WW's greatest mistake was their first one, what is called taking the "Moonshot" approach.
WW tried to simultaneously go from engineering prototypes to full production on a whole slate of interrelated and interdependent products. These products can be viewed as a matrix of sub-projects, each of which has a hardware dimension and a software/firmware dimension.
WW also tried to go from a few hand-built prototypes to final manufacturable design and full production in one step: The first "big thing" WW did after the KS campaign ended was to set up a Chinese partner company in Wuxi to handle manufacturing.
All this was done with what was, in hindsight, a stupefying level of management ignorance. Management was clearly drinking its own Delusional Kool-Aid. The penalties were immediately evident: Chinese New Year shutdowns, very long cycles for each spin, slow decay of the defect rate (especially for the molds), ability to spin up only one device at a time, and so on.
What would have been a more sane approach?
First, there is no need to reduce manufacturing costs during the relatively small production run for the KS backers: The actual production costs for the KS units are a relatively small part of the total costs needed to develop and debug them (COGS (Cost Of Goods Sold) vs. NRE (Non-Recurring Expenses/Engineering), with or without amortization).
The alternate approach would have been to use local resources to the greatest extent possible until initial low-rate (pilot) production started. Ideally, all mold (and other tooling) development and production of cases & PCBs should have been handled in Austin. The turn-around time would have been days instead of months. There would have been no timezone barrier, no shipping barrier, no language barrier, no currency barrier, no trade barrier... You get the idea.
Let's say WW had done this, at least for the first several hundred KS backers. How then to spin up volume Chinese manufacturing?
First, the best way to deal with barriers is to move all facets of the problem to the other side of the barriers. How to do this? Send known-good Austin units (along with copies of the production test equipment) to China and say "Clone these products!". The definition of success couldn't be more clear: Either it is a perfect clone, or it isn't. They can tweak molds until the Chinese cases precisely match the Austin cases. They can spin PCBs until all production (and certification) tests pass. (BTW, I've seen brilliant Chinese production engineers tweak schematics, PCB layouts and BOMs to simplify production, increase production rates and reduce costs, all with no loss of quality, functionality or reliability, and often with a boost to all three.)
There are several pressures that may be applied to ensure prompt success at the Chinese end, given the presence of successful US-based pilot production.
Point out that low-rate US pilot production is already working, and every unit made in the US is one that won't be made in China.
Initially work with multiple Chinese companies in parallel, dropping them as they fail. Eventually only two will remain. Divide production between them based on their yields. This will take significant seed money, but it's pure risk reduction and schedule acceleration.
There are US production companies who already have close Chinese partners (or owners), where local manufacturing is first done on a small scale in the US, then moved intact to identical, but larger, facilities in China.
Only after successful production in China starts will there be a need to set up a Chinese partner company to permit further reductions in production costs (mainly taxes and red tape).
Clearly, Wuxi should have been the last production step, not the first. A step to be taken near the end of the KS backer pilot production runs, and after successful initial Chinese production runs, to transition to full commercial production.
What is the risk-benefit analysis?
In technology, the most valuable factor to optimize is "Time to Market", something WW has failed at totally, utterly, and abysmally. And they are failing harder and harder with each passing day. Estimating the financial impact of the delay is difficult (proving a negative), but it surely exists and must be substantial.
The second factor for a startup to optimize is "burn rate", the optimal deployment of limited initial financial assets over time ($/week) to achieve the earliest possible goals that will entice additional investment (and, hopefully, revenue) sooner rather than later.
Burn rate is second to time to market simply because it is important to get product "out there" before newer market entries make you obsolete. If you get beaten to market, you are then in the "also ran" slot, locked in a price-based race to the bottom for market share, rather than enjoying the prestige and profits of being the "first mover" in a market niche.
Yes, domestic pilot production will be far more costly than Chinese production, but the time savings are truly immense. I've seen other startups justify over a 10x COGS differential for precisely this reason. For IoT-scale products, the COGS differential is closer to 2x-5x.
Again, with 20-20 hindsight, the funds already spent by WW would certainly have yielded faster and better domestic production runs, and attracted investors (and experienced folks) to handle the transition to Chinese production.
But that's just the hardware. An accelerated hardware schedule would have demanded accelerated software/firmware schedules as well. And three years in, WW still lacks fully functional software.
How can/could the software development process be accelerated?
First, be quite up-front with KS backers about the realities of hardware vs. software:
- Tell KS backers that feature-complete firmware/software won't arrive until just before retails sales start.
- The initial device firmware will barely be smart enough to correctly and robustly handle updates.
- Only basic devices and simple rules will be supported by the earliest software versions.
- The initial phone apps will be idiots, able to only register devices.
Use a "release often" development approach where each incremental release fixes some bugs while adding or expanding at least one feature. Plan for hundreds of releases, preferably on a daily or twice-weekly schedule.
This would get final hardware and initial software to KS backers at the earliest possible moment. Backers could advance their position in the shipping queue if they are willing to be early beta testers (something every Maker and Hacker would gladly do), and delay their queue position if not.
This would seed WW's presence in the market, despite a ramped limited-functionality roll-out to "only" 1700 KS backers.
But, sadly, this is not what WW did. Hasn't done. Won't ever do.
No, WW management went for the Moonshot, but aren't even close to getting the whole rocket to the pad.
It's like one engine (out of 6) and half the capsule are at the pad, and management is scratching their heads wondering why those parts haven't managed to get themselves into orbit yet.
Yup, there ain't no rocket scientists in WW management...