Your Startup Won’t Scale
I call bullshit.
Five years ago, out of the ashes of a failed person-to-person recommendation sharing startup, my partner Ryan and I decided we’d pivot — and dramatically — to build Bean Box, a coffee e-commerce business. In our first month, GeekWire was kind enough to write about us (see here), and not a day goes by where I don’t cringe thinking about what the first commenter said: “This program is pretty tough to scale as is.” Ouch, right? As a serial entrepreneur, I’m used to this kind of smackdown, but it’s what you do — and learn — from there that’s valuable.
Show me a business that scales easily, and I’ll show you a future-proof, enterprise-class, AI-enabled blockchain marketing automation platform that unlocks the kingdom of heaven…. Um, yeah.
As a software engineer, I’ve spent almost a quarter century fielding comments about scalability, of all kinds. What I’ve learned, and what I hope you’ll take home, is that any mention of scale that’s not clarified with a ton of nuance is effectively an empty utterance, generally leveled with some malevolence in mind. In short, the lesson we’ve learned is that in a startup, scale is as scale does, and the proof is in the pudding.
Five years in, and we’ve shipped fresh coffee to over 100k customers; in 2018 we handled approximately 50 tons of the lovely stuff. Every week, our system plans out hundreds of skus (mostly coffee samplers, and about half algorithmically personalized), and then coordinates handling and product manufacture using up to 125 different kinds of coffees sourced from 30 roasters. Every week is a two-ton Dances with Beans, in which our software, our team, and our equipment all work together to ship thousands of products. And yes, scaling has been hard, as it should be.
Scale, it turns out, can — and should — grow with your business, providing you are able to look ahead to meet it, and are willing to fail to know when the next approach is needed. By way of example, let’s look at how we’ve grown the deceptively super simple (physical) task of taking fresh coffee beans and filling bags with them. We started with four people sitting at a small table, pouring beans into measuring cups placed on scales. Easy! Within a few weeks, that process took hours, which threatened a key strand of our product DNA: freshness.
Fast forward through a stepwise learning curve: 4 wall-mounted cereal dispensers with scales underneath brought us to 8x throughput (cost: a few hundred $); then, a single industrial weigh-and-fill machine added another 4x throughput (cost: thousands of $); add three more of those (cost: tens of thousands of $), and you’re really cooking.
But like many cases of scaling incrementally, we quickly hit linear discontinuity: our 2016 holiday peak volume would have required too many people and too much floor space to accommodate 12 of those machines. Enter automation. Today, we’re operating a fully automated packing line, which feeds, opens, fills and seals 44 bags every minute (cost: hundreds of thousands of dollars). Here’s a bit more about what we implemented, and why. From here, similar machines can be added as needed, with more headroom than we’d ever need to reach $100m+ in revenue.
But the work of scaling even this part of our business is far from done. Our key learning on physical scale, is that when you scale up one process, you immediately break the processes both upstream and downstream. And when those operations break, it’s a natural part of the scaling process.
Scaling to meet demand, in this simple example, was a matter of incrementally growing how we operated with the demand curve, and then looking ahead to where and how linear scaling would break, and then investing before it overwhelmed us. Sure, a venture-backed startup (not us), banking on scale up front, and flush with cash, might have jumped for the jugular up front and just automated that specific process from the beginning. But without up- and downstream scale that have grown in sync and to match, you’re basically sunk.
One of the reasons we love this business is because we face dozens of scaling problems like this, all the time, both in the digital and physical realms. Here are some axioms we’ve learned, in many cases painfully:
- Scaling up means things breaking, and that’s OK.
- You can only pre-build for scale a little, because the problem space changes as you scale.
- Fixing a bottleneck means creating other bottlenecks both up- and downstream
All businesses have multidimensional scaling problems: business model; human infrastructure; business process; vendor supply chain; digital infrastructure; etc. And there’s no “solution” to scale, only the process of stepwise evolution, in lock-step with demand ideally. So when the inevitable happens — someone waving away your startup under the rubric of scalability — remember that it’s your work to prove them wrong, every day, one step at a time.
