Where to find ideas
A no-BS guide for new startups, new products, new GTM approaches, and new features
Starting to fill up for Q3 - get my help setting up & debugging your pre-PMF sales / GTM! More info HERE, my Calendly is HERE.
—
There are two kinds of people in startupland who bother me more than just about anyone:
The get-rich-quick influencer who says things like: “You can literally just prompt AI to find underserved niches and build a profitable AI startup.”
The academic who hasn’t built anything, yet feels comfortable telling you to use their complicated startup framework to find and validate ideas.
Here’s what it actually takes to find a startup idea. This also will help if you’re working at a startup considering building a new product, or even just a new feature. I am writing this so that you, a serious person trying to build something real, don’t feel like an idiot when you see people peddling nonsense about how easy it is.
This post digs into three fundamental questions:
What are we looking for?
Where do we find it?
How do we know we’re right?
Longer conversation on YouTube below, or listen on Spotify
1 - What are we looking for?
If we don’t know what we’re looking for, it’s hard to find it. And most of the things we are told to look for are imprecise at best, backwards at worst:
Pain points and problems? No - Life is a series of pain points and problems people mostly do nothing about. Or, to use Bob Moesta’s elegant phrase: “Bitchin’ ain’t switchin’!”
People who want our product? No - Nobody wants to buy B2B software. Nobody has ever wanted to buy B2B software, and nobody will ever want to buy B2B software.
People who would benefit from our product? No - that doesn’t mean they will buy it. Even if there’s compelling ROI, even if it will “earn them money or save them money” - “would benefit” is not the same as “has demand.”
A market need? An underserved niche? Demand? WTF do these things even mean?
Let’s get back to basics: Most people, most of the time, would be weird to buy a startup’s product - hell, any product. Most people have priorities we can’t help with; when they DO have a relevant priority, most of the time they have good enough options.
So: We are looking for a person who has an unavoidable priority, where their current options are insufficient or unworkable. This person would be weird not to buy our product.
I use the PULL framework for this:
Note that our startup idea is downstream of this. By articulating PULL, we are crafting a hypothesis of a demand-side need. Once we find this, then we can design our supply to fit the need elegantly.
The beauty of PULL is that it is supply-agnostic; we are looking for something out there in objective reality, not “desire for some product that doesn’t yet exist.” And when we find this thing, we find someone who will rip a janky product out of our hands… when most people would (rightly) not use it if we paid them to.
God, I would have saved years of suffering if someone had just told me this.
2 - Where do we find it?
Again, we are steered into the iceberg with traditional startup advice: Discovery interviews, the current “gold standard” for startup research, are mostly LARPy bullsh*t.
They are the kind of thing that sound right, that make sense, that take up time on your calendar, that give you the illusion of progress.
But.
If you actually cared about the outcome, if your life depended on the outcome, would you just do some interviews? Of COURSE not. You would go get firsthand experience, you would get your hands dirty.
Serious founders don’t rely on discovery interviews. Serious founders don’t just whiteboard, brainstorm, theorize. They become their potential customers, treating the task as a full-contact sport. When you do this, you observe things - feel things - that could form into PULL. Otherwise, you have, at best, secondhand knowledge.
Some relevant founder stories:
One graduated HBS, got certified as an emergency medical technician, and worked for $15 per hour for months - as he did this, an insight emerged about paperwork that turned into his PULL hypothesis and startup.
Another wanted to build software for greenhouses - instead of just interviewing greenhouse owners, he bought a greenhouse and ran it.
Others have built services companies first, then launched software/AI products because they know particular workflows so intimately.
Obviously you have to do something like this: Do you really think you can build a product people actually want, and actually value, that fits into their world so perfectly they rip it out of your hands… just by interviewing them? No! Go in-person. Do the work yourself. Get a firsthand view. If you’re not willing to do that, look in the mirror: Are you really serious about building something that serves customers? Or are you just in this to serve yourself?
Now, there are generally two approaches to creating a fast-growing startup (from Waves, Rivers, and Dams):
Waves: Here you need to go tinkering at the frontier of a new technology, specifically looking for new projects that are emerging on peoples’ to-do lists as they also tinker with the new technology. Pinecone for Vector Databases is an example here - the founders were in big tech building mega-scale AI infrastructure before others were; vector databases emerged as supply they needed. They assumed the rest of the world was 6-12 months behind them, and were right. They rode a wave by making the technology they needed at the frontier, accessible to others who were following.
Dams: Here you are looking at existing projects on peoples’ to-do lists, where existing options are lacking or unworkable, and leveraging new technologies to offer a superior experience. The approach is to observe existing projects and offer new options to see if buyers PULL. Think: Lovable versus Squarespace or Wix: They target the same existing project (building a website), but Lovable is wildly superior than clicking around like a fool in Squarespace.
Whatever path you’re trying to take, the path is still the same: Get firsthand experience, because that’s where real hypotheses emerge from.
3 - How do we know we’re right?
Until someone buys, we just have a hypothesis.
Given that:
this hypothesis determines whether or not we have a real business
it is very easy to create a hypothesis that sounds right, but very difficult to create a hypothesis that actually works
…it is in our best interest to figure out if our hypothesis is wrong ASAP. Anything less than someone buying means we still only have a hypothesis.
Buying is the bare minimum - what we’re eventually solving for is customer retention. But buying is the first step. It doesn’t matter if potential customers nod enthusiastically and say, “Wow this is such a big problem!” “What a great idea!” “You’re soooo dreamy!” It’s still a hypothesis until they buy, no matter what they say.
And yet: We are told to postpone this pesky reality by pursuing something called a design partner - a phrase that instantly causes potential customers to stop considering their actual priorities and options, and instead think, “this is a fun way to pretend I’m a startup person without risking anything!”
When we “convert” a design partner, we think we’ve validated demand; they think we are children.
I have seen so many entrepreneurs waste months on design partners: They spend months building the product, getting feedback, only to realize their design partners won’t buy, and even if they do buy, they didn’t have real, replicable demand.
People recommend we pursue design partners because they don’t understand pre-PMF sales: They think selling means pretending we have a fully fleshed-out product, then persuading people to buy it. When we understand PULL, we realize this is all backwards.
I wrote in depth how to run a pre-PMF sales call here. Soon I’ll write how to “test” your PMF hypothesis, but here’s the spoiler:
Write your hypothesis using the PULL framework
Talk to 5 potential customers who should, if the hypothesis is true, rip the product out of your hands. Record the conversations so you can review them.
Based on what happens on the calls, adjust and repeat steps 1-2.
—
Starting to fill up for Q3 - get my help setting up & debugging your pre-PMF sales / GTM! More info HERE, my Calendly is HERE.
"academic who hasn’t built anything, yet feels comfortable telling you to use their complicated startup framework to find and validate ideas." - Is this aimed at me 🥲?