Local Manufacturing – Why it Makes Sense

I recently attended a Make summit where one of the most discussed topics was the challenge of going from zero to less than five thousand units of production. Many ODMs (Original Design Manufacturers) prefer startups who can commit to larger production volumes (thousands of units minimum, preferably more in both quantity and frequency), but what do you do as a startup that is still building demand or who has a product that will be a consistent but lower volume seller? There are ODMs, partners, and options for this all-to-common scenario, but the summit attendees all agreed that these are under-documented and/or under-developed. I’ve offered to unite partners, thought leaders, and entrepreneurs in the “Zero to 5000” challenge during 2014 to help document and improve solutions for hardware startups as they ramp up production of their products. Feel free to reach out to me if you’d like more information on this effort.

and congrats to the Revolve team for shipping Kubi. It’s been a game changer for my frequent videoconferencing!


This is a guest post from Ilya Polyakov, co-founder of Revolve Robotics, makers of Kubi.

Having come out of the bottomless, dark cave that is Design For Manufacturing (DFM) we are finally going through the exciting process of production ramp on the Kubi. The part that makes this even more exciting for us and should for hardware in general is that we are doing this right here in our office at Lemnos Labs. That’s right, not Shenzhen, not Guadalajara but here in the South of Market neighborhood of San Francisco.  This decision was not made lightly and time will tell how this ultimately works. Revolve Robotics is a lean mean innovation machine, with an emphasis on lean. Typically, mass production of hardware is about as un-lean as it gets! There are two main reasons we decided to stay here and handle our production in-house.

assemblyThe first reason is scale. When we were putting together the Kubi development plan, we certainly talked to several Chinese contract manufacturers. Our product is well suited for the exact type of high volume manufacturing typically done in China.  One issue kept hitting us over and over – a big line item on any quote was the Minimum Order Quantity (MOQ). This is the magic word every manufacturer wants to hear as it guarantees revenue and drives their entire business model. Large MOQs along with a prior product reputation can even warrant the manufacturer paying for the tooling and other development expenses up front. Depending on the product, MOQ can be as low as 1K or over 10K units. A typical MOQ we came across was 5K units, with us paying for tooling out of pocket. This leads to a pretty dicey go-to-market process.

Classically, a hardware company with a consumer product puts down enough cash to fund development through that first production run. They show prototypes or better yet first article units to retailers and distributors who hopefully order everything that will be produced. The retailer will usually want to hit a hard delivery date like Back to School or Christmas (meeting hard deadlines with an overseas production ramp is a whole other set of pleasures I will get into another time). Because its a new product, the retailer typically takes the stock on consignment – ie you own the inventory on their shelf until it sells, if it doesn’t, you get your product back, sometimes with penalty fees that amount to having rented that shelf space – its all on you!  So, essentially you roll the dice, hoping the thousands of units you already paid to develop and build get to the store shelves on time and sell, only then do you see any returns on your investment. Oh, and you cannot rely on the store to market your product, therefore you also have to invest in marketing to move those units off the shelf . And that’s assuming you know how to market your product. One model that breaks this cycle to some extent, and not always, is of course crowdfunding and pre-sales. With crowdfunding you have to set your goals to be reasonable for the campaign (a science of its own) but also so that the number of units pre-sold reaches the China MOQ threshold. Most of the time the reasonable goal for the campaign is lower than the MOQ, that goal has to be multiplied several times to reach that magic number. Many companies that reach their goal end up having to pre-sell additional units or take a risk and buy them outright with the hope they will be able to sell them later.

partsOur presales amounted to several hundred units, not the thousands we need to turn on a contract manufacturer. We sought a path to go to market in a controllable, civilized matter.  We changed our marketing strategy to pursue enterprise applications where customers recognize the value of the product and are willing to pay a premium. The new higher price point allowed us to produce almost all our components in the US and handle assembly in house. We are able to commit to much smaller runs and can essentially build units on demand. We are able to hone our marketing message by talking to our customers and respond as opposed to dealing with a flood of Amazon reviews from an un educated customer. We are also able talk to future retailers with a working product in hand and with customer testimonials.

We invested in soft tooling at a US based injection molder that can produce tens of thousands of parts. We have the production and quality control process developed and most importantly owned by us. The supply chain is controlled by us. Currently we are producing about 25 units per day and as soon as that needs to increase past 1000 units per month we can outsource assembly to any number of local assembly houses – remember, we own the supply chain and the process. We also own the design and can bring it overseas when ready, still producing while that ramp happens.


The second big reason we went with local production is speed. In the lean startup model, the emphasis is on releasing early, putting in the minimal amount of effort to test a minimum viable product concept and iterate through updates. This works well with software where once its designed one simply needs to upload it to the app store. Hardware needs to be manufactured and going from design to production can be a long and painful process. An optimistic production cycle in China is around 6 months, this is once the product design is fully locked in and nothing extraordinary happens (I once had a product get delayed for weeks because of the tsunami in Japan!). That is just too long for us. Our molder in the mid west was able to turn around tooling and get us first article parts in 3 weeks. We had our first unit assembled and working a week later. All in all, it took us 3 months from tooling kick off to shipping first units to customers. One of the simplest pleasures of not working across 12 time zones is the turnaround in decisions. You cannot have the whole team on the ground in China so all decisions have to go over email or phone calls where one of the parties is up at 2am.  It usually takes 24hrs to resolve an issue whereas I could be on the phone with our molder during normal waking hours and solve a problem in 5 minutes.  Quality control is simpler since all units get tested by us. Iterating process and hardware changes was instant. For example, having the firmware team a few tables over from the assembly area made debugging our QC firmware a relative breeze.  Our production jigs went though 3 iterations in one week thanks to the laser cutter at TechShop just a few blocks away.