My team’s project, the Spark Core, just announced our second delay, putting us a total of 7 weeks behind schedule. We keep track of our project status here.
We’ve learned an extraordinary amount since our campaign launched on May 2. We’ve had many breakthroughs, and we’ve made some mistakes along the way too. As we approach delivery, I wanted to take the time to share some of what we’ve learned and help answer the oft-asked question: why does every Kickstarter project seem to get delayed?
If you’re not familiar with Kickstarter, let me give you a brief rundown. Creatives and entrepreneurs come up with an idea; they throw together a prototype, film a short video about it, and put it online to share with the masses and see if other folks are excited enough to contribute to bringing this idea to life. Contributions are exchanged for “rewards”, which in many cases equate to a pre-order of a new product, which are provided with an “estimated delivery date”.
Most projects (particularly hardware) are delayed by weeks, months, or sometimes more than a year. Delays tend to correlate with the complexity of the product, the level of overfunding, and the constraints on the team (e.g. are they doing this full time, is it a team or an individual, do they have other sources of funding, etc.). Sometimes the project creators run out of money and disappear, running from an angry mob of backers who believed in them once, and now want a refund.
I believe that in the end, most hardware project creators are waylaid by the same obstacles. Hopefully this will help other future Kickstarter project creators avoid the common pitfalls. But then again, it might not, because:
Founders are optimistic; that’s why we got into this in the first place
It doesn’t matter how many people tell you “Take the time you think it will take to manufacture your product, and double it.” It never sticks. Why? Because we’re optimists. Founders think we’re abnormal; we’re the exception to the rule. After all, if you think you’re just an average founder, then you should probably just close up shop here and now, because the majority of start-ups will fail, and being average puts you with the failing bunch.
So since we all think we’re above-average, we assume that the delays that hit everyone else won’t affect us. We watch other Kickstarter projects and we see how unrealistic their estimates were, and their mistakes look stupid and naive.
But the truth of the matter is that we all make mistakes, and we’re never fully prepared for the madness that ensues after the campaign closes, because:
Kickstarter prototypes tend to be rougher than you think
Video is a forgiving medium, as it’s easy to show the things that work and gloss over the things that don’t. Also, for many products, creating a beautiful cosmetic model can be very easy; you can get a high-quality SLA 3D printed model for ~$150, carefully apply a coat of spray paint, and you’re done.
For the Spark Core, the hardware was theoretically close to finished. But we got some great feedback during our campaign and decided to add some new features like a different antenna option and an RTC crystal for accurate time. These features extended our timeline in exchange for a better product, and most of us founders would gladly make that trade.
On the software side, I would say that roughly half of the features that we were promising were built when we launched the campaign. The other half we knew we could build, but there was still a lot of work ahead of us — and even the features that were already in place were super early in development. Although hardware is hard, we’ve put many more man-hours into the software than we have into the hardware, because we’re playing in relatively uncharted waters on the software side, whereas the hardware is more cut and dry.
Making a prototype “manufacturable” is tricky business
There’s a massive field of work around something called Design for Manufacturability, or DFM. At a high level, this means designing products to be easy to manufacture, with two potential outcomes: decreased manufacturing cost, and increased yield.
Yield (the percentage of products that come off the manufacturing line that are fully functional) is a tricky business, and factors into your costs. If 10% of your products fail on the manufacturing line, then your manufacturer has to charge you 10% extra for the functional units to cover the cost of the “scrap”. As your product becomes more complex, the failures compound and become both more common and more expensive.
So what do you do to fix this? You do DFM, making design tweaks that improve your yield by either relying on the experience of your manufacturing partner, doing sample runs and making tweaks based on the results, or both.
DFM isn’t necessarily rocket science, but it tends to tap into a different skill set and a different type of experience than most project creators have. It helps to have a great manufacturing partner and experienced advisors, but even so, most of us learn DFM on the job, and learning on the job slows us down.
Tooling for injection molded plastic sucks
Tooling, in the context of manufacturing, is the set of tools that must be built custom for your product. When you make something on a 3D printer, there’s no tooling; every different shape comes out of the same box, without modification. But when you move into injection molded plastic, your whole world changes.
Injection molded plastic requires a mold (natch). This mold will be hit with highly pressurized liquid goo, so it needs to pretty tough; the molds are usually created from very hard metals. The hardness of the mold makes it last for a while, so you might get 100,000 or 1,000,000 parts out of a mold before it starts to deform. But its hardness makes it very difficult to create in the first place because carving out a very hard metal is slow, precision work. And both “slow” and “precision” really mean expensive.
Tooling for an everyday product with a plastic enclosure might cost between $10,000 and $100,000. It can also take weeks or months to create. But even worse for a start-up, once you create a tool, you’re locked in.
Us entrepreneurs really like to iterate. Iterations make things better. And tooling kind of kills that for you, because it makes iterating effectively impossible, so you have to get all of your iteration done before you create your tools. This inevitably causes delays.
So what can you do about this? Avoid tooling, or find manufacturing methods that have cheaper tooling (laser cutting, thermoforming, etc.). We’re lucky, because with the Spark Core, we were able to avoid plastic entirely. Dev kits are nice that way.
Lead-times on components may often surprise you
When you go out to source components for your product, you’re generally trying to find sources with low prices and high availability, meaning that they’ve either got a bunch in stock or they can get it pretty quickly. The problem is that these two factors are in diametric opposition; the cheapest source is the factory, but their lead times can be 12-16 weeks because they have to, you know, make it. In contrast, distributors will hold inventory of the components and therefore get them to you much faster, but holding inventory is expensive, so their prices are higher; the greater the inventory they must hold to meet your needs, the larger price premium you’ll pay.
Inevitably many products have one or two components that fall into the general category of unobtainium, i.e. impossible to find, and that often causes delays. For the Spark Core, this component was the CC3000, our Wi-Fi module, which was an important part of the design and therefore could not be replaced. However there were a number of other components that seemed totally commoditized at first glance, and yet turned out to be difficult to source, so we had to replace them and re-design our circuit board to accommodate new components during the sourcing process.
How to fix this? Try to avoid any components that Digi-Key doesn’t keep ridiculously large inventories of. You probably won’t be doing your sourcing from Digi-Key, but their inventory levels tend to be a good measure of how common a part is, and therefore how widely available it is. Sometimes, though, sourcing constraints can be very different overseas. The best thing to do is to get your manufacturing partner to check stock with their preferred distributors; you’ll need to do this eventually, but your manufacturer probably won’t want to do it too many times, because it’s time-consuming.
A good way to get an early read on a component’s availability is to search for it on Taobao, China’s eBay-meets-Amazon-meets-DigiKey-meets-every-other-e-commerce-website. If the component seems to be available from a bunch of folks on Taobao, it’s probably easy to get in China; if it’s not, you might have some trouble. Google Chrome’s in-browser translation is your friend.
Supply chain partners tend to share your optimism
It can be difficult to convince a factory to work with you when you’re a start-up. Factories tend to make their money on volume, which start-ups generally can’t offer. So if a factory is willing to work with you, it’s probably because they’re excited about what you’re doing and they hope that you’ll grow quickly, and that they’ll get to come along for the ride.
It also probably means that they see you through the same rose-colored glasses that you see yourself. They probably think your product development is complete, and they probably don’t realize that you’ve never done this before.
Therefore when your partner gives you timeline estimates, they’re not factoring in engineering iterations, because they don’t realize they’ll be necessary. They may assume that you have more resources on the project than you do, and they may not realize how heavily you’ll be relying on them for support.
So while it might seem reasonable to take an estimate from your supply chain partner and pass it along to your backers with the message, “see, the experts say that it’ll only take 6 more weeks!”, there’s reason to be skeptical of even the experts.
The scope tends to broaden, especially with very successful projects
What happens when you set a $10,000 goal for your Kickstarter campaign — assuming that you could actually deliver the product for $10,000 — and you collect over a half million dollars?
You hire people, which speeds you up. And you do more, which slows you down.
Many project creators look at Kickstarter not as the goal, but as the beginning of a long journey. I hope that’s the case for us, and I hope our backers are supportive of that vision. Many of them talk about Spark as a bigger entity than just the Spark Core, which I appreciate, as that’s how I see it myself.
This does have an impact, though, on the short-term. As an example, we’re probably spending more time on security and on scalability of our software platform than might be necessary to make our backers happy because we want the platform to be suitable for not just hobbyists and Makers, but also entrepreneurs and OEMs.
And I don’t think this is uncommon. Many successful projects raise VC funding, implying that they’ve got big dreams. And pursuing those dreams inevitably bleeds into the process of delivering on the Kickstarter campaign.
Being transparent with your backers can be nerve-racking
This doesn’t really explain why projects get delayed, but it does explain why when delays happen, founders tend to communicate less, and backers start to get angry.
It’s scary to be expected to report out to thousands of people on a monthly basis. The causes of a delay can often be subtle, and you worry that backers won’t understand. Or, if problems arise, you hope that you can fix them before your next update, so you can share good news rather than bad news.
It’s a slippery slope, though. Everyone starts out super open and transparent. But the first time you miss an update, or when the delays and/or the uncertainty about your timeline become unbearable, the tone changes. At first it starts with a couple of rabble-rousers whose frustration is quelled by the supportive masses. But at some point, people start asking for refunds, and the grumblers begin to overwhelm the supporters. This is every project creator’s worst nightmare.
So what’s the solution here? Well, I don’t really know. But I’m trying an experiment with our campaign of absolute and unconditional transparency. We have a status page tracking our up-to-the-minute project status. We over-share in our updates (which makes them extremely long; sorry for any of our backers who have, you know, other things to do). We film videos explaining the engineering process in terms simple enough that we hope most of our backers understand them. We even write blog posts like this one (yes, I just linked to this post in this post. Call it meta-transparency).
Will it work? Will our diligence pay off? Will our backers still love us despite our delays? I sure hope so. And if they do, I hope other project creators will follow suit.
For most founders, the product comes first.
I asked our fantastic advisor, bunnie, for his input on this article, and he offered the following. He said it well enough that I have little to add. bunnie says:
There’s a saying, you can have schedule, price, or quality: pick only two. The two that Founders tend to pick – particularly on Kickstarter – are price/quality, over schedule. It’s a different ball-game when you deal with retail, where there is a real contractual monetary penalty if you deliver late. The contractual penalty means you can now value time, so you may go for higher priced solutions that deliver fast and good quality – as long as the solution cost is less than the contractual penalty for delivering late, it was “worth it”. Or you can cheap out and just ship whatever you get, regardless of the quality, and take the returns – but fundamentally, Kickstarters don’t penalize Founders for being late, so of the holy trinity, quality and price are the favorites. In short, there’s no reason for you not to be late, except pride, and pride is an easy meal to swallow when faced with poor quality or paying more as the other options.
Founders are like first-time parents. I don’t know if you’ve had children, but the first child always gets too much attention; the nursery is painted just the right color, the baby gets matching clothes, you take detailed notes and read all sorts of books and agonize over paper vs. cloth diapers, what formula or breast feed solution to use, and when the baby cries you rush over to comfort it.
The second baby inevitably gets the hand-me-downs of the first, and by the time you’re on the third or fourth you’ve got the process down and you know it’s okay to let the baby cry a bit, you know exactly what supplies you need, and you know when they’ll become more independent so they tend to become independent faster. You stop worrying that the nursery looks like a mess, and you don’t get torqued when the baby’s face is covered with slobber and snot. It’s not such a big deal when they go off to school for the first time, or start dating, driving, etc. – it just happens.
Products are very much similar to this – founders dote over their very first product, only to find later on that half the stuff they labored over the customers didn’t even notice or appreciate. You then have a more refined perspective of the ROI and can apply better judgment as to which features to kill or throttle back to preserve schedule, and you know what areas factories are good at doing on their own, and what areas you need to pay attention to. But generally, Founders tend to dote and fret; and there’s nothing you can do to stop them from doing that, except stand back and enjoy watching the rewarding process of new entrepreneurs bringing a new product into the world.
And that, after all, is why Kickstarter exists.