The top-down trap
What *seems* safe winds up being high-risk
We know we need to build stuff people want. Why is this so difficult to do in practice?
Here’s an idea.
There are two ways to approach the question of “what people want”:
Top-down: Focused on niches and personas (AKA: GROUPS of people)
Bottoms-up: Focused on ONE individual customer
The top-down approach seems safer to our MBA brains. We do research to find the right niche, then interview a bunch of prospective customers, and figure out what the shared needs are. We then analyze the market size for the product that will address the shared needs & build.
And guess what? It doesn’t work, and we don’t know why.
Here’s why this happens:
Groups of people are NON-ERGODIC. The group doesn’t represent the individual.
AKA, we create a version of Frankenstein’s monster based on a bunch of individual customers that’s supposed to represent individual customers, but doesn’t.
Abstractions are figments of OUR imaginations. The bias is baked-in by default.
AKA, because the Frankenstein’s monster we create exists only in our mind, we always shape our Frankenstein’s monster based on our biases.
It would be weird if the top-down approach actually worked.
It’s cruel: The seemingly low-risk approach is actually high-risk. It’s a way to build a business that works in paper, but not in practice because we haven’t built something one real person actually wants. Our safety blanket winds up strangling us.
The solution is the bottoms-up approach: We build our business without abstractions, just with individual customers. Maybe we can find one representative customer to design around, but probably we have to figure it out through volume.
The components of the bottoms-up model:
It is about replicating ONE customer case study. Not about selling a product to a niche or persona or whatever.
As we attempt to replicate ONE customer case study, we find patterns. These patterns give us the chance to generalize our case study to resonate with an increasingly broader audience.
BUT, as we generalize our case study, it must REMAIN a “hell yes” for each individual customer. The risk as we generalize is that in trying to resonate with a bunch of different customers, we wind up resonating with nobody.
How I’m thinking about this with Waffle and PMF Camp:
Waffle: The “One Customer” I’ve designed our case study around is an early-stage B2B AI startup. They want to stay lean, don’t want to spend their time thinking about infrastructure, but are going to need SOC 2 compliance. Waffle is a “hell yes” for them when it makes AWS insanely simple, by giving them SOC 2 compliant infrastructure they don’t have to think about, in just a few clicks.
PMF Camp: The “One Customer” I’ve designed our case study around is a smart B2B founder who has done a bunch of research, raised some $$, built a product, tried to start selling… and it’s not quite working, and she doesn’t know why it’s not working. PMF Camp is a “hell yes” for her when it gives her an easy-to-implement approach to figure this out, 1:1 feedback on sales calls from a founder who’s done this before, and live teardowns of other B2B startups.
As each of these has taken on customers, I’ve felt the urge to generalize these case studies. As I’ve generalized, I’ve violated the “hell yes” rule - nobody would look at my generalized case studies and say “hell yes!” This is natural, because I want to resonate with everybody, to expand my TAM, to expand optionality.
But I have to remind myself at every turn, that there is no scale transformation: There is no point at which a switch flips, and I now sell to a group of people or a persona or a niche. In order to sell to a million customers, I have to sell to one customer at a time, a million times over. And that requires a bottoms-up “hell yes.”
The bottoms-up model and “hell yes” rule keep me grounded, and help me avoid my top-down tendencies. I hope they help you too!
—
PS: