How Fellow bootstrapped for 8 years to build a coffee empire

And how it had to go back to the drawing board on a $10 million product

A decade ago, a little coffee company launched a Kickstarter campaign for a product called Duo, a coffee maker that’s the best of both worlds: a cross between a French press and a pour-over.

The campaign sold almost $200,000 worth of machines to 2,700 backers, but the product itself was a flop. The company was Fellow, which today makes some of the best pour-over coffee grinders on the market; much-copied kettles with hyperprecise temperature control, loved by pour-over baristas across the industry; and a bunch of other caffeine-related paraphernalia.

Most recently, the company launched the Opus, one of the very few sub-$200 coffee grinders that can grind the beans in a variety of ways, from fine espresso to grinds suitable for electric coffee makers, French presses, AeroPresses, pour-overs and the like.

But the machine almost didn’t see the light of day: A year into production, the team had to scrap the entire idea and start over. I sat down with Fellow CEO Jake Miller to learn about the highs and lows of manufacturing hardware that keeps us well caffeinated.

“Looking back, not raising institutional money early on was absolutely the right call. We only exist today because of that choice.” Jake Miller, CEO, Fellow

“What’s true for most entrepreneurs is that there’s just something in them that has to come out,” Miller said, who started his entrepreneurial streak by bootlegging CDs in high school, creating and selling novelty T-shirts in college, and then starting a roofing and siding company after he graduated. “I didn’t want to sell inappropriate T-shirts in a side hustle anymore, but I still loved the thrill of entrepreneurship.”

He fell in love with coffee during a stint at Caribou Coffee. From there, he earned an MBA at Stanford and decided to start building Fellow. He attempted to raise money, but after receiving 73 rejections from VCs, he gave up. Not raising a serious chunk of cash in the early days turned out to be a company-saving blessing in disguise: Fellow raised its first “real” round of institutional money last year in a $30 million round.

“I graduated from Stanford, and I had this idea for Duo. Bright-eyed and bushy-tailed, I said, ‘Let’s go,’ and we launched it on Kickstarter. It was just me at the time, and when we raised the $200,000, I thought I was rich,” Miller said. “It took two years and more than $300,000 just to deliver the Kicsktarter units. It was this incredible fall from glory and me realizing how hard product design and development was. That product flopped and doesn’t exist today. I learned a lot from it.”

The Duo was a mashup between a pour-over and a French press, with the convenience of the latter and the quality of the former. It made a great cup of coffee, but it was complicated to use, hard to clean and unreliable.

At the Fellow office, a sneaky preview of the company’s upcoming scale that makes pour-over coffee fool proof. Image Credits: Haje Kamps / TechCrunch

Fellow had to go back to the drawing board after its sub-$200 grinder was a year in development. Thus began an incredible journey in which bootstrapping and raising small angel checks turned out to be a blessing, though there were a lot of war stories along the way.

“After 73 rejections, I finally met with an investor. His name was Jerry Mix. I remember taking the train down to Redwood City, and we met at a Starbucks. He said, ‘I don’t love the first product, but I do love the vision. Let’s do this,’ and wrote me a check for $250,000, basically that day,” Miller said. “Looking back, I am so thankful that I failed to raise a meaningful amount of money. The reality is that if I could have raised 2, 3 or 4 million dollars, I would have taken this massive swing at this product that no one cared about. I would have spent $2 million on development and launched some $1,000 do-it-all coffee contraption. And there’s a pretty big graveyard of products like that.”

Fellow found benefits in being strapped for cash and having to be really deliberate about the categories it wanted to play in, the cadence of product launches and the overall focus of the company.

After Duo, the company launched a stovetop kettle called Stagg, which included a thermometer built into the lid. From there, it built an electric kettle called the Stagg EKG, which moved the market significantly, as one of the smartest, most temperature-controllable electric kettles around.

“The first electric product was the Stagg EKG. We’d never done electrics before, and I’m not an engineer, so we didn’t know how to figure it out, but once again we went to Kickstarter,” Miller said. “It was well received, and we sold about half a million worth of kettles. We were able to fund the development using those funds.”

Observing the customers

“If you watch people who really care about coffee, you realize they care deeply about temperature. We found ourselves looking for the hacks. People were taking their stovetop kettles and drilling holes in the lids to add thermometers. We were like, ‘Cool, but you don’t need a hack; we just design that as a feature,” Miller said about the development of the original stovetop kettle.

The desire to build a super-precise electric kettle came from observing the customers using the existing kettle. For the electric version, the company took it a step further, with automatic, super-precise temperature control.

“Basically all kettles are just mechanical hot plates. It has a mechanical relay that turns on to heat the water,” Miller said. The problem with relays is that they only have a certain number of cycles before they wear out, so the company added a solid state relay to the kettle, called a Triac. “Our kettle uses a mechanical relay to get really close to our set point, within about 4 degrees Fahrenheit. It then switches off the mechanical relay and switches to the Triac: That means you can’t hear the clicking, and you can use pulse-width modulation to accurately control the heating element.”

All of that is nerd-speak for, “Get the water as hot as you can as fast as you can, then essentially turn the heating element on and off very quickly for accurate temperature control.”

The downside? The electronics were expensive.

“There’s a really cool PID in there as well,” Miller said, referring to the control loop that keeps tabs on the temperature inside the kettle and controls the Triac. “That was totally overkill on those components, and that added an additional $7 to the bill of materials.”

For context: A $7 increase on the bill of materials typically results in a $20 to $25 increase in suggested retail price. That’s a lot in a world where you can pick up an electric kettle for under $20. “But what really matters is what the customer cares about. They don’t care about pulse-width modulation,” Miller said. “They care that if you set the kettle to 199º F, it will hit 199º F and stay there while you make your cup of coffee.”

Keeping the ritual

Making coffee is a ritual to a lot of people, especially those who are passionate about it. The trick is to make the coffee-making easier, without losing the sense that you are making your own coffee.

“We’ve been very deliberate through our design process and product portfolio to not remove the ritual from the process,” Miller said. “We want to make it easier, not completely take it away from you. If you want convenience, there’s Keurig and other options. What we designed is for the person who wants that five-minute pause in their day to say, ‘I’m going to create this. I care about this.’ and I believe you’ll see that in what we’ve designed. We are working to find and keep that balance.”

Of course, once the company added a plug to its products, the complexity skyrocketed. “I was spending Christmas and New Year’s in the factory trying to get these Kickstarter units out of the door, and I remember just sitting there, by myself, thinking, ‘What the hell are you doing? My family’s having Christmas dinner back home,'” Miller said. “But that’s what’s required to do hardware. You can’t send off a file and hope that great hardware shows up a year later.”

For the first eight years, Fellow says it did three or four small angel rounds, and Miller remembers that the VC industry was laughing at him and his little company. Two years ago, though, that changed, and the company found itself on the cusp of a big opportunity.

“The only way we could make that opportunity come to pass was to close a large round of funding, so we brought in $30 million for that. That was a huge unlock for us, because it enabled us to hire. We’ve hired 20 people in our product team — all ex-Apple, Tesla, Google X — they bring these specialized skills so we can do all the kinds of product development things we couldn’t do before when we were bootstrapped and angel-funded,” Miller said. “We also had to start talking about board management for the first time, we had to have regular board meetings, and there’s real pressure to grow. I can get fired now!”

The pressure is one thing, but Miller said that the accountability that comes with real money and a board is, in fact, a blessing.

Dodging a VC-shaped bullet

“Looking back, not raising institutional money early on was absolutely the right call. We only exist today because of that choice,” Miller said, though he appreciates the things the additional money unlocks. “We have a team of 100 now, which means I have 99 people who do a really great job of all the fun and cool things. There’s a part of me that misses doing that, but at the same time, I’m not sitting by myself in a factory at Christmas trying to figure out a PID loop.”

The company wants to be the global leader in coffee and espresso products. But at the same time, there are only so many coffee products you can make: Eventually, the market becomes saturated.

“It may take us like four, five, six years, but we’ll reach a point where we’ve designed everything there is to design within coffee and espresso. And then we’ll have a really fun choice to make as a company to say, what do we do next?”

The company isn’t that keen to go too far down market, however. “We care about the value we provide to the customers, and if you want to hit a lower price point, you’re stripping out a lot of the cool stuff that people love and expect from Fellow. If I were a betting man, I’d guess we’d be moving to other categories instead,” he said.

Fellow identified early on that brand reputation is valuable and important: The products are thoughtfully designed, functional and can often be repaired if something goes wrong with them. If the first interaction with the Fellow brand is a kettle like any other $20 kettle you can buy, you lose a lot of the magic.

“Right now, we essentially design tools for the home coffee lover or home barista. These are beautifully functional tools that perform well and look great. That same positioning, I think, works with the home chef, the home mixologist and many others as well,” Miller noted. “But we have to stay true to who we are as Fellow and figure out what those next products are. That’s probably five years out, so I’m not going to worry about it too much yet.”

User-centered design

“Are you making a coffee maker?” That question was met with a swift “no comment” and a cheeky smile. But I’m still curious … if the company were to make a coffee maker, how would it approach it?

“With any product, the question is if you can you find a reason for it to exist. To come to that answer for any product, there’s this idea of really empathizing with the customer. I came out of the design school at Stanford, a big believer in human-centered design. Emphathize, define, ideate, prototype,” Miller said. “A lot of what we do is just observe. What products are customers used to? What are their pain points? What do they love? What do they hate? The next step for us is the ideation round. How do we make things better?”

One of the beautiful things about having an office in central San Francisco is that you can invite your local customers to make coffee and just look and ask questions. COVID-19 made the observation step a lot harder.

“There was one point where I spray-painted a kettle white in our retail store. I put it out on the showroom floor and asked the staff to count how many times someone tried to buy the white kettle. Six months later, we launched a white kettle.”

A 12-month setback

The company just launched the Opus grinder earlier this year, with the goal of building a sub-$200 grinder. Fellow eventually realized it had made a fundamental mistake in the way the grinder was designed. After a year, Fellow had to start from scratch.

In its office, Fellow has a test rig that can dispense 20 grams of beans, run the grinder, and then do it all again a few moments later. This enables the company to test the longevity of its products. Image Credits: Haje Kamps / TechCrunch

“Three years ago, the belief was that you couldn’t get a grinder at a sub-$200 price point that can do both espresso and pour-over. We wanted to design an all-purpose grinder. To do that, you need a grinder that has a ton of torque; to achieve that torque, you just over-spec the motor,” Miller said. “But an over-specced motor is expensive, and that would push the price to above $200. So we had to design a gear box. We couldn’t find anything that worked, so we went down a path to design our own. It was a lot more complicated than we thought: We got something that worked, but we threw a lot of testing at it, and realized that this thing would work for three to six months, and then wear out.”

The gearbox was made of plastic, and it turns out that the friction and pressure of the high-torque application broke or melted the gears of the worm gear design the company had chosen. The challenge with worm gears is that they slide against each other, rather than rolling and interlocking. That means you can get really high torque, but it also means that the gear box has to be well lubricated.

The company tried plastic on plastic. Plastic on metal. Metal on metal. After a year of experiments, Fellow wasn’t any closer to a solid solution.

“All our products come with a three-year warranty. We want to keep them out of the landfill. It was a really awful day where we realized that what we were working on for a year simply wouldn’t work,” Miller said. The $200 grinders failing in less than a year would have put a serious dent in the company’s reputation. “We made a tweak, and the grinder would fail after three months. We made another tweak, and it failed after four months. Another tweak … but we never got there.”

The company took apart competitors’ grinders and found a few that were using similar designs.

“There were certainly other grinders that were using worm gears, but they just weren’t creating the torque that was needed for high-quality espresso. In at least one case, the brand just didn’t care about the liability: They must have known that this thing was going to fail, and I suspect they were crossing their fingers that it did so after the one-year warranty,” Miller said. “There are some really good higher-price-point grinders that do a pretty good job of everything. But a lot of that comes down to design geometry: How do you go from a coffee bean to coffee grounds?”

The solution, in the end, was to use a more conventional gear-box design that uses a 42-to-1 reduction. A high-speed motor spins at 15,000 rpm or so and is geared down to the burrs spinning at around 350 rpm. The higher-speed motors are much cheaper because they don’t have to be very precise or have very high torque.

On the left: The worm gear that didn’t make the cut. On the right: the gear box that made it into the final product. Image Credits: Fellow

“In the end, it was a cost choice,” Miller said, revealing that the new design, too, uses plastic gears. “The question was if we could make plastic work and make it reliable for more than three years of use. The testing said yes, so we pushed ahead.”

It seems as if it was worth doing properly: The Opus grinder has only been on the market for a relatively short time, but it’s trending toward selling $10 million worth in the first year. That’s a big deal, especially if the company is confident that it isn’t going to get half of them back with broken gearboxes after six months.

Leaning into company values

The anecdote of going back to the drawing board says a few things about Fellow as a company that’s reflected in its broader product line. The Ode grinder was recently released as a second-gen grinder, with a $350 price tag. But ultimately, there were only four major evolutions from the previous generation. Unusually, all but one of the improvements are backward compatible, meaning that if you have a first-gen grinder and it breaks, you can upgrade it with the new parts, getting the benefits of the redesigned grinder.

“One of our principles is to never stop improving, and everything is always a prototype. There were four pain points with the original Ode grinder, and we were able to fix them all. We listened to the customers and spent two and a half years making it better,” Miller said. “We made the minimum grind size smaller, we made the hopper steeper so beans wouldn’t get stuck. We made 23 new variations on the burrs until we found one that was perfect. And we added an ionizer to reduce the mess.”

The ionizer helps reduce static electricity, which means that the coffee dust doesn’t stick to stuff as easily, and that’s the only thing that isn’t backward compatible. Everything else — the burrs, the hopper and the minimum grind size adjustments — can all be installed on the original grinder.

“There’s no reason to throw an old grinder away. I think we are one of the only small-appliance brands that has a replacement parts section on our accessories page. If something breaks, you buy that part, and you can replace it,” Miller said. “We believe that people are willing to pay a premium for things you can fall in love with. That’s what we want to do.”