SnowShoe’s Journey to Mass Production

An in-depth look at our process as we start pre-orders for V2 of the SnowShoe Stamp.

A year ago this month, SnowShoe graduated from the Disney Accelerator. There was a lot to be excited about: we were closing a fantastic round of seed funding and we had lined up some amazing pilot customers.

There was also a big problem: our hardware was ugly, didn’t work very well, and was too expensive to manufacture.

We knew something had to change, so we endeavored to find a better way.

Along the way, we’ve learned basic mechanical engineering, racked up a crazy number of frequent flier miles, and formed some truly amazing partnerships (and friendships). Keep reading for a full run-down of where we have been, where we are today, and where we hope to go over the next three months.

Finding a Partner

This incredibly long and arduous journey started in February when we interviewed a large selection of engineering firms that specialize in helping startups manufacture hardware in Asia. After a lengthy selection process, we decided to work with ZaoTech, a well respected firm with offices in both the United States and Shenzhen, China.

The first major task we undertook with ZaoTech was to define the schedule of projects that would take us from our old 3D printed stamps (built in-house on our fleet of Makerbots) to a truly scalable process that could be efficiently outsourced to a contract manufacturer.

It seems trivial now, but, at the time, this felt like a very intimidating task — so many aspects of the project were unknown that it seemed impossible to lay out a roadmap that we would be able to follow. In the end, ZaoTech’s experience paid off, as the initial project timeline they laid out has been extremely (and surprisingly) accurate.

Step 1 — Prototype Parts Fabrication

Estimate: 2–3 weeks; Completed on-time in Mid May

There were some fundamental science questions that needed to be answered before we could dive head-first into designing a better manufacturing process. As we outlined in our Hardware is Hard blog post, a good number of our 3D-printed stamps were failing to function properly on Android devices.

We needed to figure out why this was the case, and, because one of our suspected culprits was the 3D printing process itself, we needed to make some test hardware that wasn’t made on a 3D-printer.

Through several weeks of discussions, we were able to specify the exact prototype hardware we needed to manufacture and ZaoTech was able to have one of their partners in Shenzhen fabricate the hardware for us.

Part of the SnowShoe mobile device test suite we used to develop our stamp specifications.

Once the prototype parts arrived at our office, it took us about two weeks to run a body of experiments that had us testing the prototypes on most smartphone handsets currently for sale in the US. These experiments resulted in a precise set of physical specifications for a stamp works on each and every device we tested.

Now that we had physical specifications for the stamps we were trying to manufacture, it was time to figure out how we were actually going to make them.

Step 2 — Stamp Design and Assembly Breakup

Estimate: 2–4 weeks; Completed on-time in Early/Mid June

ZaoTech applied their considerable expertise in injection mold tooling design to come up with a set of three different manufacturing processes that we could potentially use to fabricate SnowShoe stamps at scale. Each process was ranked based on inherent risk factors, cost of tooling, and material efficiency.

We thought we were going to be forced to choose one of the three proposed processes until ZaoTech designed a set of injection mold tools that would allow us to test the most important aspects of each of the three assembly processes.

After internal review and a small amount of internal prototyping, ZaoTech assembled a Request for Quotation (RFQ) package that we circulated amongst interested contract manufacturers.

Step 3 — Supplier Sourcing and initial quotations

Estimate: 2–3 weeks; Completed on-time in late June/early July

As startup founders that had never manufactured hardware in Asia before, finding interested manufacturers was an incredibly intimidating proposition. We knew nobody, and we had no clue how we were going to start finding the right people to work with.

Luckily, we had two networks we could lean on: our investors (especially TechStars), and ZaoTech. Several other TechStars hardware companies were able to introduce us to their contract manufacturers, and ZaoTech has several factories they routinely work with in China.

With these introductions in-hand, it was time to board a plane to Hong Kong.

Our ride for the long trip to Hong Kong

We flew into Hong Kong and took a ferry into mainland China. This is another aspect of our manufacturing process that would have simply been impossible without ZaoTech’s services. Wong Ze Shen, the lead mechanical engineer on our project, served as our interpreter, tour guide and travel agent for all aspects of the trip.

We ended up having face-to-face meetings with four contract manufacturers, and of those four we toured three of their facilities.

After returning from China, ZaoTech circulated the RFQ to the four manufacturers we had met. They each took about two weeks to produce an initial quotation.

All of the initial quotations were significantly higher than what we had been hoping for, a fact that sparked no small amount of panic in the SnowShoe office. ZaoTech patiently explained this was all part of the process and the quotations would come down to our target range as ZaoTech iterated on the manufacturers’ proposals.

ZaoTech was right. Within ten days, and after three or four back-and-forth iterations with the contract manufacturers, we had multiple quotes that were right in-line with our target price estimates. We even had a quote that beat our most aggressive price targets for this first iteration of tooling!

Step 4 — Mold and Tooling Design and Build

Estimate: 5–7 weeks (1–2 weeks design + 4–5 weeks build); Completed on-time in late August

We chose to work with the lowest bidder and ZaoTech communicated our choice to each of the parties. We then issued a purchase order for tooling to our new contract manufacturer!

It took the manufacturer a solid 4 weeks to iterate further on the tooling design (in constant communication with ZaoTech) and to finally start cutting the tools.

We received weekly updates from ZaoTech during this time but there was not a lot we could do. Both the manufacturer and ZaoTech were pretty heads-down in getting their work done.

Step 5 — Engineering Validation Tests (EVT)

Estimate: 3–4 weeks; Not yet completed, but very close!

There is a very specific suite of tests that you run to bring a manufacturing line into production. These validations tests are designed to act as a set of checks to make sure you don’t end up making a huge amount of hardware that doesn’t work, doesn’t look right, or costs too much to sell at a profit.

As soon as the tooling was finished, we boarded another long flight to Hong Kong so we could participate directly in the first of these three major exercises: the Engineering Validation Tests (EVT).

For SnowShoe Stamps, that means comprehensively testing 250 units on over 20 different mobile devices in our test suite. In total we completed over 5,000 individual tests over a three week period, and that is not accounting for any tests we may have repeated.

In the end, EVT hardware should work, though it might be ugly and is probably not something you can sell to a client.

Sheradyn, SnowShoe’s Head of Hardware Operations, conducting EVT testing in China

If everything works in EVT, you can graduate to the next major set of tests. Unfortunately we didn’t make it through EVT on our first pass.

While running our EVT stamps through all of the devices in our test suite, we identified a major problem: the stamps were only working intermittently on iPads! This was both shocking (we had never before observed problems with iOS devices) and frustrating.

We launched directly into a week of in-depth root-cause analysis and identified several factors we thought were likely contributing to the iPad failures.

The bad news was most of the errors were things that would require us to change the injection mold tools we had just cut. Making those changes would cost us another 3 weeks but we had no choice.

The good news is the changes resulted in stamps that work perfectly on nearly every device we have tested. In the end, the delay was painful, but definitely worth it.

Current Status: We are currently in the final stages of our second round of EVT testing and everything is coming up aces! We expect this testing phase to conclude early next week, at which point we will immediately begin manufacturing stamps to support DVT, the second of our three major phases of validation testing.

Step 6 — Design Validation Tests (DVT)

Estimate: 3–4 weeks; Scheduled for Completion in Mid November

Once we’re sure our tooling can be used to make a limited number of stamps and those stamps are working as intended, it is time to focus on aesthetics and repeatability. The Design Validation Tests (DVT)are designed to do just that.

For DVT, we will be manufacturing approximately 2000 stamps. The goal is to make sure all 2000 stamps work as expected. During these tests, we will also make sure each of these stamps meets the criteria we have laid out for cosmetic finish, durability, material toxicity, and several other environmental tests. The test will also help set a baseline for our expected loss rate when manufacturing thousands of stamps.

The good news is that all stamps that pass our DVT criteria are stamps that we can actually sell to customers! This means that we may be shipping our first V2 stamps by late November, if everything goes according to plan!

Step 7— Process Validation Tests (PVT)

Estimate: 4–6 weeks; Scheduled for Completion in Mid December

Finally, after we have proven our tooling can make stamps that work (EVT) and that the tooling can also make lots of stamps that all work and look the same (DVT), the last hurdle we have to jump over is to figure out how efficiently we can make stamps at large scale.

This is where the Process Validation Test (PVT) phase comes in. In PVT, we study each and every process a stamp goes through on the manufacturing line, and we optimize those steps for speed and cost. This could mean building a robot to handle an assembly step that was done by a human in EVT and DVT, or it could mean changing the layout of the manufacturing line to reduce bottlenecks or improve ergonomics for the workers.

Our current plan is to do an initial PVT run of 5000 stamps in mid December. If all goes well with those stamps, we will likely shut down the line for one final set of tooling improvements (mostly cleaning and resurfacing) before ramping up to full mass-production (50–100K stamps/month) in early Q1.

Figuring out how to manufacture SnowShoe Stamps at scale in China has been a massive challenge. We are extremely excited that we have finally solved enough of the inherent problems that we can start taking pre-orders for these new stamps.

We are extremely grateful for everyone who has helped us along the way, and we are genuinely excited to see all the amazing things our customers build with this new hardware!

SnowShoe Stamp is a startup located in San Francisco, California. We make plastic stamps that unlock the digital world through a simple touch on your mobile device screen. Click here to learn more about how our stamps work.