comes to timings.
If you have to place risk orders for parts ahead of knowing demand do so in a planned manor; manage any planned ramp-up in stages to stagger working capital commitments and to ensure you prove out production processes at each stage (start with 10 units, then 100, then 200 and then move into 1k’s).
MANAGING TEAMS OF SPECIALISTS
Consumer hardware and software engineers can be breeds apart; their culture and attitude to risk are very different (Prince 2 vs Agile). Having specialists on the consumer hardware side tends to equate to older more experienced engineers who are used to working with program and project managers whereas software engineers with a few years-experience may already be at the cutting edge in their field and are used to iterative development cycles and design by doing.
Although both groups get into the office at different times of the day and may be generations apart they actually have a lot in common.
In order to create a product these groups need to work together with an overarching aim of delivering a beautiful user experience. That requires a deep appreciation of the similarities and differences on both sides and some lite-touch program management to ensure the critical dependencies are known and managed across both disciplines. Most technical obstacles that crop-up along the way require both hardware and software engineers to work together to find a solution (and also to identify the issue in the first place).
Once working together as a team great things can be achieved — this is one area where start-ups have a major advantage over the global players; there are no existing fiefdoms that get in the way of true collaborative working. Managing positive tension between the specialist areas means you are pushing at the right boundaries — if there’s no disagreement or decisions to be escalated then chances are there’s no real innovation going on.
Whilst discussing this with an old Nokia colleague he smiled and said, ah, yes… that’s what I call ‘“Mancotta: A mix of Ricotta (software engineers) and hard aged Manchego (hardware engineers), it probably shouldn’t work but it does”: credit. #Christian Lindholm
WORKAROUNDS
Across all stage of development, you will benefit from having your software and hardware engineers (and your industrial designer, mechanical designer, supply-chain chief) in the same office, sat round the same table, solving and discussing problems together. Most challenges will require an integrated solution and will benefit from informal discussion and review (alongside more formal sessions).
If this is not possible you should make sure you can get everyone together at critical review points and have the right tools in place to allow open discussion between teams.
With many ‘experts’ in the room everything requires a decision; testing which constraints are real or whether they can be overcome in a timely manner is the only way to keep developments in the right direction. You need a team you can trust and who learn to trust and respect each other as no-one can be an expert in all the disciplines. You need to be the champion of the product; you need to decide when to hold fast or deviate from the original specification and when to call time on a specific technical challenge that is proving tricky to solve.
Only a clear product vision and mutual trust/respect can allow all specialisms to pull together in the same direction.
Craft ale also helps.
Why I wrote this
I wrote this to share some of the product development insights and workarounds I picked-up whilst creating Autographer: The world’s first intelligent wearable camera and it’s follow-on intelligent imaging platform.
Moving from Nokia to Autographer in 2012 provided a comparative context for viewing the product development challenge through the lenses of a small vs global business — it also provided an invaluable opportunity to innovate on new consumer technology and launch it into market at a time when intelligent imaging was in a very nascent stage.
Autographer launched just ahead of Google Glass and other wearables which quickly turned the market attention to questions of privacy, product form and user experience. Designing products in parallel with creating new markets focusses attentions forensically onto early users and requires a deep understanding of all interconnected interdependencies of the end solution.
Our developments in this ‘Create’ space led to the creation of valuable IP which allowed for future licensing into other adjacent sectors and inspired the creation of a new AI business focused on classifying and understanding the growing volume of images and video being captured across the world.
Who wants an easy life anyway ;-)
https://www.youtube.com/watch?v=ppsr0B-YhPI
https://www.pocket-lint.com/cameras/reviews/122616-omg-life-autographer-review
https://www.theguardian.com/travel/2014/feb/27/autographer-wearable-camera-review-berlin-city-break
This story is published in The Startup, Medium’s largest entrepreneurship publication followed by +368,675 people.
Subscribe to receive our top stories here.
