How not to screw up your manufacturing startup
Conventional startup wisdom is geared toward web apps. Here’s what we’ve learned that flies against whatever VC blogs are serving these days.
This post is meant to share the lessons I’ve learned in a manufacturing startup. It’s intended audience are people who are working on a manufacturing startup idea, are familiar with the tech startup scene, but don’t have a ton of experience in the manufacturing sector. And by “manufacturing startup idea”, I mean you’ll be operating in a warehouse or factory setting. This is not about having a formula or design for a copacker, 3d printer, or artisans in Italy — y’all got different issues.
I work as the CFO/COO of Edenworks, an indoor farming startup in Brooklyn. Given our job is to design and operate farms inside warehouses to produce salad greens as efficiently as possible, our issues are more manufacturing than farm related. A lot of the conventional startup wisdom out there doesn’t apply to us or other manufacturing startups. So given these lessons were all learned the hard way, I figured I’d share them to spare some sweat.
Process, not features, are your competitive advantage. This has been the biggest learning for me. Most startups focus on features to gain customers. Manufacturing companies largely don’t because features are easy to copy, and second movers often have the advantage if they can produce your feature for cheaper. Consequently, most manufacturers stay ahead in their industry by having the highest level of productivity, and that means focusing on process. This is what frustrates guys like Peter Thiel about manufacturing. He wanted flying cars, but instead got 40 years of relentless process innovation to drive down costs. In the auto industry, the undisputed leader of this drive has been Toyota. Toyotas largely have the same features as other cars, but they innovated like crazy on their production process. As a result, Toyota is currently worth $186 billion, 2.5x more than the second place finisher, Daimler.
Focus on the physics before software. The promise of the Internet of Things has duped folks into thinking that the traditional manufacturing process + software = orders of magnitude higher productivity. Software is helpful, but if you’re redesigning a system, software’s benefit is marginal — you get greater returns by changing the Thing than the Internet behind it. In indoor farming, we see a lot of competition focus on how data will drive yield increases, yet they haven’t figured out how to regulate air temperature in their facility. As a result, their “machine learning” algorithms aren’t really learning — they get garbage data in and spew garbage data out. At least in indoor farming, the physics of maintaining those environmental states, and doing so cheaply, has a higher ROI than software (at the least at the beginning). The right design with crap software thrown on top will be 10x more productive than the wrong design with great software. Some of the most efficient food production facilities in North America still report inventory and production metrics in text-only PDFs in Courier New. The returns on spending $100k for a new packaging machine outweigh $100k for an updated UI.
Nothing is new. The winning designs in manufacturing focus on simplicity, as complexity makes stuff expensive and risky. Most “innovations” in manufacturing are just taking simple designs and making them more complicated. As a result, your new conveyor belt design probably ain’t as good as the one designed 20 years ago. Moreover, solving the basic problems (how do we make X? How do we move X?) has been the focus of millions of firms for hundreds of years — chances are there’s an off-the-shelf solution to new problems. At Edenworks we’ve taken cues from the pharmaceutical, eyeglass lens, floriculture, and cannabis industries. We’ve looked at the major farm, greenhouse, and packaging incumbents in the US, Canada, Netherlands, and Japan. There’s no way any solution we could have designed from scratch would have beaten existing solutions which have been battle tested for decades.
The challenge for us then is how to break down the problem so that solutions from existing suppliers, potentially from other industries, can be implemented to solve our problems. This requires site visits, Google, and Youtube (in manufacturing, Youtube is our Stack Overflow…kind of). You’re our kind of manufacturing engineer if you dig absorbing all the different solutions that exist and learning how to configure and integrate them into a unified solution. In fact, there’s a whole field called manufacturing systems integration that’s dedicated to configuring off the shelf systems and components to solve custom problems. Integration doesn’t sound as fun as design, but given there’s no turning back from poor design that’s gone to market in manufacturing, it’s often the better bet.
If you do need to design custom equipment or inputs, understand that as soon as someone figures out how to do what you do with off-the-shelf solutions, that someone will have the cost advantage and beat you royally. A good, defensible manufacturing strategy is one where you’re applying and protecting (ideally via patent) a faster, cheaper, more reliable way of doing something in your industry by borrowing a proven approach from a parallel industry.
Do competitor teardowns. Paul Graham has said “A startup worrying about competitors is like a fat guy smoking a cigarette as he worries about contracting West Nile Virus”. Unless you already have deep industry experience, this is terrible advice for manufacturing startups. Doing competitor teardowns taught us:
- how much stuff actually costs — a surprisingly tough problem for new manufacturing startups
- which competitors were likely to succeed, and what the ones who weren’t needed to do in order to survive
- who the best suppliers were
- what productivity benchmarks are actually achievable — what’s reported publicly turned out to be waaaay off
To go to market without having known these things would have killed us. After doing the competitor teardown, we completely redid our design to take advantage of lower cost suppliers and reconfigured our business plan to fill the largest market gaps.
A proper teardown requires calling all of your competitors’ suppliers, asking for specs and price quotes, reading their patents, finding out their plant-level org chart, spending time understanding their growth strategy, and then making a ton of (hopefully informed) assumptions. These assumptions will change over time, so when it comes time to model these out in excel, it’s important to keep those assumptions visible and explicit. We have anywhere from a dozen to 50 line item references for each of our competitors.
Suppliers are your best consultants. Consultants spend a few weeks to a few months on a given project, make some phone calls, look at some data, and are henceforth experts worth $200+ an hour. A sales rep from a leading manufacturing supplier has spent 10+ years watching industry trends, has written specs for your competitors, and generally understands what works and what doesn’t. They also will share this information (which is usually not under confidentiality agreement) for free, given only a token amount of prodding. A good amount of our competitive edge at Edenworks comes from information gleaned from suppliers.
Do not move fast and break things. Move fast to minimize risk. Code is cheap to build and fix. Equipment is not. If you’re outfitting a factory for a major customer, you’re probably spending a few million bucks that you’re not getting back anytime soon. If you want to overhaul your factory design, your best bet is to build a new factory. We went through four complete design overhauls, each with multiple revisions, before going to market in a serious way. Had we gone to market or even raised considerable money with versions 1–3, we’d be toast. The faster you can minimize risk, the faster you can start building out for production.
On the operations front, never leave anything to ambiguity. If a supplier says the item is coming “next week”, ask when exactly next week and get the tracking code. If you’re getting inspected (we’ve been inspected by customers’ purchasing teams, two insurance companies, the USDA, and the New York State Department of Agriculture and Markets), get their inspection checklist before they arrive. If you think your customer is happy because you’ve received no complaints, visit them to make sure. All of these things take time away from getting product out the door, but they ensure you don’t get screwed. In manufacturing, you optimize for not getting screwed.