Startups are science
On marketing in conditions of extreme uncertainty.
I often tell people that in the depths of startup chaos, I discovered a love for science.
Which was quite surprising for me. I was an English major, liberal-arts type, who loves conceptual discussions and always found measurement of any kind to be a bit distasteful.
But after further entrenching my literary mentality with a graduate degree, I inadvertently stumbled my way into a friend’s startup as the first business hire. It was chaotic. Now I know that this is kind of the point, but at the time, it was overwhelming and deeply unsettling. I’ll never forget the CEO sitting me down a couple of weeks in, to tell me: “Look, JY. You’re clearly a sharp guy. You ask lots of smart questions. But… we do not know the answers. We are figuring this out, too. We need you to figure out the marketing stuff. If you can’t do that, it’s not going to work out.”
He wasn’t bluffing or exaggerating for dramatic effect. He was a recent college graduate and the average age of our team was 23. So I needed to start figuring things out. We tried lots of things. Most of them didn’t work. Some of them did, so we did more of those.
But it wasn’t until 3 years later, that things actually clicked. A particularly smart friend (who was also struggling to launch a company) told me that I should really read Eric Ries’ book, The Lean Startup. I was skeptical because, by this point, I’d discovered the hard truth that not everyone who manages to convince an Inc. or Entrepreneur Magazine editor to let them write a guest column actually knows much about how to build or grow a company. So, another successful founder promising that their approach would fix everything? I’d heard it before.
I was wrong.
Reading that book felt like getting the cheat codes. After years of trying to figure things out on my own, this was the systemic approach I’d always assumed was out there… but after years of not finding it, I’d given up.
That book showed me what a startup is and how it should actually work.
Wait, what’s a startup?
What’s the goal of a startup?
Most people would probably say… growth? Get really big and make a lot of impact/money?
Most people (I will humbly submit) are wrong.
This has gotten confusing because the concept of “startups” has become a cultural force and multi-billion-dollar organizations with thousands of employees still use the term to describe themselves.
But a real, true startup is a specific thing. My go-to definition comes from Ries: “A startup is a human institution designed to create a new product or service under conditions of extreme uncertainty.”
That last part, the “extreme uncertainty” bit, is the key. And this is exactly what I’d been experiencing.
When you’re trying to launch a new product to a new audience, you’re trying to assemble a puzzle, except you don’t really know what the completed picture will look like. And you don’t know what shape the pieces are. Or if you even have the right pieces. And you’re usually trying to do it as fast as possible. While other people are maybe trying to steal your pieces.
Stressful, right?
This is why the whole central point of Lean Startup is that the single primary goal for a startup is quantified learning. Before you can do anything else, you have to crack the code. You have to put the pieces together.
So, no, a startup is not initially about growth. A startup is about learning.
Fortunately… there’s a tried-and-tested system that mankind has developed for trying to learn stuff. It’s called science.
Hurray, science!
No, I was not much of a scientist. I did get 2nd place in the regional science fair in 7th grade, testing the comparative speeds of different shapes and materials of parachutes for GI Joe action figures. But somewhere after that, the complexity of the subject and one particularly bad year with a negligent pre-calc teacher really disrupted my interest in the quantitative realm.
So, turning back to science in the world of startups was an act of desperation. We (still) did NOT know so many things about our business, and the whole methodology described by Ries and others in the (at the time) emerging field of “Growth” (with a capital G) showed that there was a structured way to approach this.
Fittingly enough, this actually brought everything back to the pre-7th-grade era science context (my sweet spot)…
The Scientific Method! I’m sure you know the concept, but do you remember the steps? Here, let me refresh your memory:
Maybe this all feels very obvious to you now, but I’d encourage you to think about how your team does marketing, especially for new products or experimental (hint, hint - it’s right there in the name!) initiatives. Do you… actually write down your hypotheses? Design an experiment? Do any kind of analysis or review, after a program is complete?
Yeah, me, neither. That’s why this was so powerful. By realizing that this timeless and very Lindy-validated approach could help put together our startup puzzle, I could feel everything click into place.
That month, I made a slide deck and convinced the rest of the leadership team to let me present at our next all-hands, to explain our new approach. I wish I could find the slides, but the big headline was:
MARKETING = SCIENCE.
We got way more structured in how we approached validating the right customers, finding the right channels, and allocating budget to drive the most impact.
Although that company didn’t have the glorious acquisition or exit that we’d hoped it would, this approach extended our runway, allowed us to raise more money, and helped a lot of our users to be healthier and happier, along the way. And this is the approach I’ve brought to every single company I’ve worked with since, from other just-getting-started ventures to multi-billion-dollar organizations that, well, yeah, I’m not sure if I’d really even call a startup anymore.
KNOW BEFORE YOU GROW.
So, what does all of this mean for you?
Well, for one, yeah, you should go ahead and read Lean Startup. It’s incredible. Even the highly divisive “Minimum Viable Product” approach makes way more sense, if you actually get the full context (it’s a mechanism designed to specifically accelerate learning).
But beyond that, my main point is that it’s important to know what phase you’re in.
Every company wants to be at the rocket-ship stage where the business is printing money and hockey-sticking all your metrics, up-and-to-the-right. But none of that stuff works if you don’t have the foundation right. The real startup phase is all about figuring out the basics.
As Elena Verna has said (as her First Law of Growth, in her very first newsletter!): You should not be focused on Growth, before you have Product-Market Fit! That’s trying to scale something before it’s stable. Great Growth teams use tons of data to optimize the various elements of a company’s growth engine, but you barely even know what to measure, when you’re first getting started.
So, to use a phrase that I’m trying to make happen: You have to know, before you grow.
(Or, is “You have to learn, before you earn” better? Tell me what you think.)
Either way: Make sure you’ve got a healthy fire going, before you try to pour gas on it. Artificial growth accelerators (like paying for a ton of traffic) can actually do more harm than good in the early stages, because they prevent you from seeing what is actually working and what needs to be fixed.
In the Growth and Scale-up phase, you’re optimizing all kinds of things. The Balfour Fits point out that you need to assemble (at least) four separate components to get to the $100M ARR range: Market, Product, Channel, Model.
But in the true startup phase, Ries says you’re really just testing two core hypotheses:
The Value Hypothesis: Do people actually want this, and will they pay for it?
The Growth Hypothesis: Is there a repeatable method for introducing the product to those people?
If you’re in charge of marketing, #2 is your job. But that’s the thing—every company has to figure out both of these, but the second cannot work without the first. So, even if your job title says “Marketing” or “Growth,” you better make sure that the value hypothesis has been confirmed, before you start spreading the word.
This is why Product Marketing is usually the most important component for an early-stage team—it’s the product/market crossover role. In more ways than one. While the Value hypothesis is generally something that the Product team will own and improve, a product marketer can help develop the clearest way of expressing that value and how people will perceive it.
But for both of these core hypotheses, it’s wild to me how many teams don’t have something sketched out, even in theory. Again: No one really knows for sure whether their theory will be right. But if you haven’t established your hypotheses, you have nothing to test against! The whole scientific method wheel doesn’t work, because you’re jumping to the “data collection” phase without having anything to prove or disprove.
You probably won’t be shocked to hear that this abbreviated approach isn’t nearly as effective.
So read the book. In it, you’ll see the simplified, more memorable structure that Reis recommends: Build, Measure, Learn.
But I’d encourage you to put on your lab coat and do the whole dang thing. If you’re trying to discover (or rediscover) the foundational model for your business:
Write out your observations and questions.
Research what people already know about this. So fast, with AI!
Write out your actual hypothesis. What do you think is the main value people will buy? How are you expecting to sustainably let those people know about it?
Test it with an experiment. This is the MVP on the value side, but you can do “Minimum Viable Tests” on the marketing side, too: What’s the fastest way to figure out if a growth channel works for you?
Analyze the data. Seriously. You have to analyze the data.
Report conclusions. Tell other people!
Repeat! 🥳
Science is great. Trust me, it comes highly recommended.
Everyone’s a startup, now?
The reason I’m writing this newsletter and trying to get everyone to understand how startup marketing works from a first principles approach is that a lot more companies are going to be needing this approach.
As I mentioned before, the term “startup” has become so broadly applied that it pretty much just means “Tech company.” Which has been problematic in the past, since young companies often try to apply later-stage playbooks from “other startups” (which aren’t actually startups) before they are really ready.
But now… well, is it fair to say that every company is operating “under conditions of extreme uncertainty”? I think so. The double disruption of AI that I’ve described for tech companies is actually an extension of the two core hypotheses that Ries described:
AI is threatening the core value hypothesis of companies, because AI capabilities may replace software capabilities.
AI is threatening the core growth hypothesis of companies, because AI is upending the previous media landscape and distribution channels.
So I’ve been encouraging founders and marketing leaders at early teams to learn to love science. But I don’t think it stops there. A lot of the fundamentals need to be reconsidered. It’s time to go back to first principles. You’ve got to know, before you can grow again. And I’ve got just the method for you to try.
Hurray, science!